JP4661529B2 - Image processing device - Google Patents

Description

  The present invention relates to an image processing apparatus that scans image data and calculates a feature amount of a graphic composed of connected pixel groups included in the image data.

  Conventionally, as image processing devices for recognizing figures such as characters and pictures from image data captured by a camera, feature quantities such as area, center of gravity, and perimeter of a figure composed of connected pixel groups are calculated. Things are known.

As such an image processing device, the acquired image data is a graphic pixel (for example, a black pixel) which is a graphic element for each pixel, or a background pixel (for example, a white pixel) other than the graphic pixel. Is converted to binary data representing whether or not, and then a label that is graphic identification information is assigned to each graphic pixel so that the same label is attached to the connected graphic pixels, There has been proposed an apparatus for calculating a feature amount of a graphic included in image data by calculating a feature amount of the pixel group using a pixel group to which the same label is assigned as one graphic (for example, Patent Document 1). Etc.).
Japanese Patent Laid-Open No. 5-189563

  However, the proposed image processing apparatus scans the image data and assigns a label to each pixel and calculates the feature amount of the graphic. The graphic pixel to which the label is applied depends on a different label. In the case of pixels that connect other specified graphics, it is assumed that the graphics specified by these different labels form one graphic, and every time such pixels are processed, the image The graphic pixels to which these labels are assigned are searched from all the pixels constituting the data, the labels are matched, and the feature amount of the pixel group including the graphic pixels with the matched labels is calculated.

  For this reason, the processing for one pixel takes time, and the timing for starting the processing for the next pixel is delayed, which causes a problem that the scanning time of image data for one screen becomes long. The scanning time of the image data is longer than the time required for the video camera to output the image data for one screen (generally, about 33 ms because the image data for 30 screens is output per second). Therefore, in the proposed image apparatus, the image data input from the camera cannot be processed continuously.

  The present invention has been made in view of such problems, and in an image processing apparatus that calculates a feature amount of a graphic composed of connected pixel groups included in image data, the scanning time of image data on one screen is shortened, and the image The purpose is to enable continuous processing of data.

  In the invention according to claim 1, which is made to achieve the above object, when image data is inputted, the binarizing means converts the image data into graphic pixels which become graphic elements for each pixel. Or binary data representing whether the pixel is a background pixel other than a graphic pixel. Then, the pixel determination unit scans the image data converted into binary data by the binarization unit, sequentially extracts graphic pixels as the target pixel from the image data, and labels the pixels around the extracted target pixel. From the pixel information storage means, and based on the acquired labels of surrounding pixels, the pixel of interest is a new pixel that forms a new figure or an addition that becomes a part of an already formed figure It is determined whether the pixel of interest is an additional pixel. If the pixel of interest is an additional pixel, it is further determined whether the additional pixel is a connected pixel that connects a plurality of different figures.

  When the pixel determination unit determines that the target pixel is a new pixel, the new label issuing unit issues a new label, and when the pixel determination unit determines that the target pixel is a connected pixel, The information storage means stores connection information indicating that a plurality of figures are connected at the connection pixel in the connection information storage means.

  Further, when the pixel determination unit determines that the pixel of interest is a new pixel, the initial value setting unit stores the new label issued by the new label issuing unit in the label storage unit and specifies the new label. The initial value is stored in the feature amount storage unit as the feature amount of the figure to be processed, and the new label is stored in the pixel information storage unit as the label of the target pixel. In addition, when the pixel determination unit determines that the target pixel is an additional pixel, the existing value update unit includes a figure in which the additional pixel is a part of the feature amount of the figure stored in the feature amount storage unit. The feature amount is updated, and the label of the graphic in which the additional pixel is a part thereof is stored in the pixel information storage means as the label of the target pixel.

  When the scanning of the image data for one screen by the pixel determining unit is completed, the operations of the new label issuing unit, the link information storing unit, the initial value setting unit, and the existing value updating unit based on the determination of the pixel determining unit are completed. The feature amount updating means rewrites and matches the labels of the respective figures recognized as the same figure among the labels stored in the label storage means based on the connection information stored in the connection information storage means, Of the feature quantities of the figure stored in the storage means, the feature quantities of each figure specified by each label before matching the labels are synthesized, and the synthesized feature quantities are specified by the matched labels. It is stored in the feature amount storage means as the feature amount of the figure.

  That is, in the image processing apparatus of the present invention, while scanning image data for one screen, the feature amount is not completely calculated, and only the information necessary for calculating the feature amount is stored in the label storage unit, After the scanning of the image data for one screen is completed and stored in the connection information storage means and the feature amount storage means, the feature amount of the figure is calculated based on the information stored in each storage means at the time of scanning. For this reason, the scanning time of the image data can be shortened as compared with the case where the feature amount of the graphic is calculated while scanning the image data. For example, images sequentially input from the imaging device such as a camera for each screen. Even when the feature amount of a graphic included in image data is calculated from the data, the image data can be scanned in time for the timing at which the imaging apparatus outputs the image data, and processing can be performed continuously. .

By the way, image data of one screen is scanned by the pixel determining means, and the time for completing the storage of the data for calculating the feature quantity of the figure in each storage means by each of the above means, and the feature quantity updating means are stored in the storage means. Acquisition of image data for the next screen after the pixel determination means starts acquiring image data for one screen, which is the time combined with the time for calculating the feature amount of the figure based on the data stored in the group When the pixel determination means starts scanning the image data of the next screen, the calculation of the feature amount of the graphic included in the scanned image data is also completed. ing. However, at the time when the pixel determination unit starts scanning the image data of the next screen due to the performance of the logic circuit constituting each unit or the CPU, the feature amount of the graphic on the previous screen by the feature amount update unit In this case, the information on the next screen is written in each storage means being used by the feature amount update means, and the feature amount update means before the calculation is in progress. There is a problem that the feature amount of the figure on the screen cannot be calculated accurately.
Therefore, as in the first aspect of the invention, there are provided a plurality of storage means groups, and information for calculating the feature amount of the figure obtained by scanning the image data for one screen includes the initial value setting means and the existing value setting means. If the value updating unit and the linked information storage unit sequentially write to the plurality of storage unit groups, and the feature amount updating unit sequentially selects the plurality of storage unit groups, the feature amount updating unit sets the initial value. It is possible to prevent the means, the existing value updating means, and the linked information storage means from simultaneously using the storage means group that is used in common. Therefore, it is possible to prevent the information of the next image from being written into the storage means group that is being used by the feature quantity update means, so that the feature quantity of the figure can be reliably calculated. Further, since the calculation of the feature amount by the feature amount update unit and the scanning of the image data by the pixel determination unit can be performed in parallel, it is possible to increase the speed of capturing the image data.
In the invention of claim 2, as in the invention of claim 1, when image data is input, the binarizing means converts the image data into graphic pixels which are graphic elements for each pixel. Or binary data representing whether the pixel is a background pixel other than a graphic pixel. Then, the pixel determination unit scans the image data converted into binary data by the binarization unit, sequentially extracts graphic pixels as the target pixel from the image data, and labels the pixels around the extracted target pixel. From the pixel information storage means, and based on the acquired labels of surrounding pixels, the pixel of interest is a new pixel that forms a new figure or an addition that becomes a part of an already formed figure It is determined whether the pixel of interest is an additional pixel. If the pixel of interest is an additional pixel, it is further determined whether the additional pixel is a connected pixel that connects a plurality of different figures.
When the pixel determination unit determines that the target pixel is a new pixel, the new label issuing unit issues a new label, and when the pixel determination unit determines that the target pixel is a connected pixel, The information storage means stores connection information indicating that a plurality of figures are connected at the connection pixel in the connection information storage means.
Further, when the pixel determination unit determines that the pixel of interest is a new pixel, the initial value setting unit stores the new label issued by the new label issuing unit in the label storage unit and specifies the new label. The initial value is stored in the feature amount storage unit as the feature amount of the figure to be processed, and the new label is stored in the pixel information storage unit as the label of the target pixel. In addition, when the pixel determination unit determines that the target pixel is an additional pixel, the existing value update unit includes a figure in which the additional pixel is a part of the feature amount of the figure stored in the feature amount storage unit. The feature amount is updated, and the label of the graphic in which the additional pixel is a part thereof is stored in the pixel information storage means as the label of the target pixel.
When the scanning of the image data for one screen by the pixel determining unit is completed, the operations of the new label issuing unit, the link information storing unit, the initial value setting unit, and the existing value updating unit based on the determination of the pixel determining unit are completed. The feature amount updating means rewrites and matches the labels of the respective figures recognized as the same figure among the labels stored in the label storage means based on the connection information stored in the connection information storage means, Of the feature quantities of the figure stored in the storage means, the feature quantities of each figure specified by each label before matching the labels are synthesized, and the synthesized feature quantities are specified by the matched labels. It is stored in the feature amount storage means as the feature amount of the figure.
In other words, in the image processing apparatus according to the second aspect of the present invention, the feature amount is not completely calculated while scanning the image data for one screen, and only the information necessary for calculating the feature amount is labeled. A feature value of a figure is calculated based on information stored in each storage means at the time of scanning after being stored in a storage means, a connection information storage means, and a feature value storage means, and scanning of image data for one screen is completed. . For this reason, the scanning time of the image data can be shortened as compared with the case where the feature amount of the graphic is calculated while scanning the image data. For example, images sequentially input from the imaging device such as a camera for each screen. Even when the feature amount of a graphic included in image data is calculated from the data, the image data can be scanned in time for the timing at which the imaging apparatus outputs the image data, and processing can be performed continuously. .
By the way, as described in claim 1, the image processing apparatus may be provided with a plurality of storage means groups, but in this case, a large number of storage areas are required. Here, in general, the time for the feature amount updating unit to calculate the feature amount of the graphic included in the image data for one screen from the information stored in the storage unit group is the image data for each screen. The imaging device such as a camera to output is sufficiently shorter than the time from the output of image data to the output of the next image data, and in a short time after the storage of data in each storage means is completed. The calculation of the feature amount of the figure ends. That is, when image data output from an imaging device such as a camera is directly acquired from the imaging device, there are two storage means groups for storing information for calculating the feature quantity of the figure. If there is enough.
Therefore, when two storage means groups are provided and the storage means groups are used alternately as in the second aspect of the invention, in particular, image data captured by an imaging apparatus such as a camera is captured by the imaging apparatus. When the feature amount update unit calculates the feature amount, the next image data information is written to the storage unit group used by the feature amount update unit. Therefore, it is possible to reliably calculate the feature amount of the figure with the minimum storage area.
By the way, the connection information storage means may be configured to sequentially store the combination of labels of connected figures in the connection information storage means, but in this case, the storage area for storing the combination of labels is increased. I need. Therefore, as described in claim 3 , when the pixel determination unit determines that the target pixel is a connection pixel that connects the graphics, the connection information storage unit stores 2 of all the labels that specify the graphic. It is preferable to set a connected state or a non-connected state for all label combinations consisting of one label, and in this way, the connection information storage means stores again the combination of labels once stored. It is possible to prevent the storage area from being used in vain. In addition, when the feature amount updating unit calculates the feature amount, it does not attempt to rewrite the label repeatedly for the same label combination, so that the processing time can be shortened.

