JPH06251191A - Image processing device - Google Patents

Abstract

PURPOSE:To leave the image of a character that is an object of reading. etc., and to surely eliminate an unrequited image without increasing burden on work. CONSTITUTION:This device is constituted of a knowledge information storage part 18 to store color information to classify a color at every color, uniformity information to specify regularity. and shape information to specify shape for a color image, a color information extraction part 14a which extracts the image at every color based on the information stored in the knowledge information storage part 18, a uniformity extraction part 14b which extracts the image with regularity, a shape information extraction part 14c which extracts the image with prescribed shape, an image classification processing part 14d which classifies a required image part and another image part on the basis of the image extracted by each of the extraction parts 14a-14c, respectively, and an unrequired image elimination part 14e which eliminates a classified another image from the color image set as a processing object.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus suitable for an optical character reader (OCR) or a text reader which optically detects an image and processes the image.

[0002]

2. Description of the Related Art Generally, an optical character reader (OCR)
A text reader detects an image by optically scanning a sheet such as a form, and recognizes a character based on the image.

The text reader has not specified the format of a document or the like to be processed, and is adapted to automatically detect a character image and recognize the character image (convert the image into a character code). There is.

Therefore, in an existing form (sheet) that is commercially available, a ruled line, a character frame indicating a character entry position,
Other entry field names ("amount", "number", "address", etc.)
There is a form removal function for removing these unnecessary images because there are characters that are originally printed and are not recognized. By removing unnecessary images by this form removal function, subsequent processing can be easily performed.

As a method of the form removal function,
An image of a form (blank sheet) in which characters or the like to be processed are not registered is registered in advance, and when the reading process is executed, the image of the form to be processed is compared with the blank sheet image. Unnecessary parts can be deleted by deleting the matching parts.

On the other hand, in an optical character reading apparatus in which the format of the form to be processed is predetermined, information indicating the format of the form to be processed (format control information) is registered, and characters are written according to this information. Read. The format control information has detailed contents such as a ruled line position in a form, a character frame position, and positions of various items.

[0007]

As described above, in the prior art,
A blank sheet image is registered in order to remove unnecessary parts from the image detected by optical scanning, and a method to perform based on this image and format control information for the form to be processed By registering, the method of determining the position where the required information is recorded is used.

However, in the method of registering the image of the blank sheet, it is necessary to compare the image of the blank sheet and the image to be processed in correspondence with each other. There was a case where the unnecessary part could not be surely deleted due to blurring.

Further, in the method of registering format control information, it is necessary to register detailed information for each form to be processed. For example, all ruled line positions, character frame positions, item positions, etc. to be detected have to be registered for each different form, so that the work load is very large and the form to be read is There was a problem that types were limited.

The present invention has been made in view of the above points, and it is possible to leave an image of a character or the like to be read and reliably remove an unnecessary image without increasing the work load. It is an object of the present invention to provide a simple image processing device.

[0011]

SUMMARY OF THE INVENTION The present invention stores color information for classifying each color image, uniformity information for specifying regularity, and shape information for specifying shape. Information storage means for storing the color image to be processed, color information extraction means for extracting an image for each color based on the color information stored in the information storage means, and the color image to be processed Regarding the uniformity information stored in the information storage means, the uniformity extraction means for extracting a regular image based on the uniformity information stored in the information storage means, and the shape information stored in the information storage means for the color image to be processed. On the basis of the shape information extracting means for extracting an image of a predetermined shape, the color information extracting means, the uniformity extracting means, and the shape information extracting means. An image classifying means for classifying a required image part and other image parts based on the image extracted by the above, and a color image to be processed by the other image parts classified by the image classifying means. And an unnecessary image removing unit for removing the unnecessary image.

[0012]

According to this structure, in order to remove unnecessary image portions from the color image to be processed,
It is based on color, uniformity, and shape. In addition, the information referred to in order to extract these characteristics is prepared independently of the processing execution function. Therefore, a highly accurate form removal function for removing an unnecessary image portion is realized, and it is possible to easily deal with a change in an image to be processed.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the arrangement of an image processing apparatus according to an embodiment of the present invention. It is assumed that the image processing apparatus of this embodiment is realized as a character reader such as a text reader.

As shown in FIG. 1, the image processing apparatus includes a color scanner 10, an image storage unit 12, an image processing unit 14, a character recognition unit 16, a knowledge information storage unit 18, an understanding result storage unit 20, and a registration unit 22. It is composed by.

