JP3566738B2 - Shaded area processing method and shaded area processing apparatus - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to a method for processing a shaded area in a case where a shaded area is included in an input document image, and more particularly to a method for inputting a document original as image data and performing character recognition on an optical character reader (OCR). The present invention relates to a shaded area processing method suitable as a processing method.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, in a character recognition device, in a character area modified by shading, it has been difficult to accurately cut out a character because a hatched dot and a character pattern overlap. Therefore, when processing a document including a shaded area in the document original, the input image is displayed on the display device before the character recognition processing, and the document area is displayed so that the operator avoids the shaded area. It was necessary to specify the name and correct erroneous recognition caused by shading.
[0003]
As a technique for extracting characters included in an area modified by a background pattern, there is a technique for extracting a found character using a self-organizing multi-resolution filter, which was announced by Shuka et al. At the IEICE Spring National Convention (1989). Are known.
[0004]
In this method, a learning area is provided in each texture area by a self-organizing method, a new intermediate variable is created from a combination of two variables from a large number of input variables, and a number of new intermediate variables are selected based on a predetermined evaluation criterion. Self-select the intermediate variables. Judgment is made as to whether any of the selected intermediate variables has sufficiently converged to the target value as an output, and if there is no sufficiently converged intermediate variable, the self-selected one is regarded as the input variable. Then, the operation returns to the operation for generating a combination. Until a variable that satisfies the convergence condition is obtained, the above operation is repeated, and a character is extracted from the found area in the document by obtaining a highly accurate conversion function.
[0005]
However, also in the above method, it is necessary for the operator to specify a background pattern area including characters in advance as preprocessing. Further, in the process of obtaining the conversion function by the self-organizing method, it is necessary to repeat the floating-point operation, so that it takes a long time for the operation process and is unsuitable as a preprocessing method for character recognition.
[0006]
Due to the spread of word processors and DTP in recent years, the number of text-modified document manuscripts has increased, and technology for automatically and quickly detecting and removing character modification by shading as preprocessing such as character recognition processing. Is needed.
[0007]
[Problems to be solved by the invention]
In view of such circumstances, the present invention automatically and quickly detects a shaded area included in an input document document and removes the shaded information while preserving character information. It is an object of the present invention to provide a suitable shaded area processing method.
[0008]
[Means for Solving the Problems]
In order to solve the above-mentioned problem, according to the first aspect of the present invention, small dots forming a shaded area are extracted from input document image data, and the small dots are integrated to form a halftone dot in the document image data. A shading candidate area is extracted, a shading area is determined based on the size of the shading candidate area, and shading information is removed from the area determined to be the shading area .
[0009]
According to a second aspect of the present invention, in addition to the first aspect of the present invention, shading information is removed for each predetermined area of the document image data . According to the third aspect of the present invention, in addition to the first aspect of the invention, the position of the shaded area is output . Further, according to a fourth aspect of the present invention, in addition to the first aspect, small dots forming a shaded area are extracted from input document image data, and the small dots are integrated to form the document image data. Extracting means for extracting a shaded candidate area therein; determining means for determining a shaded area based on the size of the shaded candidate area; and removing shaded information from the area determined to be the shaded area. and removal means, and configured to so that with the.
[0010]
【Example】
(Example 1 Basic processing)
Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a flowchart of a basic process for detecting a shaded area. The process of detecting a shaded area is composed of two steps, a rectangle extraction step and a rectangle integration step.
[0011]
The rectangle extracting step is a step of extracting a black run from the continuity of black pixels in the input document image and extracting a circumscribed rectangle of the black run. In general, the dots that make up the shaded area are extracted as circumscribed rectangles that are much smaller than the circumscribed rectangle of the character. In the present invention, by setting a threshold (for example, a size within 3 × 3 pixels) corresponding to the character size used in the document to be processed, only the information of the halftone dots in the document image is changed to the small dots ( (Rectangle). As the threshold value to be set, a value is set that does not erroneously detect a comma, a reading point, or the like in a document to be processed as a halftone dot.
[0012]
Next, in the step of integrating rectangles, small dots (for example, a distance of 10 pixels or less) within a predetermined threshold (for example, a distance of 10 pixels or less) are subjected to horizontal and vertical threshold processing for the distance between the extracted small dots. The process of integrating the rectangles is advanced, and the rectangle is extracted as a rectangle of a shaded candidate area.
[0013]
By checking the size of the shaded candidate area from the coordinate information of the shaded candidate area generated in the step of integrating the rectangle, it is determined whether or not the area is a shaded area. In general, character modification by shading is often performed on a plurality of characters rather than on only one character. Therefore, the determination of the shaded area in the present invention is performed by, for example, the height of a line of a document. , A shaded candidate area having a length equal to or more than twice the row height is determined as a shaded area.
[0014]
The region determined to be a shaded region is subjected to a process of removing shaded information by inverting black pixels of small dots belonging to the shaded region into white pixels. Information on which small dot belongs to which shaded area depends on the subordinate of the small area to be merged and the rectangular area to be merged in the process of integrating the small dots and extracting the shaded candidate area. It can be easily grasped by managing the relationship.
[0015]
FIG. 2 is a block diagram in which the present invention is applied to a character recognition device. The document image document is input from the scanner 1 and stored in the image memory 2 as document image data. The central processing unit 3 reads the shaded area processing program from the shaded processing program storage ROM 4, executes the process of extracting the shaded area for the document image data in the work area RAM 5, and removes the shaded area from the document image. . At this time, the coordinates of the rectangle extracted as the shaded area are also stored in the RAM 5.