  By the way, in the label storage means, when the issue order area of the label with the later issue order is not rewritten, the label with the later issue order among the labels stored in the issue order areas of the label combination. The label stored in the label issue order area is always the label with the issue order later than the label stored in the issue order area of the label with the earlier issue order. Then, when the feature amount updating unit reads the state set for the combination of labels in the connection information storage unit in order from the combination including the label with the earlier issuance order, in the above-described case, the issuance order is slow. The combination that can be rewritten to the label has not been read yet, and there is no area that has been rewritten to the label with the later issue order in the label storage means.

Therefore, as described in claim 4 , the feature amount updating means rewrites the label issue order area of the label with the later issue order out of the combination of labels in which the connection state is set in the label storage means. If it has not been rewritten, this area is simply rewritten with the label stored in the issue order area of the label with the earlier issue order, and the label storage means is searched. It is possible to reduce the number of times of performing the calculation, and it is possible to shorten the calculation time of the feature amount of the figure.

In the image processing apparatus according to any one of claims 1 to 4, the initial value setting means and the existing value updating means, as claimed in claim 5, once the label of the pixel of interest in the register When storing the label of the next target pixel, the label of the previous target pixel stored in this register is stored in the area corresponding to the previous target pixel in the line buffer, and then the next target pixel is stored. The label may be stored in a register. In this case, for example, a storage unit capable of storing the labels of the pixels for one screen is provided, and the storage area can be reduced as compared with the case where the labels of all the pixels constituting one screen are stored.

On the other hand, according to the image processing apparatus according to any one of claim 1 to claim 5, as described in claim 6, the first image acquisition unit that is capable of acquiring constructed directly image data from the imaging device May be provided. In this way, image data captured by an imaging device such as a camera can be acquired directly from the imaging device, and the image data is scanned and processed continuously at the timing when the imaging device transmits image data. It can be carried out.

Then, the image processing apparatus according to any one of claims 1 to 6, as described in claim 7, the image data is configured to acquire the image data from the stored storage medium A second image acquisition unit may be provided. The image data may be obtained directly from the imaging device. However, according to the seventh aspect , when the image data is already stored in the storage medium, the image data is read from the storage medium, The feature amount can be calculated.

By the way, when an image acquisition unit that acquires image data from an imaging device such as a camera is provided, it takes 6 cycles for the image acquisition unit to acquire and transmit one pixel, and the binarization unit acquires one pixel. In the case where two cycles are required for transmission, the binarization means needs to perform binarization processing in four cycles per pixel. Therefore, as described in claim 8 , if the image data is acquired and transmitted for each of a plurality of pixels, the time required to acquire and transmit the pixels can be reduced, so that the image data is converted into binary data. Can be made long, and the restriction on the processing speed of the binarization means is relaxed.

In general, the luminance value of a pixel is composed of 8-bit data. Therefore, as described in claim 9 , when image data is transmitted every four pixels, for example, when a transmission path having a bus width of 32 bits is used, image data can be transmitted and received efficiently. it can.

On the other hand, in the image processing apparatus according to any one of claims 1 to 9, binarizing means, as claimed in claim 10, the image data, which is set by the luminance value threshold You may comprise so that it may convert into binary data with a value. In this way, the image data can be converted into binary data and graphic pixels can be extracted from the difference in luminance of each pixel constituting the image data.

In the image processing apparatus according to any one of claims 1 to 9, binarizing means, as claimed in claim 11, the image data and the background data stored in advance in the storage means It is also possible to generate difference data consisting of the difference between the first and second data as new image data, and to convert the image data into binary data using a threshold value set by the luminance value.

  In other words, by calculating the difference between the image data and the background data, new image data having different luminance values only for the changed portion of the pixel, that is, the pixel of the moved object, is formed. By converting to binary data using the threshold value set in (2), and determining the image data converted into binary data by the pixel determining means, the feature quantity of the graphic representing the moving object can be calculated.

In addition, according to the image processing device of the twelfth aspect , since the pixel determination unit determines the graphic pixel adjacent in the eight directions around the target pixel as a pixel connected to the target pixel. Even a shape object can be reliably identified as a single figure.

On the other hand, according to the image processing apparatus of the thirteenth aspect , the pixel determining means determines a graphic pixel adjacent to the target pixel in the four directions of the horizontal and vertical directions as a pixel connected to the target pixel. Therefore, for example, when two objects are connected in an oblique direction, they can be distinguished from other objects. In addition, since the number of pixels to be determined is reduced as compared to the case where graphic pixels adjacent in the surrounding eight directions are determined as connected pixels, the processing time can be shortened.

The image processing apparatus according to any one of claims 1 to 13, as described in claim 14, each independent of the connection information storage means, and an initial value setting means and the existing value updating means Thus, it may be configured to be operable. Specifically, for example, the CPU may execute parallel processing, or each unit may be configured by a separate electronic circuit. In this way, processing by each unit can be performed in parallel. As a result, the processing time can be shortened.

Meanwhile, the feature of the figure, the value requiring the scanning of the image data more than once (for example, figures such as dispersion) may be, but as described in claim 15, in a single scan Since it is not necessary to scan the scanned image data again if the feature amount of the figure can be calculated, it is necessary to provide a storage means for storing the scanned image data, a circuit for performing the scanning again, and the like. Not convenient.

Further, as the feature amount of the figure that can be calculated by one scanning, as shown in claim 16 , any one of the area, the center of gravity, the peripheral length, the average luminance, and the including rectangle of the figure is used as the feature amount of the figure. In this case, since each feature amount is calculated by one scan, the same effect as that of the fifteenth aspect can be obtained.

Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic configuration diagram showing a configuration of an image processing apparatus 10 to which the present invention is applied.
The image processing apparatus 10 according to the present embodiment scans image data captured by an imaging device such as a camera, calculates a feature amount of a figure including connected pixels included in the image data, and uses this feature amount. It is for recognizing a specific object such as a person, a car, or a building, and an image input I / F unit 12 for acquiring image data output from the external device 90, and input image data, The binarization processing unit 14 for converting to binary data for each pixel, and the image data converted into the binary data output from the binarization processing unit 14 are acquired, and the type of each pixel to be configured is sequentially A pixel determination unit 16 for determination, a rewritable memory such as SRAM, and the like, a storage unit 18 that stores various information, and a label number that is identification information of a figure in response to the determination result of the pixel determination unit 16 New label issued When the label number is input, the unit 32, the connection information storage unit 34 that stores the connection information of each figure in the storage unit 18 in response to the determination result of the pixel determination unit 16, and is specified by the input label number. The feature amount of the figure to be calculated is calculated, and the temporary feature amount update unit 36 that stores the calculated feature amount in the storage unit 18 and the determination of the image data for one screen by the pixel determination unit 16 are finished, and the new label issuing unit 32 When the operations of the connection information storage unit 34 and the temporary feature amount update unit 36 are completed, the feature amount of the connected figure is determined based on the information stored in the storage unit 18 with the connected figure as a new figure. Is calculated, and the feature quantity update unit 38 for updating the feature quantity of the graphic stored in the storage unit 18 is read out. The feature quantity of the graphic stored in the storage unit 18 is read out, and a specific object is recognized from the feature quantity of the graphic. From the CPU that executes the process for That a control unit 40, and a.

  In this embodiment, each part except for the control part 40 is constituted by one LSI 50 having a clock generation part, and the same clock signal is supplied to each part, and the operation timing of each part is determined by this clock signal. Is set.

  The image input I / F unit 12 is connected to the external device 90, acquires image data to be processed, and scans the acquired image data in a raster manner to generate a digital signal. When the upper left corner of the image is the origin, the horizontal direction is the X axis, and the vertical direction is the Y axis, the coordinates (X, Y) of each pixel and the luminance value represented by 256 gradations are set to a plurality of pixels (this embodiment). In the embodiment, every four pixels) is output to the binarization processing unit 14 together with the write signal. In the present embodiment, the external device 90 has a built-in image sensor, and outputs image data captured by the two-dimensional image sensor at a constant speed (for example, a normal video rate of 33 ms / 1 image). Device (so-called camera).

  The binarization processing unit 14 acquires the image data output from the image input I / F unit 12 and sequentially converts it into binary data for each pixel. More specifically, the binarization processing unit 14 receives the write signal output from the image input I / F unit 12 and acquires data of four pixels output from the image input I / F unit 12. Then, based on the acquired luminance value, each pixel is a pixel that is a background element (white pixel in the present embodiment) or a graphic pixel that is a graphic element (black pixel in the present embodiment). Judgment is made sequentially based on a preset threshold value, and the binarization result in which the background pixel is represented as 0 and the graphic pixel is represented as 1 and the coordinates (X, Y) of each pixel are written every 4 pixels. It outputs to the pixel determination part 16 with a signal. When the write signal is received again, the next four pixels of data are acquired, and the same operation is repeated. Of the transmission lines connecting the image input I / F unit 12 and the binarization processing unit 14, the transmission line to which the luminance value is transmitted uses a transmission line having a bus width of 32 bits.

The pixel determination unit 16 determines the type of each pixel constituting the image data (binary data) acquired from the binarization processing unit 14 for each pixel.
Specifically, the pixel determination unit 16 receives the write signal output from the binarization processing unit 14, acquires the data of the four pixels output from the binarization processing unit 14, and coordinates the acquired four pixels as coordinates. Search is performed in order from the pixel having the smallest value of (X, Y), and a graphic pixel is extracted based on the binarization result. Then, when the graphic pixel is extracted, the extracted pixel is set as the target pixel, and as shown in FIG. 2A, the target pixel and the four neighboring pixels (upper left, upper, upper right, and left four pixels) From the binarization result of the total five pixels, the type of the target pixel is determined based on the pixel determination flag issuing rule shown in FIG. In addition, one of the pixel determination flags is set according to the determination result, and an operation based on the determination flag described below is executed.

  Then, when the operation based on the determination flag ends, the pixel determination unit 16 extracts the next graphic pixel, sets the extracted pixel as a new target pixel, and sets the determination flag for all the graphic pixels included in the four pixel data. When the operation based on the set and determination flags is executed and the operations for all the graphic pixels included in the four-pixel data are finished, the operation is stopped until the next writing signal is received. 2A is an explanatory diagram illustrating five pixels used for determining the type of the pixel of interest, and FIG. 2B is an explanatory diagram illustrating a pixel determination flag issuing rule.

  Hereinafter, the operation of the pixel determination unit 16 will be described in detail for each pixel determination flag to be issued. In the following description, the label number is identification information of a figure, and the label numbers of connected pixel groups forming one figure are the same. In this embodiment, when the pixel of interest is a graphic pixel and pixels adjacent to the pixel of interest in the eight directions (up, down, left, right, diagonal) are graphic pixels, the pixel of interest and each graphic pixel are connected. Recognize figures as if they are. Further, the binarization result of the four pixels in the vicinity of the target pixel shown in FIG. 2A is determined based on a label number stored in a register 28 and a line buffer 30 constituting the storage unit 18 described later.

  The pixel determination unit 16 detects the binarization result of the four neighboring pixels, and the binarization result of the four neighboring pixels is the pattern shown in the first row in FIG. 2B, that is, around the target pixel. If it is determined that there is no graphic pixel and the target pixel is a pixel forming a new graphic, the pixel determination flag N is set, and the coordinates (X, Y) of the target image are output to the new label issuing unit 32. .

  Further, the pixel determination unit 16 determines that the binarization result of the four neighboring pixels is one of the patterns shown in the second row of FIG. 2B, that is, the target pixel is a part of the existing figure. If it is determined that it is a pixel to be formed and there is a possibility that an existing figure is connected, a pixel determination flag C is set, and two types of label numbers L1 and L2 of the four neighboring pixels (L1 ≦ L2) ) And the coordinates (X, Y) of the target pixel are output to the temporary feature amount update unit 36. Then, it is determined whether or not the label number L1 is equal to the label number L2. If not equal, the label number L1 and the label number L2 (assuming that L1 <L2) are further stored in the connection information storage unit 34. Output.

  Then, the pixel determination unit 16 determines that the binarization result of the four neighboring pixels is one of the patterns shown in the third row of FIG. 2B, that is, the target pixel is a part of the existing figure. If it is determined that the pixel is to be formed, the pixel determination flag D is set, and the label number L1 of the four neighboring pixels and the coordinates (X, Y) of the target pixel are output to the temporary feature amount update unit 36.

  The storage unit 18 includes a calculation storage unit 20 including a rewritable memory (e.g., SRAM) that forms the label table 22, the connection map 24, and the feature amount table 26, and a register 28 that can store a label number for one pixel. And a line buffer 30 composed of a rewritable memory (for example, SRAM) capable of storing the label numbers of pixels for one scanning line (refer to FIG. 5 for the register 28 and the line buffer 30).

  As shown in FIG. 3A, the label table 22 is configured such that the label numbers issued by the new label issuing unit 32 are sequentially written in the calculation storage unit 20, and the written label numbers are , The label number T [λ] is stored so as to be rewritable by the operation of the feature amount updating unit 38. However, T [λ] indicates the label number stored in the λth area of the label table 22, and the label number T [λ] and λ are equal before rewriting by the feature amount updating unit 38. .

  As shown in FIG. 3B, in the connection map 24, among the label numbers issued by the new label issuing unit 32, the smaller label number L1 is the vertical axis and the larger label number L2 is the horizontal axis. It is configured in a matrix, and for the intersections of different label numbers (that is, combinations of different label numbers) on the connection map 24, the figures identified by the label numbers L1 and L2 (that is, the label numbers are L1 and L2) (A graphic made up of graphic pixels) is connected so as to store connection flags (L1, L2) indicating whether or not they are connected. Then, for example, when the graphic identified by the label number L1 = 1 and the graphic identified by the label number L2 = 3 are connected, it is stored at the intersection of the label number 1 and the label number 3. The connected flag (1, 3) (circled in the figure) is set to a value of 1.

  In the feature quantity table 26, as shown in FIG. 3C, the figure feature quantity (in this embodiment, in other words, the number of pixels) calculated by the temporary feature quantity update unit 36 is sequentially displayed for each label number. It is configured to be written. The feature amount of the written figure is stored as a feature amount S [λ] so as to be rewritable by the operation of the feature amount update unit 38. However, S [λ] indicates the area stored in the λ th region, and before rewriting by the feature amount updating unit 38, S [λ] indicates the area of the figure specified by the label number λ. Yes.

  3A is an explanatory diagram showing the label table 22, FIG. 3B is an explanatory diagram showing the connection map 24, and FIG. 3C is an explanatory diagram showing the feature quantity table 26. In each explanatory diagram, the label numbers 1 to 4 are issued.

  The new label issuing unit 32 is a counter configured to increment the value 1 when the coordinate (X, Y) of the target image is input, and issues the incremented value as a new label number. Then, the issued label number is output to the temporary feature amount update unit 36 together with the coordinates (X, Y) of the target pixel. The new label issuing unit 32 is reset when the first pixel of one screen (in this embodiment, coordinates (X, Y) = (0, 0)) is input to the pixel determination unit 16 and the value 0 It is comprised so that. That is, it is initialized every time scanning of image data for one screen is started, and counting starts from an initial value of 0.

