JPH03164873A - Document information processing system - Google Patents

Abstract

PURPOSE:To easily confirm a place, where a character is erroneously recognized as a ruled line or the ruled line is erroneously recognized as the character, by distinguishing the colors of the character and the ruled line as well when the result of a document processing is displayed. CONSTITUTION:An input picture is read by a picture input device 3 and stored in an input picture storing area 7 and the data are dispatched to a character recognition part 62 and a ruled line recognition part 63. Then, the respective recognized results are stored in a character recognized result storing area 71 and a ruled line recognized result storing area 72. A character font generation part 64 and a ruled line pattern generation part 65 generate a character font and a ruled line pattern corresponding to stored code information. A character display part 66 and a ruled line display part 67 stores the display picture in a display picture storing area 74 according to the generated character font, ruled line pattern and stored color designation information. A display device 4 displays a picture and when it is necessary to correct the picture, the data are corrected by operating a recognized result correction part 68. Thus, the place, where the character is erroneously recognized as the ruled line and the ruled line is erroneously recognized as the character, is easily discovered.

Description

[Detailed description of the invention] [Industrial application field]

The present invention provides workstations, office processors,
This invention relates to the two-document information processing capability of a document image processing device using a personal computer, word processor, or the like.

[Conventional technology]

Users can create general documents such as delivery notes, invoices, etc. using office equipment such as those mentioned above. Various document processing systems capable of printing have been widely used in recent years. In such a system, it is first necessary to define the format, and this method requires a lot of effort and time to manually enter characters, borders, etc. using a cursor. If it exists as paper, it can be input as an image and each character and ruled line can be automatically recognized.Furthermore, in the case of an automatic form recognition system, for example, it can also recognize the logical structure between entries. Document processing devices that automate the work required for conventional document creation are being considered. It is extremely rare that recognition errors do not occur in such automatic recognition of various documents, and in most cases, the user must correct the recognition results afterwards. In correction work, it is customary to display the recognition results on a display or the like and visually check the results with the user.When displaying the recognition results on a display or the like, conventionally the inventors have filed a patent application. was displayed by a single color.

[Problem to be solved by the invention 1] When the recognition results are displayed in the conventional method (single color), when the user checks the recognition results, he or she discovers places where characters are mistakenly recognized as ruled lines or where ruled lines are mistakenly recognized as characters. In order to do this, a great deal of attentiveness was required, as the object was mistakenly recognized as having a very similar shape. For example, it is possible that part of a dotted line is formed by a period as a result of misrecognition, but it is extremely difficult to find it in a complex format. In other cases as well, the higher the recognition rate of the entire input document, the more difficult it becomes to find erroneously recognized parts. An object of the present invention is to solve this problem and facilitate the user's work of correcting recognition results. [Means for Solving the Problems] In order to achieve the purpose of item IE, an image input means and a display means for converting a document including ruled lines and characters into image data, and an image input means and a display means for converting a document including ruled lines and characters into image data, and image recognition means for recognizing characters and ruled lines from the converted image data, converting each into code data, and outputting a character font and a ruled line pattern generated based on the converted code data to the display means. In the document processing system, the present invention provides means for outputting the character font and ruled line pattern generated by the image recognition means to a display device in different colors. 1 Effect: When outputting to a display device, by distinguishing the color of the character font and the ruled line pattern, it is possible to easily find a place where a character is mistakenly recognized as a ruled line or a ruled line is mistakenly recognized as a character.

