JPH0119190B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a recognition device for figures, symbols, etc., and in particular, the present invention relates to a device for recognizing figures, symbols, etc. This invention relates to a figure separation circuit that separates figures and symbols whose size is smaller than a specified size.

Conventional methods for inputting figures and symbols into a calculator include inputting the outer periphery of the figure as a sequence of points using a coordinate input device such as a tablet or digitizer, or using keys such as an alphanumeric, kana, or kanji keyboard. The method is widely used.

However, with recent advances in scanner technology and pattern recognition processing technology, methods have been attempted to recognize desired figures and symbols from two-dimensional image data obtained by scanning drawings, etc. with a scanner, and input them into a computer. It is becoming more and more popular. In commonly used drawings such as circuit diagrams, there are a mixture of figures and symbols with different geometric dimensions and recognition methods.
In order to perform the subsequent recognition processing efficiently, it is necessary to separate these different substances prior to the recognition processing.

Conventionally, when shapes and symbols are mixed like this, the relative positional relationship between the shapes and symbols must be defined, such as writing the symbol on the right side of the shape, or the symbol can be written freely in the shape, but only in a specified position. By setting restrictions such as not being able to write, drawing specifications were stipulated so that there was no need to separate figures and characters, or even if it was necessary to separate them, they could be easily separated. However, setting such restrictions,
From the perspective of drawing a drawing, it lacks flexibility,
Moreover, it also has the disadvantage that it can only be used in limited ways. Furthermore, in the past, the figures and symbols to be separated were limited to those composed of continuous pixels. For example, for separable symbols such as kanji that are separated into "hen" and "tsukuri", "hen" and "tsukuri" are separated separately, so after separating them into one meaningful figure or symbol, It had the disadvantage of requiring re-editing.

The purpose of the present invention has been made in consideration of these circumstances, and it is a drawing in which figures and symbols, such as figures and characters, are freely drawn with geometrical dimensional features, and figures and symbols that require different processing methods thereafter. An object of the present invention is to provide a figure separation circuit capable of separating figures and symbols into meaningful units based on differences in geometric dimensions from two-dimensional image data obtained by scanning the image with a scanner.

According to the present invention, from two-dimensional image data obtained by scanning drawings, etc. that include figures and symbols with different dimensions using a scanner, etc., images of sizes less than or equal to the specified distance that exist in close proximity than the specified distance are obtained. A circuit for separating and extracting image data corresponding to a figure part includes an image memory for storing and holding the two-dimensional image data, and a raster scanning circuit for raster scanning the image memory in response to an outer edge point detection start signal. an outer edge point detection circuit that detects an outer edge point of a portion;
a figure outer edge tracking circuit that tracks the outer edge of a figure portion while sequentially detecting outer edge points adjacent to the outer edge point obtained by the outer edge point detection circuit; and a rectangle circumscribing the figure area obtained by the figure outer edge tracking circuit. Set a separation frame that is a region, or expand the separation frame side by side based on extension information of the separation frame obtained by the separation judgment circuit described later, set a new separation frame, and use the separation frame and the external information processing device in advance. A separation frame that compares the separation frame with the maximum frame of a set separation figure and generates an outer edge point detection start signal if the separation frame exceeds the maximum frame, and generates an image memory readout start signal if the separation frame does not exceed the maximum frame. a setting circuit; an image memory reading circuit that receives the image memory read start signal and sequentially reads values in the image memory corresponding to points obtained by expanding each side of the separation frame except for the upper side; and a value obtained by the image memory reading circuit. A separation function that determines whether or not it is possible to separate the figures based on the value of the signal, and outputs a cutting start signal if separation is possible, and outputs extended information for each side of the separation frame if separation is not possible. A determination circuit, a figure cutting circuit which receives the cutout start signal and cuts out the figure part included in the current separation frame and its circumscribed rectangular area, reads data from and writes data to and from the image memory, and performs external information processing. A graphic separation circuit is obtained which includes an interface operation with the device and a control circuit which controls each of the circuits and generates signals such as an edge point detection start signal for performing the graphic separation operation.