  When the label number L1 and the label number L2 are input, the connection information storage unit 34 sets the connection flag (L1, L2) stored at the intersection of the label numbers L1, L2 of the connection map 24 to a value 1.

  The temporary feature amount update unit 36 includes a new pixel update unit 36a that operates when a new label number is input, and an additional pixel update unit 36b that operates when an existing label number is input.

  When the new label number λ and the coordinates (X, Y) of the pixel of interest are input from the new label issuing unit 32, the new pixel updating unit 36a inputs the label table 22 as shown in FIG. Entered in the area of the label number of the new label (hereinafter referred to as the λth when the input label number is λ) (the fifth area indicated by the arrow in the figure) The new label number is written, and the initial area of the figure specified by the input new label number in the λ-th region (fifth region indicated by an arrow in the figure) of the feature quantity table 26 The value 1 (S [λ] = 1) is written in the λ th area of the feature quantity table 26 as a value.

  When the label number λ and the coordinate (X, Y) of the pixel of interest are input from the pixel determination unit 16, the additional pixel update unit 36 b writes in the label table 22 as shown in FIG. First, a value 1 is added to the area S [λ] of the figure stored in the λ-th region of the feature quantity table 26 (S [λ] ← S [λ] +1) to obtain the area S [λ of the feature quantity table 26. ] Value is updated. FIG. 4B shows a case where the input label number is 4, and S [4] stored in the fourth area (area indicated by an arrow in the figure) is 32. To 33.

  Here, FIG. 4A is an explanatory diagram showing the rewriting operation of the label table 22 and the feature quantity table 26 performed by the new pixel update unit 36a, and FIG. 4B is the feature performed by the additional pixel update unit 36b. FIGS. 4A and 4B are explanatory diagrams showing the rewriting operation of the quantity table 26. FIGS. 4A and 4B show the label table 22 and the feature quantity table 26 in the state shown in FIGS. 3A and 3C. In some cases, λ = 5 and λ = 4 are input as label numbers.

  Then, when the temporary feature amount update unit 36 updates the feature amount table 26 as described above, next, the target pixel (that is, the coordinates (X-1,. As shown in FIG. 5B, the label number of the pixel at the coordinates (X-1, Y-1) in the line buffer 30 (that is, the line buffer) 30 (X-1th region), and after writing the label number of the pixel at coordinates (X-1, Y) into the line buffer 30, the pixel of interest being processed (that is, the pixel at coordinates (X, Y)) ) Is written in the register 28.

  As described above, the label number of the target pixel (specifically, the label number input together with the coordinates of the target pixel) is temporarily stored in the register 28, and then stored in the line buffer 30. As illustrated in FIG. 5A, the determination unit 16 determines the upper left of the target pixel at coordinates (X, Y) from the label numbers stored in the X−1, X, and X + 1 regions of the line buffer 30. The binarization result of the upper and upper right three pixels, that is, whether each pixel is a graphic pixel is determined, and from the label number stored in the register 28, the binarization result of the left pixel of the target pixel Judging.

  5A is an explanatory diagram showing the register 28 and the line buffer 30 before the target pixel is processed by the temporary feature amount update unit 36, and FIG. 5B is a diagram of the target pixel by the temporary feature amount update unit 36. It is explanatory drawing which shows the register | resistor 28 and the line buffer 30 after a process.

  As described above, the binarization processing unit 14, the pixel determination unit 16, the new label issue unit 32, the connection information storage unit 34, and the temporary feature amount update unit 36 receive external input, Operations such as conversion to binary data, determination of pixel type, calculation of feature values of a figure for each label number, and the like are started. Then, until the processing of the image data for one screen is completed, that is, while the image data for one screen is output from the image input I / F 12, each of the above units receives the input from the outside and repeats the above operation. Execute. When the processing of the image data for one screen is completed, the temporary feature amount update unit 36 erases all the label numbers stored in the line buffer 30 and initializes the line buffer 30.

  By the way, when the pixel determination flag C is set in the pixel determination unit 16, the target pixel is a pixel that forms a part of an existing graphic, and there is a possibility that the existing graphic is connected. When the label number L1 is not equal to the label number L2, the target pixel connects two figures specified by the label number L1 and the label number L2. However, the graphic feature quantity for each label number calculated in the operation of the temporary feature quantity update unit 36 and stored in the feature quantity table 26 does not reflect the graphic connection information, and the provisional graphic feature is not reflected. Amount.

  Therefore, in the feature amount update unit 38, after the scanning of the image data of one screen in each unit is completed and the storage of the data in the calculation storage unit 20 is completed, the feature amount update unit 38 stores each of the label numbers stored in the feature amount table 26. The feature amount of the figure is updated by reflecting the connection information stored in the connection map 24 on the feature amount of the figure.

  The feature amount update unit 38 is configured to start an operation at a preset timing based on a signal supplied from the clock generation unit. In the present embodiment, the feature amount updating unit 38 starts the operation when the scanning of the image data for one screen in each unit is completed and the storage of the data in the calculation storage unit 20 is completed in each unit. It is set as timing.

  Then, the feature amount updating unit 38 is based on connection information between figures specified by different label numbers stored in the connection map 24, that is, based on a combination of label numbers for which the connection flag is set in the connection map 24. In addition to matching the label numbers for which the concatenation flag is set, the feature values of the graphic are updated by integrating the provisional graphic feature values for each label number for each matched label number.

  Hereinafter, the operation of the feature amount updating unit 38 will be described with reference to the flowchart shown in FIG. The feature amount updating unit 38 is actually configured with a logic circuit so as to operate according to the procedure shown in FIG. 6, but here, in order to facilitate understanding of the operation of the feature amount updating unit 38. This will be described with reference to a flowchart.

  As shown in FIG. 6, when the feature amount update unit 38 starts its operation, it scans the connection map 24 in a raster fashion. First, in S210, it is determined whether or not the concatenation flag (L1, L2) is set, and if it is determined that the concatenation flag (L1, L2) is set, then in S220, the L1th label table 22 is displayed. And the larger label number Lmax among the label numbers stored in the L2th area is extracted.

  When the label number Lmax is extracted in S220, the area where T [λ] = Lmax is stored in all areas of the label table 22 in order from the first area of the label table 22 (that is, Lmax is stored). And the operation of rewriting the label number in S240 to S260 described below is performed.

  More specifically, in S230, it is determined whether or not the label number T [λ] stored in the λth area of the label table 22 is equal to Lmax. If T [λ] = Lmax, the process proceeds to S240. Then, the label number T [λ] is rewritten to the smaller label number Lmin of the label numbers stored in the L1 and L2 areas. In subsequent S250, the area S [Lmin] of the figure specified by the label number Lmin (that is, the area stored in the Lmin-th region) stored in the feature quantity table 26 and the label number T [ The sum of the area S [λ] of the figure specified by λ] = Lmax (that is, the area stored in the λth region) is calculated, and S [Lmin] is rewritten to the calculated value, and in S260 Then, the area S [λ] stored in the feature table 26 is rewritten to 0, and the process proceeds to S270.

  On the other hand, if it is determined in S230 that the label number T [λ] is not equal to Lmax, the label number T [λ] is not rewritten or the like from S240 to S260. The process proceeds to S270.

  Then, in S270, it is determined whether or not the search of all areas in the label table 22 has been completed. If not, the process proceeds to S230, and the determination of the next area in the label table 22 is performed. The operations from S230 to S270 are repeatedly executed until the search of all areas of the label table 22 is completed. On the other hand, if the search of all the areas in the label table 22 is completed in S270, the process proceeds to S280. If it is determined in S210 that the link flag (L1, L2) is not set, the process proceeds to S280 without performing the above-described operations from S220 to S270 for the combination of label numbers. .

  In S280, it is determined whether or not the determination for all the connection flags in the connection map 24 is completed. If it is determined that scanning of all the connection flags in the connection map 24 has been completed, the process proceeds to S290. On the other hand, if it is determined that the scanning of all the connection flags is not completed, the process proceeds to S210, the determination of the next connection flag (L1, L2) is executed, and the scanning of all the connection flags is completed until the scanning of all the connection flags is completed. Repeat the operation.

  As described above, when the operations from S210 to S280 are executed and the scanning of all the connection flags of the connection map 24 is completed, the feature value table 26 stores the feature values of a new graphic in which the connected graphics are integrated. Is done.

  Next, in S290, all the connection flags of the connection map 24 are deleted and the connection map 24 is initialized. In S300, the number of feature quantities stored in the feature quantity table 26, that is, the total number of figures is set as an interrupt signal. At the same time, it is output to the control unit 40 and the operation is terminated.

