JP4853305B2 - Image processing apparatus and method - Google Patents

Description

  The present invention relates to a technique for converting a multi-valued image into a binary image.

  A color or grayscale multi-valued image may be converted into a binary image and output without being output as it is. At this time, for example, if simple binarization is applied to lightness information, there is a possibility that useful information such as characters may be lost mainly in a low contrast region. It is desired to obtain a binary image that retains useful information as it is.

  The following is a technical document related to the present invention.

  Japanese Patent Application Laid-Open No. 2004-228561 proposes to calculate a halftone portion, an edge portion, and a flat portion by calculating a saturation edge together with a brightness edge. Depending on the determination result, the character is edge-enhanced and the halftone dot is smoothed to reproduce an easy-to-view image.

  In Patent Document 2, an input document is separated into a character area, a photographic area, and a halftone dot area. In addition, a density value is set for determining whether or not a color image is included in the image data, an area above this density is determined as a photographic area, a low density area is determined as a character area, and spatial filter processing is performed. Proposed to perform gradation processing, edge enhancement processing, and simple binarization processing.

  Conventionally, a binary image created only from lightness information cannot reproduce a region with a chroma difference without a lightness difference. When a low-contrast portion is reproduced as a binary image, a region determination process such as a photographic region / character region is necessary.

Note that the above-described prior art and its problems are described only to explain a part of the background of the present invention. It should be noted that the present invention is not limited to the above-described prior art and problems.
JP 2000-357237 A Japanese Patent Laid-Open No. 2002-10094

  The present invention has been made in consideration of the above-described circumstances, and an object thereof is to provide an image processing technique for acquiring a binary image with little missing information from a multi-value image without performing processing such as region separation. It is said.

  According to the fundamental configuration of the present invention, a binary image that reproduces the level of brightness, a binary image that reproduces the edge of brightness (also referred to as an outline), and a binary image that reproduces the edge of saturation At least two of them are subjected to a logical operation to generate a binary image with little information loss.

  In this configuration, by directly referring to the lightness binary image and the saturation information binary image, it is possible to obtain a binary image in which a lightness low contrast region is reproduced without performing region determination. A visually natural binary image is obtained that reproduces a low-contrast region and has no missing information.

  The present invention will be further described.

  According to one aspect of the present invention, in order to achieve the above object, an image processing apparatus includes: an input unit that inputs a multi-valued image; a first binary image that represents the level of brightness of the multi-valued image; Generating means for generating at least two of a second binary image representing a lightness edge of the multivalued image and a third binary image representing a saturation edge of the multivalued image; Logical operation means is provided for performing a logical operation on at least two of the binary image of 1, the binary image of 2, and the third binary image and outputting the operation result.

  In this configuration, a binary image with few missing information is integrated without using region separation by using at least two of brightness information, brightness edge information, and saturation edge information in an integrated manner. Can be acquired.

  According to another aspect of the present invention, in order to achieve the above-mentioned object, the image processing apparatus includes: an input unit that inputs a multi-valued image; and a first two that represents the level of brightness of the multi-valued image. A value image, a second binary image representing the lightness edge of the multivalued image, a third binary image representing the saturation edge of the multivalued image, and the saturation of the multivalued image Generating means for generating at least two of the fourth binary images; the first binary image, the second binary image, the third binary image, and the fourth binary image; Logical operation means for performing a logical operation on at least two of them and outputting the operation results.

  Even in this configuration, it is possible to obtain a binary image with little information loss by using the information of brightness level, brightness edge information, and saturation edge information in an integrated manner.

  A binary image representing the height can be generated by threshold processing. The threshold processing may use a fixed threshold (also referred to as fixed binarization) or a floating threshold (also referred to as floating binarization). In the fixed binarization, an image is binarized with reference to a threshold prepared in advance for the image. The fixed threshold value may be automatically determined by Otsu's method using discriminant analysis (Otsu, “Automatic threshold selection method based on discriminant and least square criterion”, IEICE Transactions, Vol. J63-D, No. 4, pp. 349-356, 1980), but not limited thereto. Floating binarization is to binarize an image by adjusting a threshold value for each pixel of interest, for example, according to the density of a peripheral region.

