JPH04262660A - Picture recognition output device - Google Patents

Abstract

PURPOSE:To realize a picture recognition device with a high graph recognition rate, the picture recognition output device outputting a graph picture in an accurate graph shape, for example, and a digital copying machine provided with the picture recognition output device. CONSTITUTION:A picture recognition device is provided with a character connection line segmentation circuit 2 reading a picture data stored in a picture memory and segmenting a prescribed area in the picture, a character recognition circuit 3 comparing a graphic picture in the area with a data of a graph picture registered in advance so as to recognize the graphic picture and a connection recognition circuit 4 detecting, e.g. a branch or swell of a line picture end ridge, recognizing an arrow, recognizing a line picture connecting areas and converting the result into a line picture data. The picture output device is provided with an output picture generating circuit 5 generating a synthesis picture of the graphic shape recognized by the picture recognition device with the line picture and outputs the picture data of the synthesis picture to, e.g. a digital copying machine, in which the data is printed out on a recording sheet.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention relates to an image recognition device that optically reads a figure created by printing or handwriting, or a manuscript image containing figure data, and recognizes the image. For example, the present invention relates to an image recognition output device that outputs images in accurate graphic shapes, and a digital copying machine equipped with the image recognition output device.

0002

2. Description of the Related Art Conventionally, a common method for obtaining a fair drawing has been to request an expert (tracer) with drafting ability to redraw the draft drawing into a fair drawing. However, the number of experts is limited, and some require a long time to create drawings, which often results in high costs. Based on these circumstances, figure recognition devices have been developed as copying devices for copying handwritten drawings, drawing reading devices, and input devices for CAD (automatic drawing devices). FIG. 20 is a block diagram showing a schematic configuration of one example. In the figure, 91 is an input device that reads a document image and converts it into an image signal, 92 is a vectorization device that converts the image signal into vector data, and 93
94 is a computer (host computer) that performs document image recognition processing by comparing the vector data with built-in graphic data, and 94 is an output device that outputs the graphic data of the image that has undergone graphic recognition to, for example, a printer.

0003

As shown in the figure, the figure recognition device is composed of a host computer and dedicated image processing hardware, so it is quite expensive. Moreover, its operation required a certain degree of computer expertise and an understanding of a complex operational flow. Moreover, the figure recognition rate of the figure recognition device is not very high, and therefore, in order to correct erroneously recognized figure data, it is necessary to rely on interactive correction processing by the host computer, reducing the usability of the figure recognition device. Ta. The present invention has been made to solve the problems of the prior art described above, and provides an image recognition device with a high figure recognition rate, an image recognition output device that outputs the recognized figure image in, for example, an accurate figure shape, and It is an object of the present invention to provide a digital copying machine equipped with the image recognition output device at a low cost.

0004

[Means for Solving the Problems] In order to solve the above problems, the present invention is an image recognition device that reads image data stored in an image memory, performs image processing, and recognizes the shape of an image. An image cutting means for cutting out a predetermined area; a figure image recognition means for performing figure image recognition by comparing a figure image in the cut out area with data of a figure image registered in advance; Alternatively, the apparatus includes line image recognition means that recognizes line images connecting cut out areas and converts them into line image data, in the same manner as detecting bulges and recognizing arrows. Also,
The image output device is a line image drawn based on a pre-registered figure shape corresponding to the figure image recognized by the figure image recognition means of the image recognition device and the data of the line image recognized by the line image recognition means. The image composing means has an image composing means for creating a composite image by composing the images, and outputs image data of the composite image composed by the image composing means. Furthermore, the digital copying machine has a document reading means that optically reads the document image and converts it into an electrical signal.
Image data read by the document reading means from a form manuscript with a character entry frame printed in drop color is inputted to the image output device, and an image is recorded on a recording sheet based on the image data output from the image output device. This is what it is. In addition to the above-mentioned means, the digital copying machine has a data input means in the operation section, and the data of the graphic image recognized by the graphic image recognition means can be modified through the data input means. Also good.