The color scanner 10 detects a multi-valued color image by optically scanning a form or the like to be processed. Image storage unit 12
Is for storing the form image detected by the color scanner 10. Image processing unit 14
From the image stored in the image storage unit 12,
While referring to the information stored in the knowledge information storage unit 18,
This is a process for removing an unnecessary image portion. The image processing unit 14 includes a color information extraction unit 14
a, a uniformity extraction unit 14b, a shape information extraction unit 14c, an image classification processing unit 14d, and an unnecessary image removal unit 14e are provided.

The color information extraction unit 14a classifies each area in the form image by color. Uniformity extraction unit 14
In b, a portion showing uniformity (details will be described later) is extracted from each region in the image extracted by the color information extracting unit 14a. Shape information extraction unit 14c
Is for classifying each area in the form image based on its shape. The format of the form image is understood by the extraction units 14a to 14c.

The image classification processing unit 14d classifies the image portion to be read and the non-reading image portion based on the extraction results of the extraction units 14a to 14c. The non-read target image portion classified by the unnecessary image removing unit 14e and the image classification processing unit 14d is deleted.

The character recognition unit 16 performs character recognition processing on the image portion from which the non-read target image portion has been deleted by the image processing unit 14 with reference to the result of format understanding.

The knowledge information storage section 18 is for storing knowledge information necessary for reading characters from a form image. The knowledge information storage unit 18 includes a color information storage unit 18
a, the uniformity information storage unit 18b, and the shape information storage unit 18c
Is stored as knowledge information for understanding the format.
In addition, a character recognition dictionary 18d that is referred to when the character recognition unit 16 recognizes a character is stored.

The understanding result storage unit 20 includes a color information extraction unit 14
This is for storing the result of format understanding obtained by a, the uniformity extraction unit 14b, and the shape information extraction unit 14c. The information stored in the understanding result storage unit 20 is
It is referred to in the character recognition processing in the character recognition unit 16.

The registration unit 22 controls the man-machine interface and registers the information stored in the knowledge information storage unit 18. The registration unit 22 is provided with an input / output device such as a keyboard and a display device. In addition,
The operation of each unit shown in FIG. 1 is assumed to be controlled by a control device (not shown).

Here, various types of information for format understanding stored in the knowledge information storage unit 18 will be described with reference to FIG. The color information storage unit 18a has registered therein information about colors in the form image. For example, the color of the ruled line or the character frame in the form image is registered as “red” and the character is registered as “black”. The color information extraction unit 14a classifies the image of the designated color for each color according to these pieces of information.

The uniformity information storage section 18b has registered therein information relating to the regularity of the arrangement of bits in the image, that is, information identifying the straight line, the dotted line, the circle, etc. of the arrangement of bits. For example, if the number of bits of the coordinate value x is n or more and consecutive, the registration is made as a vertical line (straight line). The uniformity extracting unit 14b extracts an image portion showing uniformity according to these pieces of information. At this time, the uniformity extraction unit 14b
Considers the inclination and blurring of the form image to be processed, and extracts the image with a certain width.

The shape information storage section 18c stores information about the shape of an image. For example, if a rectangle is formed by four straight lines, it is a table, and the vertical line existing in the rectangular frame is registered as a position where columns of the table are divided. The shape information extracting unit 14c classifies the image according to the shape according to these pieces of information.

As a result, the understanding result storage unit 20 stores information for distinguishing the image portion to be read from the image portion not to be read, and the information on the images classified by the extraction units 14a to 14c. It

Next, the operation of this embodiment will be described.
Here, it is assumed that the form image as shown in FIG. 3 is detected by the color scanner 10 and stored in the image storage unit 12. In order to simplify the explanation, it is assumed that the line (frame) portion of the image shown in FIG. 3 is “red” and the character portion is “black”, and that the image has no unnecessary inclination. In addition, it is assumed that the knowledge information storage unit 18 is registered with information having the content shown in FIG.

First, the color information extraction unit 14a of the image processing unit 14 classifies the image for each color with reference to the color information registered in the color information storage unit 18a. Here, the image is classified into a frame image and a character image.