  An image representing an edge is generated by, for example, floating binarization. The edge strength may be calculated using a filter and binarized by threshold processing.

  In the logical operation, the content of the operation may be changed depending on whether the image representing the brightness level is an edge around the target pixel.

  For example, in a binarized image with high and low lightness, when the lightness is high (white), OFF is set when the lightness is small (black), and in a lightness edge image, the edges are turned ON and the others are turned OFF. Perform exclusive OR processing. When both the lightness high / low binary image and the lightness edge binary image are ON, when the periphery of the pixel of interest corresponds to the edge of the image representing the lightness high / low, it is ON, black solid (filled with ON) When it corresponds to, it is turned OFF.

  For example, when the lightness binary image is ON and the saturation binary image is ON, when the periphery of the target pixel corresponds to the edge of the lightness binary image, ON, black solid (filled with ON) When it corresponds to, it is turned OFF.

  According to still another aspect of the present invention, in order to achieve the above-mentioned object, the image processing apparatus includes: an input means for inputting a multi-value image; a predetermined measure of a predetermined color space of the multi-value image; Generating means for generating a first binary image representing a level of the image and a second binary image representing an edge of a predetermined measure in a predetermined color space of the multi-valued image; and the first binary image and the above Logical operation means for performing a logical operation on two binary images and outputting the operation result is provided.

The predetermined color space is, for example, L ^* a ^* b ^* , but is not limited to this. A high and low binary image is based on, for example, a lightness value (L ^* ). Edges can be detected by thresholding the edge strength. For example, the edge intensity of saturation is calculated as the Euclidean distance between the edge intensity of the a ^* image and the edge intensity of the b ^* image, but is not limited thereto. Alternatively, the binary image of the edge may be based on, for example, metric saturation (((a ^* ) ² + (b ^* ) ² ) ^1/2 . It may be more.

  Even in this configuration, it is possible to obtain a visually natural binary image with less information loss by utilizing the height information and edge information in an integrated manner.

  Each means may be constituted by subdivided means, or conversely, may be constituted as an integrated means.

  The present invention can be realized not only as an apparatus or a system but also as a method. Of course, a part of the invention can be configured as software. Of course, software products used to cause a computer to execute such software are also included in the technical scope of the present invention.

  These and other aspects of the invention are set forth in the appended claims and will be described in detail below with reference to examples.

  According to the present invention, it is possible to acquire a binary image with little information loss without performing region separation.

  Examples of the present invention will be described below.

  FIG. 1 shows the basic configuration of the present invention and shows an image processing apparatus according to a first embodiment. The image processing apparatus according to the first embodiment can be realized in various modes such as a copying machine, an imaging apparatus, and a computer program.

1, the image processing apparatus includes an image input unit 10, a color space conversion unit 11, a smoothing unit 12, an element binary image generation unit 13, a logical operation unit 14, and the like. The image input unit 10 inputs multivalued image data defined in a predetermined color space. The input image data is, for example, R, G, B data, but is not limited to this. The color space conversion unit 11 converts the input image data into L ^* , a ^* , and b ^* data. Any type other than L ^* a ^* b ^* color system can be used as long as it can distinguish lightness information and color information. The smoothing unit 12 performs preprocessing such as removal of noise such as isolated points. For example, as illustrated in FIG. 2, the element binary image generation unit 13 includes a brightness high / low binary image generation unit 130, a lightness edge binary image generation unit 131, a saturation high / low binary image generation unit 132, and a saturation edge binary. A value image generation unit 133 is included to generate a lightness high / low binary image, a lightness edge binary image, a saturation high / low binary image, and a saturation edge binary image. The logical operation unit 14 performs a logical operation on two or more types of element binary images and outputs the result as a generated binary image. The logical operation executed by the logical operation unit 14 is, for example, exclusive OR processing, but any operation that stores both the high-low binary image and the edge binary image may be used, depending on the surrounding situation. The contents of the arithmetic processing may be appropriately changed.