  Hereinafter, operations performed by the feature amount updating unit 38 when the input image illustrated in FIG. 7A is input will be described in order along the flowchart illustrated in FIG. 6. However, the label image in FIG. 7A is a diagram in which the label number of each graphic pixel is described on each graphic pixel, and is not generated in the image processing apparatus 10 of the present embodiment. It is a reference diagram for facilitating understanding.

  7B is an explanatory diagram showing the label table 22 and the feature amount table 26 when the scanning of the image data of one screen is completed and the storage of the data in the calculation storage unit 20 is completed. FIG. 7C is an explanatory view showing the connection map 24 at the same time point, and FIG. 7D is an explanatory view showing the label table 22 and the feature amount table 26 rewritten by the operation of the feature amount update unit 38.

  When the operation is started, the feature amount updating unit 38 scans the connection map 24 shown in FIG. 7C in a raster manner. First, in S210, it is determined whether or not the connection flag (L1, L2) is set. Judgment is made and it is detected that the connection flag (1, 3) is set. In the subsequent S220, the larger label number 3 of the label numbers stored in the first and third areas of the label table 22 is extracted, and in the operation of S230 to S270, in the label table 22, An area having a label number of 3 is detected.

  More specifically, in S230, it is determined that T [3] = 3, and in subsequent S240, the third area of the label table 22 is stored in the first and third areas of the label table 22. Rewrite the label number 1 which is the smaller of the numbers. In S250, the figure area S [1] specified by the label number 1 stored in the first area of the feature table 26 and the figure area S [3] specified by the label number 3 Is calculated (ie, the areas 8 and 12 stored in the first and third regions), the area S [1] is rewritten to 20, and the area S [3] is rewritten to 0 in S260. In S280, it is determined that the determination for all the flags of the connection map 24 is not completed, and the process proceeds to S210.

  Next, in S210, it is detected that the connection flag (2, 3) is set. In subsequent S220, the larger label number 2 of the label numbers stored in the second and third areas of the label table 22 is extracted, and in the operation from S230 to S270, in the label table 22, An area having a label number of 2 is detected.

  More specifically, in S230, it is determined that T [2] = 2, and in subsequent S240, the second area of the label table 22 is stored in the second and third areas of the label table 22. Rewrite the label number 1 which is the smaller of the numbers. In S250, the figure area S [1] specified by the label number 1 stored in the first area of the feature quantity table 26 and the figure area S [2] specified by the label number 2 Is calculated (ie, the areas 20 and 4 stored in the first and second regions), the area S [1] is rewritten to 24, and the area S [2] is rewritten to 0 in S260. In S280, it is determined that the determination for all the flags of the connection map 24 is not completed, and the process proceeds to S210.

  Next, in S210, it is detected that the connection flag (3, 4) is set. In subsequent S220, the larger label number 4 of the label numbers stored in the third and fourth areas of the label table 22 is extracted, and in the operation from S230 to S270, in the label table 22, An area having a label number of 4 is detected.

  Specifically, in S230, it is determined that T [4] = 4, and in subsequent S240, the fourth area of the label table 22 is stored in the third and fourth areas of the label table 22. Rewrite the label number 1 which is the smaller of the numbers. In S250, the area S [1] of the graphic specified by the label number 1 stored in the first area of the feature quantity table 26 and the area S [4] of the graphic specified by the label number 4 Is calculated (ie, the areas 24 and 1 stored in the first and fourth regions), the area S [1] is rewritten to 25, and the area S [4] is rewritten to 0 in S260.

In S280, it is determined that the determination for all the connection flags in the connection map 24 is completed, and the process proceeds to S290.
Here, as shown in FIG. 7A, this graphic image is stored in the feature table 26 when scanning of all the connection flags in S210 to S280 is completed because all the graphic pixels are connected. The area thus obtained is S [1] = 25.

  Next, in S290, all the connection flags stored in the connection map 24 are deleted, and in the subsequent S300, the number of feature values stored in the feature value table 26, that is, the total number of figures 1 together with the interrupt signal is displayed. It outputs to the control part 40 and complete | finishes a process.

  The control unit 40 receives the interrupt signal from the feature amount update unit 38, calculates the total number of graphics input together with the interrupt signal from the feature amount update unit 38 from the feature amount of the figure stored in the feature amount table 26. To read out feature quantities (for example, when the total number of figures is 5, read out feature quantities in the first to fifth areas of the feature quantity table 26), and to recognize a specific object from the feature quantities of the readout figures Execute the process. Note that the method for recognizing an object from a feature amount of a graphic is a well-known method and is not directly related to the present invention, and thus detailed description thereof is omitted.

  As described above, the image processing apparatus 10 according to the present embodiment includes the calculation storage unit 20 including the label table 22, the connection map 24, and the feature amount table 26, the register 28, and the line buffer 30. The image input I / F unit 12 acquires image data from the external device 90, and the binarization processing unit 14 converts the image data input from the image input I / F unit 12 into binary data for each pixel. Then, the pixel determination unit 16 sequentially determines the type of each pixel constituting the image data converted into binary data, and the new label issuing unit 32 receives the determination result of the pixel determination unit 16 and uses the graphic identification information. A link number is issued, the link information storage unit 34 receives the determination result of the pixel determination unit 16, stores the link information of each figure in the link map 24, and the temporary feature amount update unit 36 receives the input label number. Figure identified by The feature amount of each figure is calculated and stored in the feature amount table 26. If the input label number is a new label number, a new label is added to the label table 22. Store. Then, when the determination of the image data for one screen by the pixel determination unit 16 is completed and the operations of the new label issuing unit 32, the connection information storage unit 34, and the temporary feature amount update unit 36 are completed, the feature amount update unit 38 The feature quantity of the figure is updated by reflecting the connection information stored in the connection map 24 to the feature quantity of the figure for each label number stored in the feature quantity table 26.

  Therefore, according to the image processing apparatus 10 of the present embodiment, it is possible to shorten the scanning time of the image data as compared with the case where the feature amount of the figure is calculated while scanning the image data. For this reason, the image data can be scanned from the external device 90 so as to be in time for the sequential input of the image data for each screen, and the processing can be continuously performed.

  Further, in the image processing apparatus 10 of the present embodiment, the connection map 24 is obtained by displaying the smaller label number L1 of the label numbers issued by the new label issuing unit 32 on the vertical axis and the larger label number L2 on the horizontal side. Concatenated flags (L1, L2) indicating whether or not the figures identified by the label numbers L1 and L2 are connected to different combinations of label numbers on the connection map 24. ) Is stored. When the pixel determination unit 16 determines that the figures having different label numbers are connected, the connection flag is set by the operation of the connection information storage unit 34.

  Therefore, according to the image processing apparatus 10 of the present embodiment, even if the pixel determination unit 16 repeatedly determines the connection between the graphics specified by the same label number, the connection information storage unit 34 sets the connection flag that has been set. The storage area to be used can be reduced as compared with a case where all combinations of graphic label numbers determined to be connected are stored, for example.

  On the other hand, the feature amount updating unit 38 scans the connection map 24 to detect the set connection flag, and rewrites the label table 22 and the feature amount table 26 for the combination of the label numbers indicated in the detected flag. The same label number combination is not rewritten twice. For example, all combinations of figure label numbers determined to be connected are stored, and this is read and rewritten. Compared to the case, the processing time for calculating the feature amount of the figure can be shortened.

  In addition, the image processing apparatus 10 according to the present embodiment includes a register 28 that can store a label number for one pixel, and a line buffer 30 that can store a label number of a pixel for one line scan. When the quantity update unit 36 updates the feature quantity table 26, the label number of the target pixel processed immediately before stored in the register 28 is written in the area corresponding to the target pixel in the line buffer 30. After the writing, the label number of the target pixel being processed is written to the register 28, and the pixel determination unit 16 determines the neighborhood 4 of the target pixel based on the label number stored in the register 28 and the line buffer 30. The binarization result of the pixel is determined.

Therefore, according to the image processing apparatus 10 of the present embodiment, it is not necessary to store the label numbers of the target pixels for all the pixels forming the image data for one screen, and the storage area can be reduced.
Further, in the image processing apparatus 10 according to the present embodiment, the binarization processing unit 14 determines whether each pixel is a pixel serving as a background element or a graphic element based on the acquired luminance value. Since it is sequentially judged whether it is a pixel or not by a preset threshold value, image data is converted into binary data and graphic pixels are extracted from the difference in luminance of each pixel constituting the image data Can do.

  Further, the binarization processing unit 14 acquires the image data every four pixels, converts the image data into binary data, and outputs the image data converted into the binary data to the pixel determination unit 16 every four pixels. Since it is configured and the time required for pixel acquisition and transmission can be reduced, the time for converting image data into binary data can be increased, and the restriction on the processing speed is eased. In this embodiment, the luminance value of the pixel is composed of 8-bit data, and the bus width of the transmission path through which the luminance value is transmitted is composed of 32 bits. By transmitting, image data can be transmitted and received efficiently.