[0016]
The shaded document image data is subjected to character recognition processing such as pattern matching with a character recognition dictionary (not shown) by the central processing unit 3 to convert the image data into character code data. 6 is displayed.
[0017]
(Example 2 division processing)
Here, when the processing of the first embodiment is performed on the entire original image as a processing target, a considerable work area RAM such as rectangular data is required. In a system having a sufficient work area RAM, processing can be performed by the method of the first embodiment. However, in order to operate on a system such as a personal computer in which the mounting memory is limited, memory saving must be performed. Have to get started.
[0018]
In the second embodiment, the entire original image is processed by dividing the range for the extraction processing of the shaded area and performing the extraction processing of the shaded area for each divided area. FIG. 3 shows a flowchart of a basic process of detecting a shaded region in the division process.
[0019]
In the second embodiment in which the processing unit is divided, the shading area may exist over the boundary of the divided area, and therefore, processing at the boundary of the unit area is a feature. In the division processing according to the second embodiment, a shading area determination process is performed based on a positional relationship between a shaded candidate area and an overlap area, that is, an area having an appropriate height existing at the boundary of the unit area currently being processed. Is performed.
[0020]
A description will be given by taking as an example the shaded candidate areas (1), (2), and (3) shown in FIG. The area (1) does not cover the overlap area, that is, the area having an appropriate height existing at the boundary of the unit area currently being processed. In the same manner as the basic processing, the shaded area is determined, and if the size is appropriate, the shaded area is removed as the shaded area.
[0021]
The area (2) is an example of a state where a part of the shaded candidate area covers the overlap area, and the area (3) is an example of a state where the shaded candidate area is included in the overlap area. . When the shaded candidate area is in the state of (2) or (3), there is a high possibility that the shaded area exists across the next unit area, and thus the boundary processing shown in the flowchart in FIG. 5 is executed. .
[0022]
First, a shaded area is determined for a shaded candidate area in the same manner as in the basic processing, and if the area is of a predetermined size, the removal processing is performed as a shaded area. Here, when the shaded candidate area is determined to be a shaded area, the Y coordinate (Yos) of the start point of the overlap area and the Y coordinate (Yae) of the end point of the shaded area are compared, and Yae is equal to or greater than Yos. In the case of, since the shaded area is likely to exist over the next processing area, this shaded area is carried over to the next processing area and used in the next processing area.
[0023]
By this processing, even when the shaded area extends over the next unit area as in (2) of FIG. 4, the shaded area is carried over to the next unit area, and is integrated and grown. Therefore, the hatched area at the boundary can be extracted without any problem.
[0024]
If the shaded candidate area does not have an appropriate size and is not determined to be a shaded area, the Y coordinate (Yos) of the start point of the overlap area and the Y coordinate of the start point of the shaded candidate area are determined. The coordinates (Yae) are compared. If Yae is equal to or less than Yos, it is considered that a part of the shaded area is in the unit area currently being processed. Is a candidate for the starting point of the next processing area.
[0025]
By this processing, the unit area of the next processing is returned to the start point coordinates Yas of the shaded candidate area not determined as the shaded area, and the processing areas are partially overlapped. By performing the overlap processing, even when a shaded area is included in the overlapped area as shown in (3) of FIG. 4, it is possible to respond to the shaded area determination in the next unit area. .
[0026]
Further, in the present invention, the position of the rectangle (the coordinate value of the start point / end point, or the coordinate value and the width / height of the start point) detected as the shaded area can be output to an application such as a word processor or DTP software. . Thus, after performing the character recognition processing by removing the hatching, the application reproduces the character modification by the shading, applies the hatching to the recognition result, displays the result on the display, or outputs the result to the printing device. It becomes possible.
[0027]
Further, in the present embodiment, the description of the determination processing of the shaded area is advanced by paying attention to the size of the small dot after the rectangle extraction step, but in addition to the size of the small dot, the input image data before the rectangle extraction step is added. It is also effective to use the shape information of the shaded dots in the middle.
[0028]
【The invention's effect】
As described above, according to the method of processing a shaded area according to the present invention, a shaded area existing on an image can be automatically detected and removed, and each shaded area can be designated by an operator. The operability of the system is improved and the overall processing time is reduced. In addition, the document to be subjected to the character recognition processing is expanded, and improvement in input efficiency can be expected.
[Brief description of the drawings]
FIG. 1 is a flowchart of a basic process for detecting a shaded area according to the present invention.
FIG. 2 is a block diagram in which the present invention is applied to a character recognition device.
FIG. 3 is a flowchart of a basic process of detecting a shaded region in the division process.
FIG. 4 is a diagram illustrating a shaded candidate area existing at a boundary portion.
FIG. 5 is a flowchart illustrating boundary processing according to the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Scanner 2 Image memory 3 Central processing unit 4 Shading area processing program ROM
5 Work area RAM
6 Display

Claims (4)

Extract the small dots that make up the shaded area from the input document image data,
By integrating the small dots, a shaded candidate area in the document image data is extracted,
The shaded area is determined based on the size of the shaded candidate area,
Removing the shaded information from the area determined to be the shaded area ;
A method for processing a shaded area. 2. The method according to claim 1, wherein the shading information is removed for each predetermined area of the document image data . 2. The method according to claim 1, wherein the position of the shaded area is output . Extracting means for extracting small dots forming a shaded area from the input document image data, and integrating the small dots to extract a shaded candidate area in the document image data;
Determining means for determining a shaded area based on the size of the shaded candidate area;
Removing means for removing shaded information from the area determined to be the shaded area;
A shaded area processing device comprising:

Images