The lightness high / low binary image generation unit 130 performs threshold processing on the lightness data (L ^* ), and generates data representing the lightness (L ^* ) data in binary. In the threshold processing, for example, the brightness data is binarized with a threshold prepared for each image by fixed (simple) binarization. Simple binarization is performed by, for example, Otsu's method using discriminant analysis (Otsu, “Automatic threshold selection method based on discriminant and least square criterion”, IEICE Transactions, Vol. J63-D, No. 4, pp. 349-356, 1980).

The lightness edge image generation unit 131 forms a binary image representing the edge of lightness data (L ^* ). For edge detection, edge strength is detected using an edge detection filter or the like, and this is subjected to threshold processing. Alternatively, floating binarization that adaptively determines a threshold value for each pixel may be employed. In order to detect an edge by floating binarization, for example, the floating threshold value may be set so that the threshold value decreases as the difference between the maximum value and the minimum value of the brightness of surrounding pixels increases. For example, a value th ₀ sufficiently larger than the difference between the maximum value and the minimum value of the brightness of surrounding pixels, an average value th _{ave of the} brightness at the periphery, and a maximum value Max, which are planned for an image portion not including an edge, Using the minimum value Min, the threshold value can be expressed as th0 + thave−1 / 2 · θ · (Max−Min) (where θ is an emphasis coefficient of about 1 or less).

The saturation high / low binary image generation unit 132 performs threshold processing on the saturation data (a ^* , b ^* ) to generate data representing the high / low of the saturation (a ^* , b ^* ) data in binary. The threshold value processing binarizes the saturation data with a threshold value prepared for each image by, for example, fixed (simple) binarization, or with a threshold value prepared for each pixel by floating binarization, similarly to the lightness data. The saturation data is, for example, metric saturation (((a ^* ) ² + (b ^* ) ² ) ^1/2 , but is not limited to this. Values using a ^* and b ^* data derived from one side are used. May be.

The saturation edge image generation unit 133 forms a binary image representing the edge of the saturation data (a ^* , b ^* ). In the edge detection, the edge intensity is detected by using an edge detection filter or the like as in the edge detection of the brightness data, and this is subjected to threshold processing. Alternatively, floating binarization may be employed.

  FIG. 3 shows an example of binarization processing of this embodiment, and details thereof are as follows. Please refer to the example of the image shown in FIG.

[Step S10]: Image data, for example, R, G, B image data is input.

  The example of the input image in FIG. 4 includes “training guidance” as content, and includes character information as a whole. The brightness is, for example, 0 to 255, the black part is “0” side, and when binarized, it becomes “OFF” (0), and the white part is “255” side, and when binarized, “ON” (1 ) The background of the “program” character in the center of the input image has the same brightness and saturation as the character. The triangular part is a background with high brightness and saturation. The circular part has low brightness and saturation. The character part of “2005 New Training” at the bottom consists of characters with slightly low brightness in the background of high brightness, and there is no saturation.

[Step S11]: The color space of the input image is converted from RGB to the L ^* a ^* b ^* color system.
[Step S12]: The smoothing unit smoothes each data of L ^* a ^* b ^* . The smoothing uses, for example, a median filter or adaptive smoothing processing in the vicinity of the target position, but is not limited thereto.
[Step S13]: An element binary image is generated.

  In the example of FIG. 4, a brightness high / low binary image (a), a lightness edge binary image (b), and a saturation edge binary image (c) are generated as element binary images.

  The brightness high / low binary image (a) is generated by, for example, fixed binarization. In the example of FIG. 4, the rectangular background and circular background of “program” are crushed in black (OFF).

  The brightness edge binary image (b) is generated by, for example, floating binarization. In the example of FIG. 4, the center “program” is white except for the outline of the background (ON). Although not clear in FIG. 4, as shown in FIG. 7, only the outline of the character of the brightness edge binary image is stored in black, and the inside of the outline is outlined.