  Further, since the pixel determination unit 16 is configured to determine a graphic pixel adjacent in the eight directions around the pixel of interest as a pixel connected to the pixel of interest, the shape of the graphic is complicated. Can also be reliably identified as a single figure.

  Since the link information storage unit 34, the new pixel update unit 36a, and the additional pixel update unit 36b are each configured by separate electronic circuits that can operate independently, an operation of setting a link flag; The rewriting operation of the feature quantity table 26 and the like can be performed in parallel, and the processing time can be shortened.

  Further, in the image processing apparatus 10 of the present embodiment, the area is calculated as the feature amount of the graphic, so that it is not necessary to scan the scanned image data again, and a memory for storing the scanned image data. There is no need to provide a means or a circuit for performing scanning again, which is convenient.

  In the present embodiment, the image input I / F unit 12 corresponds to the first image acquisition unit of the present invention, the binarization processing unit 14 corresponds to the binarization unit of the present invention, and the pixel determination unit 16 Corresponds to the pixel determination means of the present invention, the calculation storage unit 20 corresponds to the storage means group of the present invention, the label table 22 corresponds to the label storage means of the present invention, and the connection map 24 The feature quantity table 26 corresponds to the feature quantity storage means of the present invention, the new label issuing unit 32 corresponds to the new label issue means of the present invention, and the linked information storage section 34 corresponds to the linked information storage means of the invention. Corresponds to the linked information storage unit of the present invention, and the new pixel update unit 36a of the temporary feature amount update unit 36 corresponds to the initial value setting unit of the present invention, and the additional pixel update unit 36b of the temporary feature amount update unit 36. Corresponds to the existing value updating means of the present invention, and the feature amount updating unit 38 Corresponding to the characteristic value update section.

As mentioned above, although one Embodiment of this invention was described, it cannot be overemphasized that this invention can take a various form.
For example, as a first modification of the image processing apparatus 10 of the present embodiment, the operation of the feature amount update unit 38 may be performed as shown in FIG. 8 in order to shorten the detection time of the feature amount of the graphic. FIG. 8 is a flowchart showing the operation of the feature amount updating unit 38 of the first modification.

  As shown in FIG. 8, when the feature amount update unit 38 starts processing, the connection map 24 is scanned in a raster manner. First, in S310, it is determined whether or not the concatenation flag (L1, L2) is set, and if it is determined that the concatenation flag (L1, L2) is set, then in S320, the L2th label table 22 is displayed. It is determined whether or not the label number stored in the area is equal to L2, that is, whether or not the L2th area has been rewritten. When it is determined that the L2th stored label number is equal to L2, that is, the L2th area has not been rewritten yet, L2> L1, so the L2th stored label number Since T [L2] is a number larger than the L1 stored label number T [L1] (that is, T [L2]> T [L1]), it is stored in the label table 22 at L2 in S330. The label number T [L2] is rewritten to the L1-th stored label number T [L1].

  Supplementally, since the feature amount update unit 38 scans the connection map 24 in a raster manner, for example, it detects that the connection flag (3, 4) is set while scanning L1 = 3. When the label number 4 is stored in the fourth area of the label table 22, that is, when the label number of the fourth area has not been rewritten to 1 or 2, the row with the label number L1 = 4 is still present. Since it is clear that the area with the label number 4 does not exist in any area other than the fourth area, the fourth area is searched without searching the area with the label number 4 from the label table 22. The label number 4 (T [4]) stored in the first area can be rewritten with the label number 3 (T [3]) stored in the third area.

  On the other hand, when the label number of the fourth area is rewritten to 1 or 2, the label number T [3] stored in the third area and the label number T [4 stored in the fourth area ] Is larger, and it is necessary to search an area rewritten with the value of the larger label number of T [3] and T [4]. This is because, for example, when T [3] = 2 and T [4] = 1, the scanning of the row with label number L1 = 2 has been completed, so the label number has been rewritten to 2. This is because there is a possibility that there is another area, and the area in which the label number 2 is stored also needs to be rewritten to the label number 1.

  In S340, the area S [L1] of the figure specified by the label number L1 (that is, the area stored in the L1th area) stored in the feature table 26 and the label number L2 are stored. The area S [L2] (that is, the area stored in the L2th region) of the figure specified in this way is calculated and rewritten with the calculated value, and the area S [ L2] is rewritten to 0, and the process proceeds to S420.

  On the other hand, if it is determined in S320 that the L2 stored label number is not L2, that is, the label number has been rewritten, the process proceeds to S360 and the operations from S360 to S410 are performed. Since this is the same as the operation from S220 to S270 in the above embodiment, description thereof is omitted. Also, the operations from S420 to S440 are the same as the operations from S280 to S300 in the above-described embodiment, so this description is also omitted.

  As described above, in the image processing apparatus 10 according to the first modification, the feature amount update unit 38 scans the connection map 24 in a raster manner and detects that the connection flag (L1, L2) is set. It is determined whether or not the label number T [L2] stored in the L2 area of the label table 22 has been rewritten. If the label number stored in this area has been rewritten, the label table 22 is stored. A search is performed to extract the larger label number Lmax of the label numbers stored in the L1 and L2 areas of the label table 22, and the extracted label number Lmax is extracted from the L1 and L2th labels of the label table 22. The label number Lmin that is smaller of the label numbers stored in the area is rewritten. On the other hand, when the label number T [L2] stored in the L2th area has not been rewritten, L2 is rewritten to the label number T [L1] stored in the L1th area.

  Therefore, according to the image processing apparatus 10 of the first modification, the feature amount updating unit 38 selects the subsequent operation depending on whether or not the label number T [L2] is rewritten, and T [L2] is rewritten. If not, the label table 22 is not searched, so that it is possible to shorten the calculation time of the graphic feature amount.

Further, as a second modification of the image processing apparatus 10 according to the present embodiment, as shown in FIG. 9, two calculation storage units 20 may be provided in the storage unit 18.
In other words, in the present embodiment, the storage unit 18 includes one calculation storage unit 20 including a label table 22, a connection map 24, and a feature amount table 26. Although it was comprised so that it might be used for calculation, two calculation memory | storage parts 20 comprised from the label table | surface 22, the connection map 24, and the feature-value table 26 are provided, the connection information storage part 34 and the temporary feature-value update part 36 uses one of the two calculation storage units 20 in common, and alternately changes the calculation storage unit 20 to be used for each image data of one screen, and the feature amount update unit 38. However, the two calculation storage units 20 may be selected in order, and processing may be performed based on the data stored in the selected calculation storage unit 20.

  According to the image processing apparatus of Modification 2 configured as described above, the calculation storage unit 20 is alternately used. Therefore, for example, it takes time to calculate the feature amount of the figure by the feature amount update unit 38, and the figure Even if the next image data is input before the calculation of the feature amount is completed, the information on the next image data is not written to the calculation storage unit 20 used by the feature amount update unit 38. Therefore, it is possible to reliably calculate the feature amount of the figure.

  The same effect can be obtained by providing a plurality of calculation storage units 20 and using each calculation storage unit 20 in sequence. However, in the second modification, the external device 90 such as a camera is used. The output image data is processed, the timing interval at which the image data captured by the imaging device such as a camera is output for each screen, and the image processing device 10 scans the image data for one screen. Since it is presumed that the time for calculating the feature amount of the graphic included in the image data is almost equal, two calculation storage units 20 are provided and used alternately, thereby minimizing the storage area. It is possible to reliably calculate the feature amount of the figure.

  On the other hand, as a third modification of the image processing apparatus 10 of the present embodiment, the image input I / F unit 12 is connected to an external device 90 such as a camera and configured to be able to acquire image data to be processed. In addition, the image data stored in a storage medium such as a CDR may be acquired.

  In this way, even when image data is stored in a storage medium such as a CDR, it is possible to acquire this image data and extract a feature amount of a figure. In the third modification, the image input I / F unit 12 corresponds to a first image acquisition unit and a second image acquisition unit of the present invention.

  The image processing apparatus 10 according to the present embodiment is configured to extract a graphic representing a moving object from the acquired image data and calculate a feature amount of the graphic, as shown in Modification 4 below. Also good.

Hereinafter, an image processing apparatus of Modification 4 configured as described above will be described.
As shown in FIG. 10, the image processing apparatus 10 according to Modification 4 includes a rewritable memory (for example, SRAM, SDRAM, etc.) for storing reference data generated based on image data acquired by the image input I / F 12. ) Is provided.

  The image input I / F unit 12 scans image data acquired from the outside into a digital signal, transmits the coordinates (X, Y) and luminance values of each pixel to the binarization processing unit 14, and acquires, for example, The reference data representing the background is generated from the obtained image data and the past reference data stored in the image storage unit 60, and stored in the image storage unit 60.