According to the figure separation circuit of the present invention, there is no need to control the drawing specifications of drawings, such as drawing figures or symbols to be separated in predetermined positions or frames. The burden can be greatly reduced, and therefore the number of man-hours required for creating drawings can be reduced.

Further, according to the figure separation circuit of the present invention, when one figure or symbol is composed of a plurality of image data parts, such as the Kanji characters "hen" and "tsukuri", each image data is separated individually. - It is possible to cut out all the image data existing within a specified area as one figure or symbol without cutting out. Therefore, even for two-dimensional image data such as drawings that include figures and symbols that are composed of multiple image data parts, there is an advantage that the recognition process after cutting out the figures and symbols is simplified. arise.

The present invention will be described in detail below with reference to the drawings.
FIG. 1 is a diagram showing a part of a flowchart as an example of a drawing in which figures and symbols of different geometric dimensions coexist. The diagram includes processing blocks, large continuous figures represented by connecting lines connecting each block, and small symbols represented by characters written inside and outside each block.

FIG. 2 is a diagram illustrating a method for detecting the outer edge points of a figure part or a symbol part as shown in FIG. 1. Note that in the following explanation, the target two-dimensional image data will be treated as a binary pattern with values of 1 and 0. A black circle in the figure indicates a dot having a value of 1 in the image data. The white circle indicates a one-dot point with a value of 0 in the image data, and for the sake of simplicity, only those necessary for explanation have been described, but the other descriptions have been omitted.

Detection of the outer edge points of a graphic or symbol portion represented by a group of black circles is obtained by raster scanning the entire two-dimensional image data as shown by the dotted line in FIG. First, scan the i-th line from left to right and find the point where the value changes from 0 to 1 (white to black).
Store that position a. Next, check the values of the two points at positions a-1 and a for the i->1st line, and if the values of all points are 0, the point a of the i-th line is one of the black circles representing figures and symbols. This means that it has been detected as an outer edge point. However, if at least one point at position a-1, a on the i-1th line has a value of 1, the point at position a on the i-th line is either after the outer edge that has already been tracked or at the inner edge. Therefore, the raster scanning of the two-dimensional image data is continued without performing any processing subsequent to the outer edge detection.

FIG. 3 is shown to explain a method of finding a separation frame represented by a rectangle circumscribing the black circles by tracing the outer edge points starting from the outer edge points of the black circles detected by the method explained using FIG. 2. It is a diagram. Figure 3a
2 is a diagram showing a state in which the outer edge portion is tracked starting from the outer edge point detected in FIG. 2. FIG. The tracking of the outer edge is started from point 1 shown in FIG.
Along with the tracking process, the positions of the tracked points 1, 2, 3, . . . are stored. FIG. 3b is a diagram showing a state in which a separation frame, which is a rectangular area circumscribing a group of black circles, is set based on the position information of the outer edge obtained by the outer edge tracking in FIG. 3a. In the figure, the width of the separation frame is 7.
The height is 5 bits.

FIG. 4 is a diagram showing a state in which the separation frame detected in FIG. 3 is expanded by a preset width in each side direction except for the upper side. Figure 4a shows the expanded separation frame (indicated by a solid line) and the separation frame before expansion (indicated by a dotted line).
FIG. Read the value of the expanded part, which is the area between the frame shown by the solid line and the frame shown by the dotted line in Figure 4a, and if there is no part with a value of 1, the separation before expansion shown by the dotted line. The black circles can be completely separated by the frame. FIG. 4b shows a case where the expanded portion includes a portion with a value of 1, that is, a group of black circles indicating other graphic portions or symbol portions. In this case, the outer edges of the black circles newly detected by expanding the separation frame are tracked in the same manner as shown in FIG. 2a. Using the position information of the new black circle group obtained as a result of tracking, a new circumscribed rectangular area is set that includes both the black circle group surrounded by the separation frame indicated by the dotted line and the newly detected black circle group. Separate frame. FIG. 4c is a diagram showing the separation frame newly set in this way.