In the saturation edge binary image (c), typically, the Euclidean distance between the edge intensity of the a ^* image and the edge intensity of the b ^* image is set as a pseudo saturation edge intensity, and the pseudo saturation edge intensity is expressed as a floating binary value. However, the present invention is not limited to this. Here, the Euclidean distance between the edge intensity of the a ^* image and the edge intensity of the b ^* image is referred to as c * edge intensity. By calculating the c * edge intensity, it is possible to extract an edge between regions where colors having the same brightness and saturation and different hues are adjacent to each other. If the colors are not adjacent to each other, metric saturation (((a ^* ) ² + (b ^* ) ² ) ^1/2 may be adopted as a simple method to detect an edge from this. In the example of Fig. 4, the contour of the portion with saturation is stored (Fig. 8) The image of the center "program" missing in the lightness high / low binary image (a) and lightness edge binary image (b) Note that the part is preserved.

[Step S14]: Perform a logical operation on the binary data of the brightness high / low binary image (a), the lightness edge binary image (b), and the saturation edge binary image (c). Typically, the edge binary images are ORed, and the high / low binary image and the edge binary image are exclusively ORed. The output image may be generated and selectable by various combinational logics.
[Step S15]: A final binary image is output (printed or displayed). In the example of FIG. 4, not only the lightness high / low binary image but also the lightness edge binary image and the saturation edge binary image are referred to, so that information loss is small.

  In the above operation example, the saturation high / low binary image is not described in detail, but this may be used.

  Next, an image processing apparatus according to a second embodiment showing the detailed configuration of the present invention will be described.

  FIG. 5 shows an image processing apparatus according to the second embodiment. In this figure, the image processing apparatus includes an image input unit 20, a color space conversion unit 21, an adaptive smoothing unit 22, a fixed threshold binarization unit 23, a floating binary. It includes a valuation unit 24, logical operation units 25 and 26, a median filter 27, an edge detection unit 28, a c * calculation unit 29, and the like.

  The image input unit 20 inputs RGB image data, for example. In FIG. 5, a low-lightness character “e” (no saturation) (upper left) in a high-lightness background, a slightly high-lightness character “e” (no saturation) in a high-lightness background (lower left) ), Slightly lighter character “e” (no saturation) (lower right) in low lightness background, and slightly lighter character “e” (same lightness, different color) in slightly lighter background A processing example of an image consisting of (degree) (upper right) is shown.

The color space conversion unit 21 converts RGB image data into the L ^* a ^* b ^* color system. The L ^* data output from the color space conversion unit 21 is adaptively smoothed and then supplied to the fixed threshold value binarization unit 23 and the floating binarization unit 24. The fixed threshold binarization unit 23 binarizes the brightness data. Details thereof will be described later. The output image of the fixed threshold binarization unit 23 stores the information of the upper left part of the image example. Other parts of the information are missing. The floating binarization unit 24 binarizes the brightness data and acquires the contour information. As the contour information, information on the upper left portion, the lower left portion, and the lower right portion of the image example is stored as contour information. The information in the upper right part is missing. The logical operation unit 25 performs logical operation, typically exclusive OR processing, on the output binary images of the fixed threshold binarization unit 23 and the floating binarization unit 24.

The a ^* data and b ^* data output from the color space conversion unit 21 are smoothed by the median filter 27. The edge detection unit 28 detects the edge strength of each of the a ^* data and b ^* data, and the c * calculation unit 29 calculates the Euclidean distance between the edge strengths of the a ^* data and b ^* data to obtain the c ^* edge strength. Output as (pseudo saturation edge intensity). The floating binarization unit 30 binarizes the c ^* edge intensity with a floating threshold value to form a saturation edge image. The saturation edge image holds information on the upper right part of the image example as edge information.

  The logical operation unit 26 performs logical operation (for example, exclusive OR) on the logical operation result of the brightness high / low binary image and the lightness edge binary image and the saturation edge binary image, and outputs the result.

  In the output image, the portion saved in the lightness high / low binary image is output as a normal solid character, not saved in the lightness high / low binary image, and the portion saved in the edge binary image is output as a contour character. There are few omissions.

  FIG. 6A shows an example of an image binarized by the fixed threshold th in FIG. The threshold th can be selected as a value that optimizes the variance of the brightness histogram using the method of Otsu.

  FIG. 7 shows an example of a brightness edge binary image. FIG. 8 shows an example of a saturation edge binary image. In FIGS. 7 and 8, a small white circle indicates a character, and the character is a white outline character. Of these, the outline may be lost depending on the size of the surrounding pixels.