[0005]

[Operation] The image cutting means cuts out a predetermined area in the image, such as the area surrounded by the character entry frame of a form printed in drop color. The graphic image recognition means recognizes graphic images such as characters by comparing the graphic image in the area with the data of the graphic image registered in advance. The line image recognition means recognizes a line image connecting the areas and converts it into line image data, for example, arrow data in which an arrow is recognized by detecting a branch or bulge at the edge of the line image. The image synthesis means creates a composite image by combining a pre-registered figure shape corresponding to the figure image recognized by the figure image recognition means and a line image drawn based on the data of the line image recognized by the line image recognition means. do.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings. 1 to 19 are drawings according to an embodiment of the present invention, in which FIG. 1 is a block diagram showing a schematic configuration of an image processing device of a digital copying machine according to an embodiment of the present invention, and FIG. A schematic conceptual diagram showing input/output; FIG. 3 is a diagram showing the format of a form and an example of entry in the form; FIGS. 7 to 9 are diagrams similarly explaining character recognition processing, FIGS. 10 to 13 are diagrams similarly explaining connection recognition processing, FIG. 14 are diagrams similarly showing output images, and FIGS. 15 to 19 are diagrams similarly explaining recognition data correction. It is a figure explaining processing. As shown in the schematic conceptual diagram of FIG. 2, in this embodiment, by setting a document with handwritten figure shape numbers and connection lines on a predetermined form and pressing the copy button, the figure shape numbers and connection lines are automatically set. This is an example of a digital copier that reads lines and outputs them as a clean copy. As shown in FIG. 3, black square-shaped reading reference marks 10 are printed at equal intervals along the top and left sides of the form, and the center is located at the intersection of each of the vertical and horizontal straight lines passing through the center of the reading reference marks 10. Rectangular number entry frames 11 for writing figure shape numbers in dropout colors that are not copied are printed vertically and horizontally. The figure shape number must be written in this number entry frame 11 so as not to touch the frame.

The processing operation of the image processing apparatus of the digital copying machine will be described with reference to the block diagram of FIG. 1 and the explanatory diagrams of processing modes shown in FIGS. 4 to 6. For example, as shown in the entry example in Figure 3, if you place a document with a shape number and connection line written in the specified position on the document table on the document table and press the copy button,
The scanner of the document reading unit 1 scans the surface of the document in the sub-scanning direction and converts optical signals into electrical signals, thereby reading the document image. As mentioned above, fill in the number box 1.
1 is printed in a dropout color that is not copied, so the original image read by the original reading unit 1 is the same as the image shown in FIG. The image signal is converted into digital data by analog/digital conversion and sent to the character cutting circuit 2. The character connection line extraction circuit 2 first recognizes the position of the reading reference mark 10 in the read original image. That is, as shown in FIG. 7, a black pixel histogram of the peripheral portion of the read form image is calculated in the main scanning direction and the sub-scanning direction, and the peak position of the histogram is recognized. The peak position becomes the center position of the reading reference mark 10. Next, as shown in FIG. 8, the positions of the intersections Oij of straight lines passing through the center position of the reading reference mark 10 and extending in the main scanning direction and the sub-scanning direction are determined, and the number entry frame 11 is drawn vertically and horizontally from the intersection Oij. By extracting the image data of the area corresponding to the size of , it is possible to cut out the character image data of the graphic shape number as shown in FIG. The extracted character image data is sent to the character recognition circuit 3. The character recognition circuit 3 calculates the histogram of the extracted character image data again in the main scanning direction and the sub-scanning direction, and recognizes the center position of the histogram. FIG. 9 is a diagram showing character recognition processing of character image data centered on the intersection O12 of the entry example of the form shown in FIG. Based on the calculation result of the histogram in the scanning direction, an area containing only the character image is further cut out as shown in (b). Next, in order to compare the character image with data in a character dictionary registered in memory in advance, the character image is enlarged or reduced to the same size as the data in the character dictionary, as shown in (c). I do. Thereafter, the character image and the data in the character dictionary are compared, and the character with the highest degree of similarity is recognized as the corresponding character, and its character data is output. In addition,
The data contents of the character dictionary include data such as the standard character image itself, the direction of the contour vector of the standard character image, and the number of holes.