The above-described steps are repeated to sequentially expand the separation frame in the direction of each side except the top edge, and repeat until no new part with a value of 1 is detected in the expanded area. No expansion beyond the maximum possible dimensions is necessary. Therefore, each time the separation frame is expanded and a new separation frame is determined, the dimensions of the maximum frame and the separation frame are compared, and when the size of the maximum frame is exceeded, it is determined that separation is impossible, and this is explained in Fig. 2. Returning to the outer edge point detection process, the outer edge point of the next figure or symbol is detected.

An example will be described in which the method described above is used to separate and extract characters written inside and outside the processing blocks in the flowchart shown in FIG. 1.
FIG. 5 is an enlarged view of a part of the flowchart shown in FIG. 1 for explanation. The black solid line portion in the figure is a portion where the value is 1 in the image data, and the other portions are areas where the value is 0. For simplicity, the image data is not expressed in units of dots, but in the form of continuous line information. The entire two-dimensional image data is raster scanned as indicated by the dotted line in FIG. While scanning the i-th line from left to right, a point where the value changes from 0 to 1 (white to black) is found and its position a is memorized. Next, since the values of the two points at positions a-1 and a on the i-1th line are both 0, the point at position a on the i-th line is detected as the outer edge point of the black data group representing the figure. That means that. Position a of the outer edge point of the next detected black data group
Starting from , trace the outer edge of the black data group.
Here, the outer edges are the outermost black dots of the ellipses, rectangles, and rhombuses representing processing blocks, and the connecting lines with arrows connecting each processing block. The outer edge is traced until it returns to the starting point position a, and from the information on each point of the outer edge obtained during tracking, the black data group, in this case a figure group consisting of each processing block and connection line, is circumscribed. A separation frame, which is a rectangular area, is also set. Since the size of the separation frame is larger than the maximum frame of the geometric dimensions of the characters to be separated, it is determined that it is impossible to separate the black data group obtained by tracking from point a, and raster scanning is performed. Continue to perform the outer edge point detection operation for the next figure.

When raster scanning is further performed, the jth line is scanned from left to right and a change point where the value changes from 0 to 1 is found at position b, but the point at position b starts from position a on the ith line. Since this point has already been tracked during the outer edge tracking operation, raster scanning on the i+1th line is continued. Further, when scanning the j-th line, a point where the value changes from 0 to 1 is found at position c. When the values of the two points at positions c-1 and c on the j-1st line are both 0, the point at position c on the j-th line is detected as the outer edge point of the black data group representing the figure. become.

Next, by tracing the outer edge of the black data group starting from the position c of the outer edge point of the detected black data group, it circumscribes the black data group, in this case, the hen ``woman'' of the character ``beginning''. A separation frame 1, which is a rectangular area indicated by a dotted line in the figure, is set. Next, the set separation frame 1 is expanded by a preset width in each side direction except for the upper side. A separation frame 2 indicated by a dashed line in the figure is an expanded separation frame. Since there is a part with a value of 1 in the expanded part, in this case a part of the ``base'' of the character ``Hajime'', the outer edge of the newly detected black data group is traced. Using the position information of the new black data group obtained as a result of tracking, a circumscribed rectangular area including both the black data group surrounded by the separation frame 1 shown by the dotted line and the newly detected black data group is created. The newly set separation frame 3 is indicated by a dotted line in the figure. If the separation frame 3 is further expanded by a preset width and no new black data group is included, the data within the area surrounded by the separation frame 3, in this case the character ``Hajime'', is separated and extracted. Process as data. As shown here, the feature of this method is that it is possible to separate and extract separated characters, such as the kanji character that is divided into "hen" and "tsukuri", into meaningful single character units.

Since the separation of one character is completed, scanning of the jth line is continued to detect the outer edge point of the next figure. Next, the point at position d where the value changes from 0 to 1 is
Since either the point at the j-1th position d or the point at position d-1 has a value of 1, it is determined that it is an inner edge point of the symbol block. Therefore, the raster scanning of the two-dimensional image data is continued without performing any subsequent processing following the detection of the outer edge.

By performing the above-described operations over the entire flowchart shown in FIG. 1, the characters in the flowchart can be separated and extracted.