  FIG. 9 shows an improved example of the fixed binarization process. In this method, an image having a halftone can be appropriately determined as a threshold value. An example of fixed binarization processing in FIG. 9 is as follows.

[Step S20]: A density (lightness) histogram is input.
[Step S21]: Calculate (1) peak position (maximum value), (2) valley position (minimum value), (3) peak area, etc. in the histogram.
[Step S22]: It is determined whether the maximum peak area is equal to or greater than the threshold value. If the determination is affirmative, the process proceeds to step S23, and if not, the process proceeds to step S24.
[Step S23]: Set the binarization threshold = (the valley position (minimum value) on the low brightness side of the maximum peak) (FIG. 10A). Thereafter, the process proceeds to step S29.
[Step S24]: A reference density range used in the Otsu method is set.
[Step S25]: A temporary threshold value is calculated using the method of Otsu within the reference concentration.
[Step S26]: It is determined whether or not the temporary threshold = the valley position (minimum value) of the histogram. If it is the valley position (minimum value), the process proceeds to step S27.
[Step S27]: Binarization threshold = temporary threshold is set. Thereafter, the process proceeds to step S29.
[Step S28]: Binarization threshold = (Set a valley position (minimum value) on a lower density side than the temporary threshold (FIG. 10B), and then proceed to Step S29.
[Step S29]: The image density (brightness) is binarized with a set threshold.
[Step S30]: A binary image is output.

  According to this binarization method, a threshold value can be appropriately set not only for a document composed of a background and characters (FIG. 10A) but also for a document having a halftone (FIG. 10B).

11 to 13 show examples of chroma edge detection methods. In this example, the edge intensity of each of the a ^* image and the b ^* image is calculated by, for example, the Robinson operator shown in FIG. The product sum of each of the six operators in FIG. 11 and the image is calculated, and the edge strengths in the eight directions corresponding to each operator can be calculated. For example, the maximum value is the edge strength. You may confirm adjacent coupling | bonding from the edge intensity | strength of each direction.

In this example, as shown in FIG. 13A, the pseudo chroma edge strength (c ^* edge strength) is calculated from the Euclidean distance of the edge strength of the a ^* image and the edge strength of the b ^* image (FIG. 12). In order to select an effective range for edge detection, c ^* edge strength is assigned to classes 0 to 255 as shown in FIG. A histogram is calculated for this class and binarized.

  FIG. 14 shows a calculation example of a brightness high / low binary image and a brightness edge binary image. In this example, the calculation result when the fixed threshold processing pixel is “ON” (black) and the floating binarization pixel is “ON” (edge) is adapted according to the situation of the brightness high / low binary image values of the surrounding pixels. (Indicated by an asterisk in FIG. 14). In this way, the inside of the contour is drawn in solid with “black” (ON) or “white” (OFF), so the display is natural.

  FIG. 15 shows a calculation example of a lightness binary image and a saturation binary image (saturation edge binary image). In this example, the calculation result when the lightness binary pixel is “ON” and the saturation binary pixel is “ON” is adaptively changed according to the lightness situation of the surrounding pixels (in FIG. 15, the asterisk). ). In this way, when the background area is divided by brightness or saturation, the boundary can be displayed with a line of “black” (ON) or “white” (OFF).