Number entry frame 11 performed by connection recognition circuit 4
A connection recognition process that recognizes connection lines connecting . First, as shown in FIG. 6, the image of the connecting line between the number entry frames 11 is divided into rectangular divided areas 12 in the same way as the character image.
A connecting line image is cut out for each divided area 12 as connecting line image data of the divided area 12. The cut out connection line image is divided into block B units smaller than the minimum allowable line width (e.g. 8 pixels x
8 pixels), and raster scanning is performed for each block B unit to count the number of black pixels within the block B unit. When the ratio of the number of black pixels to the total number of pixels in a block B unit is equal to or higher than a set threshold (50% in this example), the block is designated as a black block BB, and when it is less than the set threshold, the block is designated as a white block WB.
Block color determination processing is performed to determine that. FIG. 10 is a diagram showing how a connection line image is traced for each block unit B, FIG. 11 is a diagram showing the block tracing direction in block color determination processing, and FIG. 12 is a diagram showing the block color determination processing described above. FIG. 2 is a diagram showing the results and connection lines extracted based on the results. Block color determination processing will be explained with reference to these figures. In this embodiment, in order to extract connection lines based on the connections between black blocks BB, a process of searching and tracing black blocks BB connected to black blocks BB is performed. First, as shown in FIG. 10, color determination processing is performed for each block B in the main scanning direction of the first line, and when a black block BB is found, the surrounding blocks starting from direction 4 in FIG. A color determination process for B is performed to search for a black block BB. Next, starting from the direction (current tracking direction - 2), the black block BB is searched again in the counterclockwise direction. When the above tracking process is repeated until the edge of the cut out connection line image is reached, the tracking process is terminated, and the process returns to the block where tracking started and performs the color determination process for each block B again in the main scanning direction. The color determination process for block B is sequentially repeated until there are no more blocks that are not subjected to the determination process. When the color determination process for all blocks B is completed in this way, Figure 1
2, a straight line connecting the black blocks BB is extracted and determined to be a connecting line. Note that 1 is included in the cut out connection line image.
If there are only two line segments, the processing time can be shortened by tracking the black block BB from the edge where the black block BB is found and terminating the block color determination process when the other edge is reached. Can be done. In addition, when the connection line is connected to a place other than the number entry frame 11, it is assumed that the connection line extends into the number entry frame 11, and the image connected to the number entry frame 11 is also cut out as an image of the connection line. connection recognition processing. Next, an arrow recognition process for recognizing an arrow when the connecting line image includes the arrow will be described. FIG. 13 is an explanatory diagram of arrow recognition processing. As a result of tracing the black block BB and extracting the connection line in the block color determination process, if the connection line branches or the black block BB swells, use the center of the connection line edge as a reference. Setting an arrow recognition frame 13 corresponding to the maximum allowable length of the arrow,
All the connections of black blocks BB connected to the edge of the connection line fall within the arrow recognition frame 13, and the ratio of the number of black blocks BB within the arrow recognition frame 13 to the total number of blocks B within the arrow recognition frame 13 is a preset threshold. It is determined whether or not it is equal to or greater than the threshold value, and if it is equal to or greater than the threshold value, it is determined that there is an arrow on the edge of the connection line. The arrow information determined in this way is stored together with the extracted connection line information in the memory as connection line information.

The output image creation circuit 5 uses pre-registered graphic shape image data corresponding to the recognition data of the character (number) image recognized by the character recognition circuit 3 and the connection recognition circuit 4.
The image data of the connection line information recognized in is combined and stored in the output image memory as output image data. The output image data read from the output image memory is sent to the copier's printer, and the output image is printed on transfer paper according to well-known electrostatographic processes. FIG. 14 shows an output image according to the entry example of the form shown in FIG. In this embodiment, as shown in the entry example of the form shown in FIG. 3, image data is registered in advance in the memory by associating figure shapes with numbers, such as 1 for a rectangle and 4 for a diamond. In this embodiment, if there is an error in the recognition data of the character image recognized by the character recognition circuit 3, the character image data can be corrected to the correct character image data by operating the input keys on the operation section. This recognition data correction process will be explained below. Figures 15 to 17 show the guidance display on the operation display section and the input format using the numeric keypad when the recognition data correction processing mode is selected, and Figures 18 and 19 respectively show examples of output images that have been misrecognized and corrected correct output. An image is shown. In FIG. 18, the hexagonal graphic shape is output by misrecognizing a character (number) image corresponding to a rectangular graphic shape. In addition, in FIGS. 15 to 17, 14 is a guidance display that guides the operation method, 15 is a numeric keypad,
16 is a numeric keypad input display. First, when the recognition data correction processing mode is selected by pressing the mode selection key on the operation unit (not shown), a guidance message asking for the line number representing the position of the erroneously recognized figure is displayed on the guidance display 14, as shown in FIG. , the operator operates the numeric keypad 15 to input the line number of the erroneously recognized figure. Next, Figure 16
As shown in , a guidance message asking for the column number representing the position of the erroneously recognized figure is displayed on the guidance display 14, so the operator operates the numeric keypad 15 to input the column number of the erroneously recognized figure. Next, as shown in FIG. 17, a guidance message asking for the correct figure number is displayed on the guidance display 14, and the operator similarly operates the numeric keypad 15 to input the correct figure number. In this way, the recognition data of the character image erroneously recognized by the character recognition circuit 3 is transferred to the numeric keypad 15.
will be replaced with the correct shape number entered from . Note that the row number and column number representing the position of the figure are the numbers assigned by counting the number entry frames 11 printed on the form from top to bottom and from left to right, respectively.