Next, the uniformity extracting unit 14b extracts the image portion showing the uniformity while referring to the uniformity information registered in the uniformity information storage unit 18b. Here, from the image of the frame portion, a bit string continuous with n bits or more in the x direction is extracted as a horizontal line segment, and a bit string continuous with m bits or more in the y direction is extracted as a vertical line segment.

Next, the shape information extraction unit 14c classifies the images based on the shape of the image with reference to the shape information registered in the shape information storage unit 18c. Here, the uniformity extraction unit 14
The connection state of the line segment extracted by b is determined, and the line segment (vertical line segment, horizontal line segment) that divides the rectangular frame or the area within the rectangular frame is classified.

These results are stored in the understanding result storage section 20. The image classification processing unit 14d classifies the image portion to be read and the non-reading image portion based on the contents stored in the understanding result storage unit 20. For example, the image classification processing unit 14d regards the inside of the rectangular frame as the image portion to be read, and the other as the non-read target image portion.

The unnecessary image removing unit 14e removes the portion classified by the image classification processing unit 14d as the non-read target image, that is, all the images including the characters outside the rectangular frame and all the line segments within the rectangular frame. To do. As a result, only the image of the character to be recognized remains.

The character recognition unit 16 includes an unnecessary image removing unit 1
The character recognition processing is performed on the image to be subjected to the recognition processing obtained by 4e with reference to the character recognition dictionary 18d. Here, the character recognition unit 16 uses the understanding result storage unit 20.
Character recognition is performed for each unit of the columns separated by the line segment based on the positional relationship of the line segments indicating the column divisions stored in, and the recognition result is output.

In this way, the format image of the form image is understood based on the information stored in the knowledge information storage unit 18, and the character recognition process is performed using the result. That is, for the form image, color, uniformity,
Since the format is understood on the basis of the shape and the shape, it is possible to reliably classify the image to be processed and the unnecessary image, and it is possible to realize an accurate and high quality foam removal function.

Further, since it is not strict information like the conventional format control information, it is easy to register the information and has the flexibility to cope with the positional deviation and the inclination of the form image, so that it can be processed. The versatility is improved because the type of form (sheet) to be processed is not specified.

Furthermore, since information about color, uniformity, and shape can be registered independently in the knowledge information storage unit 18, it is possible to easily deal with changes in the form to be processed. it can.

In the above embodiment, the example realized as a character reading device has been described, but it can be applied to an image processing device having a function (form removal function) of removing an unnecessary portion from a color image. it can. Further, the knowledge information storage unit 1 in the above embodiment
The information stored in 8 is an example for simplifying the description, and the content can be arbitrarily registered.

[0037]

As described above, according to the present invention, unnecessary image portions are removed from a color image to be processed on the basis of color, uniformity, and shape. The information referenced to extract
Since it is prepared independently of the processing execution function, you can flexibly deal with changes in the image to be processed without registering detailed information for each processing target, and the image of characters etc. to be read with high accuracy Therefore, it is possible to remove unnecessary images without fail. It is possible.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an image processing apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining various types of information for format understanding stored in a knowledge information storage unit 18 in the present embodiment.

FIG. 3 is a diagram showing an example of a form image for explaining the operation in the present embodiment.

[Explanation of symbols]

10 ... Color scanner, 12 ... Image storage unit, 14
Image processing unit 14a Color information extraction unit 14b Uniformity extraction unit 14c Shape information extraction unit 14d Image classification processing unit 14e Unnecessary image removal unit 16 Character recognition unit 18 Knowledge information Storage unit, 18a ... Color information storage unit, 18b ... Uniformity information storage unit, 18c ... Shape information storage unit, 18d ... Character recognition dictionary, 20 ... Understanding result storage unit, 2
2 ... Registration department.

Claims (1)

[Claims] 1. Color information for classifying color images into color images, uniformity information for specifying regularity,
And information storing means for storing shape information for specifying the shape, and a color for extracting an image for each color based on the color information stored in the information storing means for the color image to be processed. Information extracting means, a uniformity extracting means for extracting a regular image based on the uniformity information stored in the information storing means for the color image to be processed, and the color image to be processed With respect to each of the shape information extracting means for extracting an image of a predetermined shape based on the shape information stored in the information storing means, the color information extracting means, the uniformity extracting means, and the shape information extracting means. Image classification means for classifying required image parts and other image parts based on the images extracted by The image processing apparatus characterized by other image portions which are classified by kind means, equipped with a, and unnecessary image removing means for removing from a color image to be the processing target.