FIG. 6 is a block diagram when the graphic separation circuit according to the present invention is used in an automatic drawing reading device. In FIG. 6, reference number 4 is an information processing device, 5 is a drum scanner, 6 is a binarization circuit, 7 is a graphic separation circuit according to the present invention, and 8 is a graphic recognition circuit.

The drum scanner 5 is controlled by the information processing device 4 through a signal line 100, scans the drawing to be read attached to the drum surface, and sends the obtained density value, that is, multivalued data, to the binarization circuit 6 through the signal line 200. . The binarization circuit 6 is controlled by the information processing device 4 through a signal line 300, and binarizes the multivalued data sent from the drum scanner 5 through the signal line 200 through the signal line 300 using a threshold value preset by the information processing device 4. , signal line 40
0 to the graphic separation circuit 7. After the graphic separation circuit 7 stores the previous binary data sequentially sent through the signal line 400 in a built-in binary image memory,
Figures within specified geometric dimensions are separated according to various parameters sent from the information processing device 4 through the signal line 500, and two-dimensional image data representing the binary pattern is sent to the figure recognition circuit through the signal line 600. Send to 8. The figure recognition circuit 8 recognizes the two-dimensional image data representing the binary pattern of the separated figures by a method such as pattern matching, and sends the recognition result to the signal line 7.
00 to the information processing device 4.

FIG. 7 is a block diagram of the main part of the graphic separation circuit 7 according to the present invention. Reference number 7 in FIG.
0 is a control circuit, 71 is an outer edge point detection circuit, 72 is an outer edge tracking circuit, 73 is a separation frame setting circuit, 74 is an image memory reading circuit, 75 is a separation determination circuit, 76 is a figure cutting circuit, and 77 is an image memory It is. As described in the explanation of FIG. 2, the outer edge point detection circuit 71 is configured to raster scan the image memory 77 through the signal line 707 and detect the outer edge points of a group of black circles corresponding to a figure. Every time an outer edge point of a group of black circles is detected, the position coordinates of the detected outer edge point are sent to the outer edge tracking circuit 72 through a signal line 702.

The outer edge tracking circuit 72 tracks the outer edge of the separation starting from the outer edge point detected by the outer edge point detection circuit 71 as described in the explanation of FIG. It is output to the separation frame setting circuit 73.

The separation frame setting circuit 73 sets the initial separation frame from the position information of each point on the outer edge detected by the outer edge tracking circuit 72 as described in the explanation of FIG. As mentioned above, the values read by the image memory reading circuit 74 are sent to the separation judgment circuit 7.
The separation frame is configured to be expanded side by side based on the expansion information of the separation frame obtained by the determination in step 5.
In addition, the separation frame setting circuit 73 determines whether the new separation frame obtained by expansion exceeds the maximum frame size previously given by the information processing device 1 in the built-in maximum frame register, and If so, an outer edge point detection start signal is generated and sent to the control circuit 70, thereby causing the outer edge point detection circuit 71 to start detecting the outer edge point of the next group of black circles. On the other hand, if the current separation frame value is not exceeded, the value of the current separation frame is sent to the image memory reading circuit 74 through the signal line 704, and at the same time, an image memory reading start signal is sent to the control circuit 70, so that the image memory reading circuit 74 reads the image memory. Start the action.

As described in the explanation of FIG. 4, the image memory reading circuit 74 is configured to read values on the image memory corresponding to areas expanded in the direction of each side of the current separation frame except for the upper side. The read result is sent to the separation determination circuit 75 through a signal line 705.

The separation determination circuit 75 checks the values on the image memory corresponding to the newly expanded part, and if all the values are 0, the current separation frame value is sent to the signal line 706.
At the same time, a cutting start signal indicating that separation is possible is generated and sent to the control circuit 70 through the signal line 701.
The cutting operation in the figure cutting circuit 76 is started. On the other hand, if there is at least one point with a value of 1 in the area expanded in the three side directions excluding the top edge of the original separation frame, the separation frame is expanded to include the area with the value 1 as well. A separation frame extension signal is generated and sent to the separation frame setting circuit 73 via the control circuit 70.