00010]

As described above, according to the present invention, the invention as claimed in claim 1 compares an image cut out from a predetermined area in an image with data of a graphic image registered in advance to generate a graphic image. Since it has a figure image recognition means for performing recognition and a line image recognition means for recognizing a line image connecting the areas and converting it into line image data, it is possible to recognize an image composed of a predetermined figure image and a line image. can be done easily and quickly. According to the second aspect of the invention, recognition processing of an image including an arrow in a line image becomes easy. In the invention described in claim 3, a pre-registered figure shape corresponding to the figure image recognized by the figure image recognition means and a line image drawn based on the data of the line image recognized by the line image recognition means are combined. Since the image data of the synthesized image is output, for example, image data of an image composed of a graphic image and a line image that conform to a desired standard can be obtained. The invention as claimed in claim 4 provides a document document in which numbers corresponding to pre-registered figures are written in character entry frames printed by hand in drop color, and line drawings connecting the character entry frames are written. By reading the image, you can obtain a recording sheet on which a clean-cut image is recorded. In the invention as claimed in claim 5, the data of the figure shape recognized by the figure image recognition means can be corrected through the data input means, so even if the figure image recognition means erroneously recognizes the figure image of the document, it can be corrected with a simple operation. You can correct the data of erroneously recognized graphic images.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a schematic configuration of an image processing device for a digital copying machine according to an embodiment of the present invention.

FIG. 2 is a schematic conceptual diagram showing input and output of images.

FIG. 3 is a diagram showing a format of a form and an example of filling in the form.

FIG. 4 is a diagram showing a document image read by a digital copying machine.

FIG. 5 is an explanatory diagram of character extraction processing of the digital copying machine.

FIG. 6 is an explanatory diagram of connection recognition processing of the digital copying machine.

FIG. 7 is an explanatory diagram of a black pixel histogram recognition operation of a peripheral portion of a form image of a digital copying machine.

FIG. 8 is an explanatory diagram of the character image cutting position calculation operation of the digital copying machine.

FIG. 9 is an explanatory diagram of character recognition processing of character image data of the digital copying machine.

FIG. 10 is an explanatory diagram showing how a connection line image of a digital copying machine is traced block by block;

FIG. 11 is an explanatory diagram showing a block tracking direction in block color determination processing of a digital copying machine.

FIG. 12 is an explanatory diagram showing the results of block color determination processing of the digital copying machine and extracted connection lines.

FIG. 13 is an explanatory diagram of arrow recognition processing of the digital copying machine.

FIG. 14 is an explanatory diagram showing an example of an output image of a digital copying machine.

FIG. 15 is an explanatory diagram showing a guidance display asking for the line number of a erroneously recognized figure on the operation display section of the digital copying machine and an input form using a numeric keypad.

FIG. 16 is an explanatory diagram showing a guidance display asking for the column number of a erroneously recognized figure on the operation display section of the digital copying machine and an input form using a numeric keypad.

FIG. 17 is an explanatory diagram showing a guidance display asking for the correct figure number on the operation display section of the digital copying machine and an input form using a numeric keypad.

FIG. 18 is an explanatory diagram showing an example of an output image erroneously recognized by the digital copying machine.

FIG. 19 is an explanatory diagram showing an output image corrected by recognition data correction processing of the digital copying machine.

FIG. 20 is a block diagram showing a schematic configuration of a conventional figure recognition device.

[Explanation of symbols]

1 Original reading unit 2 Character connection line cutting circuit 3 Character recognition circuit 4 Connection recognition circuit 5 Output image creation circuit 6 Output circuit

Claims (5)

[Claims] Claim 1. An image recognition device that includes an image memory, reads image data that has been converted into digital data and is stored in the image memory, performs image processing, and recognizes the shape of the image. an image cutting means for cutting out a region; a graphic image recognition means for performing graphic image recognition by comparing a graphic image in the region with data of a graphic image registered in advance; and a graphic image recognition means for recognizing a line image connecting the regions. An image recognition device characterized by having a line image recognition means for converting into line image data. 2. The image recognition device according to claim 1, wherein the line image recognition means recognizes an arrow by detecting a branch or swell of an edge of the line image. 3. A pre-registered figure shape corresponding to the figure image recognized by the figure image recognition means of the image recognition apparatus according to claim 1 or 2, and data of a line image recognized by the line image recognition means. An image output device includes an image compositing means for creating a composite image by combining line images drawn based on the image compositing means, and outputs image data of the composite image synthesized by the image compositing means. 4. A document reading unit that optically reads a document image and converts it into an electrical signal, and the image data is image data obtained by reading a form document in which a character entry frame is printed in a drop color. A digital copying machine that records an image on a recording sheet based on image data output from the image output device according to claim 3. 5. The image forming apparatus according to claim 4, wherein the operating section has a data input means, and the data of the graphic image recognized by the graphic image recognition means can be modified through the data input means. A digital copy machine.