  In addition, this invention is determined based on description of a claim, and is not limited to the specific structure of the Example, a subject, and an effect. The present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a first embodiment of the basic configuration of the present invention as a whole. 3 is a diagram illustrating a configuration example of a part of the first embodiment. FIG. 3 is a flowchart for explaining an operation example of the first embodiment. 6 is a diagram illustrating an example of image processing according to Embodiment 1. FIG. It is a figure which shows more concrete Example 2 of this invention as a whole. FIG. 6 is a diagram illustrating a lightness high / low binary image according to a second embodiment. FIG. 10 is a diagram illustrating a brightness edge binary image according to the second embodiment. FIG. 10 is a diagram illustrating a saturation edge binary image according to the second embodiment. 10 is a flowchart for explaining an operation example of fixed binarization according to the second embodiment. FIG. 10 is a diagram illustrating an example of a fixed binarization threshold according to the second embodiment. It is a figure explaining the Robinson operator for calculating the edge intensity | strength of Example 2. FIG. It is a figure explaining the pseudo | simulation saturation edge intensity | strength of Example 2. FIG. FIG. 10 is a diagram illustrating correction of a histogram of pseudo saturation edge intensity according to the second embodiment. FIG. 10 is a diagram illustrating an example of a logical operation according to the second embodiment. FIG. 10 is a diagram illustrating an example of a logical operation according to the second embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image input part 11 Color space conversion part 12 Smoothing part 13 Element binary image generation part 14 Logic operation part 20 Image input part 21 Color space conversion part 22 Adaptive smoothing part 23 Fixed threshold value binarization part 24 Floating binarization Unit 25 logic operation unit 26 logic operation unit 27 median filter 28 edge detection unit 29 c ^* operation unit 30 floating binarization unit 130 lightness high / low binary image generation unit 131 lightness edge binary image generation unit 132 saturation high / low binary image Generation unit 133 Saturation edge binary image generation unit

Claims (6)

An input means for inputting a multi-valued image;
A first binary image representing the brightness level of the multi-valued image, and a second binary image representing the brightness edge of the multi-valued image or a third binary image representing the chroma edge of the multi-valued image. Generating means for generating one binary image of the value images;
Logical operation means for executing a logical operation on the two binary images generated by the generation means and outputting an image based on the operation result;
The image processing apparatus in which the logical operation is exclusive OR. In addition to the first binary image and one of the second binary image or the third binary image, the generation means includes the second binary image or the third binary image. Generate the other of the binary images,
The logical operation means includes the first binary image, one of the second binary image or the third binary image, the second binary image, or the third binary. The image processing apparatus according to claim 1, wherein an exclusive OR operation is performed with the other of the images. An input means for inputting a multi-valued image;
A first binary image representing the brightness level of the multi-valued image, and a second binary image representing the brightness edge of the multi-valued image or a third binary image representing the chroma edge of the multi-valued image. Generating means for generating one binary image of the value images;
Logical operation means for executing a logical operation on the two binary images generated by the generation means and outputting an image based on the operation result;
In the first binary image, black is the ON value, white is the OFF value, the edge is the ON value in the one binary image, and the non-edge is the OFF value,
The logical operation outputs an ON value when the first binary image at the target pixel position is an ON value and the one binary image is an OFF value, and the first binary image at the target pixel position is output. When the binary image is an OFF value and the one binary image is an ON value, an ON value is output, and the first binary image at the target pixel position is an OFF value. When the binary image is an OFF value, an OFF value is output, and when the first binary image is an ON value at the target pixel position and the one binary image is an ON value, the target pixel is output. When the first binary image at the peripheral position of the position is an ON value, an OFF value is output, and when the first binary image at the peripheral position of the target pixel position is an OFF value, the ON value is output. An image processing apparatus for outputting. An input means for inputting a multi-valued image;
A first binary image representing the level of lightness of the multivalued image; a second binary image representing the lightness edge of the multivalued image; and a third binary image representing the saturation edge of the multivalued image. Generating means for generating a binary image;
Exclusive OR operation result between the second binary image and the third binary image generated by the generating unit and the first binary image generated by the generating unit An image processing apparatus having logical operation means for performing an OR operation and outputting an image based on an exclusive OR operation result. An input step for inputting a multi-valued image;
A first binary image representing the brightness level of the multi-valued image, and a second binary image representing the brightness edge of the multi-valued image or a third binary image representing the chroma edge of the multi-valued image. A generating step for generating one of the value images and a binary image;
An output step of performing a logical operation on the two binary images generated in the generation step and outputting an image based on the calculation result;
An image processing method in which the logical operation is exclusive OR. Computer
An input means for inputting a multi-valued image;
A first binary image representing the brightness level of the multi-valued image, and a second binary image representing the brightness edge of the multi-valued image or a third binary image representing the chroma edge of the multi-valued image. Generating means for generating one binary image of the value images;
Logical operation means for executing a logical operation on the two binary images generated by the generation means and outputting an image based on the operation result;
A computer program for image processing to function as
The computer program for image processing, wherein the logical operation is exclusive OR.