[Example 1] Here, a document processing system 41I, which is a type of document processing system, will be used.
+ A recognition system is given as an example. FIG. 1 is a basic conceptual diagram of the present invention. 4-1゜4-2
calculates the recognition results for characters and ruled lines for the input image.
It shows how it is displayed in 12 colors and 2 colors. 41 is in the character recognition section! This is an enlarged view of a part of the # line that was incorrectly recognized as the alphanumeric character 1. 42, in the ruled line recognition unit, some of the ruled lines of the input image are rubbed, and the LU
This is an enlarged version of the parts that were not recognized. 4 is a recognition result display screen displayed by the present invention, and 4-1.
4-2 was synthesized. As can be seen from 40, which is an enlarged view of the erroneously recognized portion of 4, according to the present invention, the portions where characters are erroneously recognized as ruled lines, or where ruled lines are erroneously recognized as characters, are clearly expressed by different colors. FIG. 2 is a hardware configuration diagram in the above embodiment of the present invention. In the figure, 1 is a keyboard for inputting commands from the operator, 2 is a processor for executing form recognition processing, 3 is an image input device for importing image data into memory, and 4 is for displaying form processing results. display device. 5 is a controller for the display device 4; 6 is a memory for storing programs executed by the processor 2; 7 is a memory for storing data used by the processor 2; 8;
is a bus for interconnecting devices. The memory 6 includes an overall control section 60 that stores a main program for overall control, an image input section 61 that stores a subroutine program for inputting images, and a character input section 61 that stores a subroutine program for character recognition. Recognition unit 6
2, a ruled line recognition unit 63 that stores a subroutine program for recognizing ruled lines, and a character font generation unit 64 that stores a subroutine program for generating a character font based on the character code generated by 62;
63, a character display section 66 that stores a subroutine program for displaying the result of character F2, and a ruled line recognition section It is comprised of a ruled line display section 67 that stores a subroutine program for displaying results, and a recognition result modification section 68 that stores a subroutine program for modifying recognition results. The memory 7 includes an input image storage area 70 for storing image data and a character F2 result storage area F! for storing the result of character recognition. f7.71, a ruled line recognition result storage area 72 for storing the ruled line tIa recognition results, a work area 73 for the processor 2, and a display image storage area 74 for storing one display image. Figure 3 is a system configuration diagram of the form recognition result display device.
It shows a series of data flows. An input image is read by the image input device 3 and stored in the input image storage area 70. The data stored in the character recognition unit 62
, the ruled line recognition unit 63, and the respective recognition results are stored in the character recognition result storage area 71. The result is stored in the ruled line recognition result storage area 72. The character font generation unit 64 and ruled line pattern generation unit 65 generate character fonts and ruled line patterns corresponding to the code information stored in 71.72. The character display section 66 and the ruled line display section 67 are based on the character fonts and ruled line patterns generated in 64 and 65, and the color designation information stored in 71 and 72. A display image is stored in the display image storage area 74. The display device 4 displays the image 74, and if correction is necessary, activates the recognition result correction unit 68 to correct the data 71 and 72. This makes it possible to easily discover where characters are mistakenly recognized as ruled lines or where ruled lines are mistakenly recognized as characters. The operation of this device will be explained according to FIG. Figure 4 shows
5 is a flowchart of a form recognition result display program stored in memory 6. FIG. Step 100-700,900
correspond to the subroutine programs stored in areas 61-68, respectively. In the image input step 100,
The image input device 3 reads the paper on which the format of the form is written and stores it in the input image storage area 70 as digital image information. In the character recognition step 200, characters are extracted and recognized, and the left-hi, lower-right coordinates of the 11-character area, 2-character period code, font, and display color information are stored as character code data in the character recognition result storage area 71. do. In the M-line recognition step 300, a ruled line is extracted and recognized, and the start point, end point coordinates, line width, line type, and color information at the time of indication are stored in the ruled line recognition result storage area 72 as ruled line code data. A character font generation step 400 generates a character font corresponding to the code data stored in 71, and a ruled line pattern generation step 500 generates a ruled line pattern corresponding to the code data stored in 72. In the character display step 600, the character font generated in step 400 is stored in the display dot area 74 in color 1. In the ruled line display step 700. The ruled line pattern generated in step 500 is stored in the display double point area 74 in one color and two. However, color 1 and color 2 are different. After displaying the recognition results as shown in L. When an operator inputs an input 1 for correction processing, such as a delete key or an execution key, the correction processing program 900 is activated. If other keys are manually operated, the process moves to routine 1000 for executing processing according to the key operation. The correction process will be explained with reference to FIGS. 5(A) to 5(C). In Figure (A), first it is determined whether a single character or a ruled line has been selected as the object to be corrected. If a character is selected, the process moves to a character correction step 940, and if a ruled line is selected, the process moves to a ruled line correction step 960. (
In B), in the character correction step 940, it is determined whether addition or deletion has been selected as the correction function. If append is selected, font specification 9
44, position and size specification 946, additional writing 948 is performed, and if erasure is selected, area specification 950 and erasure 952 are performed. In Figure (C), the ruled line correction step 96
At 0, it is determined whether addition or deletion has been selected as the function selection. If append is selected, the line type specification 964, start point and end point specification 966 are performed, and then the append 968 is selected.
and if deletion is selected, start point and end point specification 9
70 and erase 972. With the above steps, a series of processes for automatic recognition of an input image is completed. Note that the character recognition step 200. Ruled line recognition step 30
In addition to distinguishing color 1, which is the color of one character display, and color 2, which is the color of ruled line display, it is also conceivable to change the density of the color in accordance with the certainty of recognition. That is, the single character recognition unit 62. The ruled line recognition unit 63 uses the character recognition result storage area 71 . It is stored in the ruled line recognition result storage area 72. As a result, by focusing on the density of one displayed color, candidates for characters and ruled lines that are causing erroneous recognition can be easily found. FIG. 6 shows basic flowcharts 1 to 1 for character density determination, in which a color in which the likelihood of character recognition is greater than Rl-1 and less than or equal to R4 can be determined as color 1 with density n+1-i (however, R0 =OO≦i≦n. 0<R, <R2<...<R-t<R- is a value representing the certainty of recognition, and the larger the value, the more likely it is). Similarly, FIG. 7 is a basic flowchart for determining the density of ruled lines, in which a line whose reliability of line recognition is thicker than R4-8 and less than or equal to RL can be determined as color 2 with density n+1-i. (However, R, = OO≦i≦n, O<R, <R, <-...<R,-[<R is a value that represents the certainty of recognition. The larger the value, the more likely it is. ). In addition, one character recognition step 200, ruled line recognition step 3Q
In O, in addition to distinguishing between the character display color, color 1, and the ruled line display color, color 2, the density of the color is changed according to the certainty of recognition as described above, and the color density is also changed on the screen using a mouse etc. It is also conceivable that by specifying an arbitrary area, input image information for that part can be viewed. Regarding the method of displaying input image information, there are other ways to display input image information, such as replacing that part with an input image whose color is different from the others, or changing the color of the recognition result and the other for those with a threshold of 4ff or less. One possibility is to display changed input images alternately. As a result, it is possible to determine whether 2% L'l m or not.
When correcting an erroneously recognized part, it can be compared with the input image on both sides, so correction work can be done accurately and quickly. Effects of the Invention As described above, according to the present invention, since the colors of characters and ruled lines are distinguished when displaying document processing results, nine characters can be changed to ruled lines in the correction work after automatic recognition of one library. , or where a ruled line is misrecognized as a character. Furthermore, if the color 91 degrees is changed in accordance with the degree of certainty of recognition, candidates for misrecognition locations can be easily identified by focusing on the density of one color.