  When the binarization processing unit 14 receives a write signal from the image input I / F unit 12, the binarization processing unit 14 acquires the image data transmitted from the image input I / F unit 12 every four pixels and sequentially stores them in the image storage unit 60. A difference in luminance value from the reference data based on the image data one screen before is calculated, and difference data is calculated.

  Then, based on the difference between the calculated luminance values, each pixel constituting the difference data is converted to a pixel (a white pixel in the present embodiment) as a background element or a moving object by a preset threshold value. The result is converted into binary data representing any of the graphic pixels (black pixels in the present embodiment) that are the elements of the graphic to be represented, and the binarization result in which the background pixel is represented as 0 and the graphic pixel is represented as 1 The coordinates (X, Y) are transmitted to the pixel determination unit 16 together with the write signal every four pixels. When the write signal is received again, the next four pixels of data are acquired, and the same operation is repeated.

  As described above, in the image processing apparatus 10 according to the fourth modification, the binarization processing unit 16 calculates difference data from the difference between the acquired image data and the reference data sequentially stored in the image storage unit 60. Each pixel of the difference data is converted into binary data by a threshold value set with a luminance value.

  For this reason, according to the image processing apparatus 10 of the fourth modification, a pixel whose luminance value has changed between two pieces of image data, that is, a pixel of a moving object is detected, and a binarization result using the pixel as a graphic pixel Is output to the pixel determination unit 16, and the feature amount of the figure representing the moving object can be calculated.

  Further, as a fifth modification of the image processing apparatus 10 according to the present embodiment, the pixel determination unit 16 determines that when the pixels in the four directions (up, down, left, and right) around the target pixel made up of graphic pixels are graphic pixels, You may comprise so that it may judge that each figure pixel is connected.

  Hereinafter, the pixel determination unit 16 that performs pixel determination as described above will be described. Similarly to the pixel determination unit 16 of the above embodiment, the pixel determination unit 16 of Modification 5 receives the write signal transmitted from the binarization processing unit 14 and receives the four pixels transmitted from the binarization processing unit 14. The obtained four pixels are searched in order from the pixel having the smallest coordinate (X, Y) value, and a graphic pixel is extracted from the binarization result of each pixel. Then, when the graphic pixel is extracted, the graphic pixel is set as the target pixel, and as shown in FIG. 11A, a total of 3 of the target pixel and the two neighboring pixels (upper and left two pixels) of the target pixel. From the binarization result of one pixel, the pixel of interest is classified based on the pixel determination flag issuing rule shown in FIG. 11B, and one of the pixel determination flags is set, and the pixel determination flag described below Perform actions based on. When the operation based on the pixel determination flag ends, the next graphic pixel is extracted, and the pixel determination flag set and the pixel determination are performed for all graphic pixels included in the four pixel data, with the extracted pixel as a new target pixel. Perform an action based on the flag. When the operation for all the graphic pixels included in the data of the four pixels is completed, the operation is stopped until the next write signal is received. FIG. 11A is an explanatory diagram showing three pixels used for classifying a target pixel, and FIG. 11B is an explanatory diagram showing a pixel determination flag issuing rule.

Hereinafter, the operation of the pixel determination unit 16 will be described in detail for each pixel determination flag to be issued.
The pixel determination unit 16 detects the binarization result of the neighboring two pixels, and the binarization result of the neighboring two pixels is the pattern shown in the first row of FIG. 11B, that is, around the target pixel. If it is determined that there is no graphic pixel and the target pixel is a pixel forming a new graphic, the pixel determination flag N is set, and the coordinates (X, Y) of the target image are output to the new label issuing unit 32. .

  Further, the pixel determination unit 16 has a binarization result of two neighboring pixels as a pattern shown in the second row in FIG. 11B, that is, the pixel of interest forms a part of an existing figure. If it is determined that there is a possibility of linking existing figures, the pixel determination flag C is set, and two types of label numbers L1 and L2 (where L1 ≦ L2) of two neighboring pixels are set. Label number L1 and the coordinates (X, Y) of the target pixel are output to the temporary feature amount updating unit 36. Then, it is determined whether or not the label number L1 is equal to the label number L2. If not equal, the label number L1 and the label number L2 (assuming that L1 <L2) are further stored in the connection information storage unit 34. Output.

  Then, the pixel determination unit 16 determines that the binarization result of the neighboring two pixels is one of the patterns shown in the third row of FIG. 11B, that is, the target pixel is a part of the existing figure. If it is determined that the pixel is to be formed, the pixel determination flag D is set, and the label number L1 of the two neighboring pixels and the coordinates (X, Y) of the target pixel are output to the temporary feature amount update unit 36.

  In the image processing device 10 of the modification 5 configured as described above, the pixel determination unit 16 determines whether or not the target pixel is a graphic pixel, and when the target pixel is a graphic pixel, the target pixel is determined. From the binarization result of a total of three pixels including the two neighboring pixels (upper and left two pixels) of the target pixel, the target pixel is classified based on the pixel determination flag issuing rule, Either one is set, and the operation based on the pixel determination flag is executed.

  For this reason, according to the image processing apparatus 10 of the modification 5, for example, when two objects are connected in an oblique direction, it is determined that the figures representing these objects are not connected. It can be distinguished from another object. Further, since the pixel of interest can be classified by identifying the binarization result of the two neighboring pixels, the operation time can be shortened compared to the case of identifying the binarization result of the four neighboring pixels.

  Further, in the image processing apparatus 10 of the present embodiment and the modification, the area of the figure is calculated as the feature amount of the figure. However, the area is not necessarily the area, and the center of the figure, the perimeter of the figure, the average of the figure A rectangle including luminance and a figure may be calculated.

  In this way, as in the case of calculating the area of the graphic, the feature amount of the graphic can be calculated by one scan. Therefore, it is necessary to provide an area for storing image data and store the image data. There is no convenience. In addition, when an area for storing image data can be secured, if the image data is temporarily stored in this area and can be scanned again, for example, two or more scans such as dispersion of graphics are required. It is also possible to calculate the feature quantity of the figure.

  On the other hand, in the image processing apparatus 10 according to the present embodiment and the modification, the binarization processing unit 14 acquires image data every four pixels and transmits the image data every four pixels. When there is a margin in speed, it may be acquired for each pixel and transmitted for each pixel.

  In the present embodiment, each unit except for the control unit 40 is configured by one LSI 50 including a clock generation unit. However, it is configured by software such as a microcomputer that includes a CPU, a RAM, and the like and executes a predetermined program. However, it is also possible to execute the processing of each unit.

It is a schematic block diagram showing the structure of the image processing apparatus 10 of this embodiment. It is explanatory drawing which shows the pixel determination flag issue rule based on four neighboring pixels. It is explanatory drawing which shows the label table | surface 22, the connection map 24, and the feature-value table | surface 26 which comprise the memory | storage part 20 for calculation. It is explanatory drawing showing the rewriting operation | movement of the memory part 20 for a calculation which the temporary feature-value update part 36 performs. 3 is an explanatory diagram showing details of a register 28 and a line buffer 30. FIG. It is a flowchart which shows operation | movement of the feature-value update part 38 of this embodiment. FIG. 10 is an explanatory diagram for explaining an operation performed by a feature amount update unit. 10 is a flowchart illustrating an operation of a feature amount update unit 38 according to Modification 1; It is a schematic block diagram showing the structure of the image processing apparatus 10 of the modification 2. It is a schematic block diagram showing the structure of the image processing apparatus 10 of the modification 4. It is explanatory drawing which shows the pixel determination flag issue rule based on 2 neighboring pixels.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Image processing apparatus, 12 ... Image input I / F part, 14 ... Binarization process part, 16 ... Pixel determination part, 18 ... Memory | storage part, 20 ... Memory | storage part for calculation, 22 ... Label table, 24 ... Connection map , 26 ... feature quantity table, 28 ... register, 30 ... line buffer, 32 ... new label issuing section, 34 ... linked information storage section, 36 ... temporary feature quantity update section, 36a ... new pixel update section, 36b ... additional pixel update Unit 38, feature amount updating unit 40, control unit 50, LSI, 60 image storage unit 90, external device

Claims (16)