The figure cutting circuit 76 cuts out binary data corresponding to the separating frame from the image memory 77 based on the value of the separating frame sent from the signal line 706, and transmits it to the automatic drawing reading device shown in FIG. As explained in the embodiment, the signal is sent to the figure recognition circuit 8 through the signal line 600.

The image memory 77 has a capacity to store a binary pattern for one drawing, and the fifth
The binary pattern sent from the binarization circuit 6 in the figure is stored.

As described above in the description of each of the circuits 71 to 77, the control circuit 70 transmits and receives commands and data to and from the information processing device 4, and controls each of the circuits 71 to 77 to perform the graphic separation operation.

As explained above using the drawings, according to the present invention, figures and symbols within specified dimensions can be separated from two-dimensional image data representing a drawing that includes a mixture of figures and symbols of different dimensions. Recognition processing can be performed efficiently when recognition methods are different.

Furthermore, if separated figures and symbols exist close to each other within a specified distance, they are cut out as one unified figure or symbol, so editing processing after separation and cutout is not necessary. Furthermore, when creating drawings, it is possible to create drawings with great flexibility by being freed from restrictions such as writing symbols etc. in lattice frames. In addition, since the graphic separation circuit of the present invention has a simple circuit configuration, it can be easily implemented in hardware, thereby realizing high-speed graphic separation.

[Brief explanation of drawings]

Figure 1 is a diagram showing an example of a drawing in which shapes and characters of different sizes coexist, Figure 2 is a diagram explaining a method for detecting outer edge points, and Figure 3 is a diagram showing how to detect a figure based on the detected outer edge points. FIG. 4 is a diagram explaining a method of expanding the separation frame to separate adjacent graphic parts together. FIG. FIG. 6 is a block diagram showing an embodiment in which the figure separating circuit of the present invention is applied to an automatic drawing reading device, and FIG. 7 is a block diagram specifically showing the figure separating circuit of the present invention. It is. In the figure, 1 to 3... set separation frame, 4
...Information processing device, 5...Drum scanner, 6...
...binarization circuit, 7...figure separation circuit, 8...figure recognition circuit, 70...control circuit, 71...outer edge point detection circuit, 72...outer edge tracking circuit, 73...separation frame expansion circuit, 74... Image memory reading circuit, 75
... Separation judgment circuit, 76 ... Figure cutting circuit, 77
...Indicates image memory.

Claims (1)

[Claims] 1. In a circuit that separates and extracts image data corresponding to a graphic part of a specified size or less that is located close to a specified distance or less from two-dimensional image data obtained by scanning a drawing, etc., the two-dimensional image an image memory for storing and holding data; an outer edge point detection circuit that receives an outer edge point detection start signal and detects outer edge points of the graphic portion while raster scanning the image memory; and the outer edge point detection circuit. A figure outer edge tracking circuit that tracks the outer edge of a figure part while successively detecting adjacent outer edge points from the obtained outer edge point, and separation of a rectangular area circumscribing the figure area obtained by the figure outer edge tracking circuit. A new separation frame is set by setting a frame or extending the separation frame side by side based on the extension information of the separation frame obtained by the separation determination circuit described later, and the separation set in advance between the separation frame and the external information processing device is performed. a separation frame setting circuit that compares the separation frame with the maximum frame of the figure and generates an outer edge point detection start signal again if the separation frame exceeds the maximum frame, and generates an image memory read start signal if the separation frame does not exceed the maximum frame; an image memory reading circuit that receives an image memory reading start signal and sequentially reads values in the image memory corresponding to points obtained by expanding each side of the separation frame except for the upper side; and a value of the signal obtained by the image memory reading circuit. a separation determination circuit that determines whether or not the figures can be separated based on the separation frame, and outputs a cutting start signal if separation is possible, and outputs extended information for each side of the separation frame if separation is not possible; A figure cutout circuit that receives a cutout start signal and cuts out a figure portion included in the current separation frame in its circumscribed rectangular area, reads and writes data to and from the image memory, and interfaces with an external information processing device. A figure separating circuit comprising: a control circuit for controlling the operation and each of the circuits and generating a signal such as an outer edge point detection start signal for performing a figure separating operation.