[Brief explanation of the drawing]

Fig. 1 is a basic conceptual diagram of the present invention, Fig. 2 is a hardware diagram of the automatic form recognition system, Fig. 3 is a system configuration diagram of the automatic form recognition system, and Fig. 4 is a basic flowchart of the automatic form recognition system. , Figures 5(A) to (C) are the basic flowcharts of the recognition result correction program, Figure 6 is the basic flowchart of the character color density determination program, and Figure 7 is the basic flowchart of the character color density determination program.
The figure shows the basic flowchart 1 of the color density determination program for ruled lines.
- indicates. Description of symbols 1... Keyboard, 2... Processor, 3... Image input device, 4... Display device, 5... Controller for display device, 6... Memory for program storage, 7...・
- Data storage memory, 8... bus. Yogu-ku Haya 2 drawing Broom 3 ￥ S drawing CA) (C) CB) Fig. C Fig.

Claims (1)

[Claims] 1. An image input means for converting a document including characters and ruled lines into image data, a display means, and a system for recognizing characters and ruled lines from the image data input from the image input means. , respectively into code data, and outputs a character font and a ruled line pattern generated based on the converted code data to the display means, the image recognition means includes the A document characterized in that a predetermined first color is given to a character font, a second color different from the above color is given to a ruled line pattern, and the recognition result is output to the display means. Information processing method. 2. In the document information processing method described in Section 1, when outputting character fonts and ruled line patterns to the display means, it not only distinguishes between colors, but also changes the intensity of each color depending on the certainty of recognition. A document information processing method characterized by: 3. In the document information processing method described in Section 1, when outputting character fonts and ruled line patterns to the display means, it not only distinguishes between colors, but also distinguishes between colors for which the certainty of recognition is below a certain threshold. is a document information processing method characterized by replacing that part with an input image whose color is different from that of the rest. 4. In the document information processing method described in Section 1, when outputting character fonts and ruled line patterns to the display means, it not only distinguishes between colors, but also distinguishes between colors for which the certainty of recognition is below a certain threshold. is a document information processing method characterized by alternately outputting an input image whose color is different from that of the other portions, and a character font and a ruled line pattern generated by an image recognition means. 5. In the document information processing method described in Section 1, when outputting character fonts and ruled line patterns to the display means, it not only distinguishes between colors, but also changes the intensity of each color depending on the certainty of recognition. Furthermore, a document information processing method is characterized in that, on a screen that outputs character fonts and ruled line patterns, an arbitrary area specified by a mouse or the like is replaced with an input image whose color is different from that of the others.