An image processing apparatus that scans image data composed of a plurality of pixels arranged in a vertical direction and a horizontal direction, and calculates a feature amount of a graphic composed of connected pixel groups included in the image data,
A pixel information storage unit in which a label which is identification information of the graphic is stored for each pixel constituting the image data;
Binarization means for converting the image data into binary data representing whether each pixel is a graphic pixel as an element of the graphic or a background pixel other than the graphic pixel;
The image data converted into binary data by the binarization means is scanned, and the graphic pixel is sequentially extracted as a target pixel from the image data, and the labels of the pixels around the extracted target pixel are the pixel information. Based on the labels of the surrounding pixels acquired from the storage means, the target pixel is a new pixel that forms a new graphic, or an additional pixel that becomes a part of a graphic already formed Pixel determining means for determining whether or not the target pixel is an additional pixel, and further determining whether or not the additional pixel is a connected pixel that connects a plurality of different figures;
A new label issuing means for newly issuing a label when the pixel determining means determines that the target pixel is the new pixel;
When the pixel determination unit determines that the target pixel is the connected pixel, the connection information storage unit stores connection information indicating that a plurality of figures are connected to the connection pixel in the connection information storage unit. When,
When the pixel determining unit determines that the target pixel is the new pixel, the new label issued by the new label issuing unit is stored in the label storage unit, and the graphic specified by the new label is stored. An initial value setting means for storing an initial value as a feature quantity in the feature quantity storage means, and further storing the new label as a label of the target pixel in the pixel information storage means;
When the pixel determination unit determines that the target pixel is the additional pixel, among the feature amounts of the figure stored in the feature amount storage unit, the feature amount of the figure that the additional pixel is a part of is calculated. Updating an existing value updating means for storing in the pixel information storage means a label of a graphic of which the additional pixel is a part thereof, as a label of the target pixel;
Scanning of image data for one screen by the pixel determination unit is completed, and the new label issuing unit, the connection information storage unit, the initial value setting unit, and the existing value update unit are based on the determination of the pixel determination unit. When the operation is completed, based on the connection information stored in the connection information storage means, the label of each figure recognized as the same figure among the labels stored in the label storage means is rewritten and matched, Among the feature quantities of the figure stored in the quantity storage means, the feature quantities of each figure specified by each label before the labels are matched are synthesized, and the synthesized feature quantities are used as the matched labels. Feature quantity updating means for storing in the feature quantity storage means as the feature quantity of the figure specified by
The equipped Rutotomoni,
A plurality of storage means groups comprising the label storage means, the connection information storage means, and the feature quantity storage means;
The initial value setting means, the existing value updating means, and the connection information storage means commonly use one of the plurality of storage means groups, and for each image data for one screen, Change the storage means group to be used in order,
The image processing apparatus, wherein the feature amount updating unit sequentially selects the plurality of storage unit groups and performs processing based on data stored in the selected storage unit group . An image processing apparatus that scans image data composed of a plurality of pixels arranged in a vertical direction and a horizontal direction, and calculates a feature amount of a graphic composed of connected pixel groups included in the image data,
  A pixel information storage unit in which a label which is identification information of the graphic is stored for each pixel constituting the image data;
  Binarization means for converting the image data into binary data representing whether each pixel is a graphic pixel as an element of the graphic or a background pixel other than the graphic pixel;
  The image data converted into binary data by the binarization means is scanned, and the graphic pixel is sequentially extracted as a target pixel from the image data, and the labels of the pixels around the extracted target pixel are the pixel information. Based on the labels of the surrounding pixels acquired from the storage means, the target pixel is a new pixel that forms a new graphic, or an additional pixel that becomes a part of a graphic already formed Pixel determining means for determining whether or not the target pixel is an additional pixel, and further determining whether or not the additional pixel is a connected pixel that connects a plurality of different figures;
  A new label issuing means for newly issuing a label when the pixel determining means determines that the target pixel is the new pixel;
  When the pixel determination unit determines that the target pixel is the connected pixel, the connection information storage unit stores connection information indicating that a plurality of figures are connected to the connection pixel in the connection information storage unit. When,
  When the pixel determining unit determines that the target pixel is the new pixel, the new label issued by the new label issuing unit is stored in the label storage unit, and the graphic specified by the new label is stored. An initial value setting means for storing an initial value as a feature quantity in the feature quantity storage means, and further storing the new label as a label of the target pixel in the pixel information storage means;
  When the pixel determination unit determines that the target pixel is the additional pixel, among the feature amounts of the figure stored in the feature amount storage unit, the feature amount of the figure that the additional pixel is a part of is calculated. Updating an existing value updating means for storing in the pixel information storage means a label of a graphic of which the additional pixel is a part thereof, as a label of the target pixel;
  Scanning of image data for one screen by the pixel determination unit is completed, and the new label issuing unit, the connection information storage unit, the initial value setting unit, and the existing value update unit are based on the determination of the pixel determination unit. When the operation is completed, based on the connection information stored in the connection information storage means, the label of each figure recognized as the same figure among the labels stored in the label storage means is rewritten and matched, Among the feature quantities of the figure stored in the quantity storage means, the feature quantities of each figure specified by each label before the labels are matched are synthesized, and the synthesized feature quantities are used as the matched labels. Feature quantity updating means for storing in the feature quantity storage means as the feature quantity of the figure specified by
  With
  Two storage means groups comprising the label storage means, the connection information storage means, and the feature quantity storage means are provided,
  The initial value setting means, the existing value update means, and the connection information storage means commonly use one of the two storage means groups, and for each image data for one screen, Change the storage means group to be used alternately,
  The image processing apparatus according to claim 1, wherein the feature amount updating unit alternately selects the two storage unit groups and performs processing based on data stored in the selected storage unit group. The connection information is obtained by setting a connected state or a non-connected state for all combinations of two labels among all labels issued by the new label issuing means. It is information that represents whether or not the figures specified in (2) are connected, and as an initial state, a non-connected state is set for all label combinations,
When the pixel determination unit determines that the target pixel is the connection pixel, the connection information storage unit sets a connection state for a combination of labels of the figures connected by the connection pixel. The image processing apparatus according to claim 1 or 2 . The new label issuing means issues a code representing an issue order as the label,
The initial value setting means stores the labels in the label storage means in the order of issue,
The feature amount update means includes:
From the connection information storage means, the state set for the combination of labels is read in order from the combination including the label with the highest issuance order,
When the combination of the labels set with the connection state is extracted, the label storage means rewrites the region in the issue order of the label with the later issue order among the combination of labels set with the connection state. To determine whether or not
In the label storage means, when the issue order area of the label with the later issue order is rewritten, among the labels stored in the issue order areas of the label combination, the issue order is earlier Rewrite the area in which the label with the later issuance order is stored among the respective issuance order areas,
From the label storage means, an area where the label with the later issue order is stored is detected, and among the labels stored in the respective issue order areas of the combination of labels, the label with the earlier issue order, Rewriting the detected area;
On the other hand, in the label storage means, when the issue order area of the label with the lower issue order is not rewritten, the area is stored in the issue order area of the label with the earlier issue order. Rewrite to label,
The image processing apparatus according to claim 3 . The pixel information storage means includes
A register capable of storing a pixel label for one pixel;
A line buffer capable of storing labels of pixels for one scan line;
Consists of
The initial value setting unit and the existing value updating unit temporarily store the label of the target pixel in the register, and when storing the label of the next target pixel, the label of the previous target pixel stored in the register and after storing in a region corresponding to the previous target pixel in said line buffer, to any one of claims 1 to 4, characterized in that storing a label of the next pixel of interest in the register The image processing apparatus described. The image processing apparatus according to any one of claims 1 to 5, characterized in that it comprises a first image acquiring means from the imaging device configured to be acquiring the image data directly. The image processing apparatus according to any one of claims 1 to 6, characterized in that it comprises a second image acquiring unit that the image data has been acquired can configure the image data from the stored storage medium . The binarization means acquires the image data for each of a plurality of pixels, converts the image data into binary data, and transmits the image data converted to the binary data to the pixel determination means for each of the plurality of pixels. the image processing apparatus according to any one of claims 1 to 7. The image processing apparatus according to claim 8 , wherein the binarization unit acquires and transmits the image data for every four pixels as the plurality of pixels. It said binarizing means, the image data, the image according to any one of claims 1 to 9, characterized in that to convert the binary data by the threshold set by the luminance value Processing equipment. The binarization means generates difference data consisting of a difference between the image data and background data stored in advance in the storage means as new image data, and the image data is set with a luminance value. the image processing apparatus according to any one of claims 1 to 9, by threshold and converting the binary data. The pixel determining means, any one of claims 1 to 11, characterized in that the graphic pixels adjacent around eight directions around the pixel of interest is determined as a pixel that is connected to the pixel of interest An image processing apparatus according to 1. The pixel determining means, the graphic pixels adjacent in four directions of the horizontal and vertical directions around the pixel of interest, of claims 1 to 11, characterized in that to determine the pixels connected to the pixel of interest The image processing apparatus according to any one of the above. According to any one of claims 1 to 13, wherein said connection information storage means, the initial value setting means and the existing value updating means, that is operatively configured each independently Image processing apparatus. The feature amount of a figure calculated by the feature amount update unit is a feature amount of a figure that can be calculated by the pixel determination unit scanning the image data once. the image processing apparatus according to any one of claim 14. The image processing apparatus according to claim 15 , wherein the feature amount of the figure calculated by the feature amount update unit is any one of an area, a center of gravity, a perimeter length, an average luminance, and an enclosing rectangle of the figure. .