JPH05128306A - Attribute discriminating method for sentence picture - Google Patents

Abstract

PURPOSE:To classify simply binarized constituting elements into 5 classifications (character (string), graphic, photograph, table, and field separator) regardless of the complication of layout. CONSTITUTION:When a binarized document picture is inputted, the document picture is reduced, and an inclination correction processing S19 is used also for constituting elements in this picture to obtain the number of intra-area black picture element, the aspect ratio and the thickness of the circumscribing rectangle, and rectangularity and the thickness of the area in steps S10, S12, nd S16 respectively, and they are compared with prescribed thresholds to classify constituting elements into 5 classifications, namely, character (string), graphic, photograph, table, and field separator. The contact area between a character (string) and a field separator at the time of reduction of the document picture is subjected to area division and is discriminated after restoration to the original picture by a contact area separating and attribute discrimination processing S18.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character area in a document image in a communication device such as a facsimile, a document image database input device, an optical character reader (OCR), and the like.
The present invention relates to a document image attribute discriminating method for discriminating the attributes of constituent elements such as a graphic area, a photograph area, a table area, and a field separator area.

[0002]

2. Description of the Related Art Conventionally, as this kind of attribute discrimination method,
For example, there is one described in Japanese Patent Laid-Open No. 03-126181.

In this document image attribute discrimination method, an image sensor scans an input document image to read multivalued image data. The read multi-valued image data is converted into the binary image data by the optimum binarization process without impairing the characteristics and properties of the multi-valued image data. That is, in this optimal binarization process, multi-valued image data is input and characters
The field separator area is subjected to area separation, and the chart / photo area is subjected to dither processing (halftone dot processing) to synthesize the images of both.

The image binarized by this optimal binarization processing is divided into a plurality of blocks by area division, and then attribute discrimination processing is performed. In the attribute discrimination processing, the aspect ratio, the area, and the black / white inversion density (the number of black / white inversions in the block / area) are calculated as the feature amount for each block obtained by the area division, and the area is determined based on this. Classify into field separators, pictures, graphics, and other areas of text.

[0005]

However, the conventional attribute discrimination method has the following problems.

(A) In the conventional attribute discriminating method, since the optimum binarized image is the object of attribute discriminating processing, it is necessary to perform complicated optimal binarizing processing as preprocessing. Moreover, since the figure and the table are not discriminated, there is a problem that the figure and the table in the document image cannot be separated.

(B) In order to solve the problem of (a), the applicant of the present application previously proposed an attribute discrimination method in Japanese Patent Application No. 03-082704. In this proposal,
The document image is simply binarized to create a binarized image (original image), the entire binarized image is further reduced, the reduced image is divided into regions, and the divided region is divided into regions. hand,
The number of black pixels in the reduced image area, the aspect ratio of a rectangle circumscribing the area,
By the process of obtaining the thickness of the region and the rectangularity of the region and comparing them with the thresholds for them, and the inclination correction process,
Obtaining the ratio of the length of the longest black run in the reduced image area to the size of the circumscribed rectangle of the reduced image area, and comparing it with a corresponding threshold value, and obtaining the black and white inversion density in the reduced image area. , A process of comparing with a threshold value corresponding to the reduced image area for a character (column), a figure,
The areas of the photograph, table, and field separator are identified.

However, in this attribute determination method, since the reduced image is divided into areas and the attribute is determined, when the document image is reduced, characters and lines represented by black circles are used as in the example shown in FIG. When the displayed field separators come into contact with each other, the regions cannot be correctly divided, and the attribute determination is erroneous.

The present invention solves the problem that the above-mentioned conventional art has with respect to the difficulty of accurately determining the attribute when a character (column) and a field separator come into contact with each other when the document image is reduced. It provides a method of discriminating the attributes of.

[0010]

In order to solve the above-mentioned problems, the present invention simply binarizes a document image to create a binarized image (original image), and further reduces the entire binarized image. After that, the reduced image is divided into components, and the number of black pixels in the reduced image region, the aspect ratio of the rectangle circumscribing the region, the thickness of the region, and the rectangularity of the region are obtained and compared with the thresholds for them. Processing, determining whether or not there is a black run longer than a predetermined threshold value in the reduced image area, obtaining an area of the original image corresponding to the reduced image area, and determining black pixels in the area of the original image. A region in which the original image is divided by obtaining a line in which no pixel exists and a line in which at least one black pixel exists, and a process for obtaining the reduced image region corresponding to the divided original image region; Longest black run in the area The length and the thickness of the area are obtained, and by the processing of comparing with the threshold value corresponding thereto and the inclination correction processing, the length of the longest black run in the reduced image area with respect to the size of the circumscribed rectangle of the reduced image area is calculated. A process of obtaining a ratio and comparing it with a threshold value corresponding thereto and a process of obtaining a black-white inversion density in the reduced image region and comparing it with a threshold value corresponding thereto are performed to perform the reduction image region with a character (string) and a figure. , Photo, table, and field separator areas are discriminated.

[0011]

According to the present invention, since the method of discriminating the attribute of a document image is configured as described above, by the process of determining whether or not there is a black run longer than a predetermined threshold value in the reduced image area,
When the image is reduced, it is determined whether or not the character (row) and the field separator are in contact with each other to form the same area.
When it is determined in the processing that the character (string) and the field separator are in contact with each other and are in the same area, an area of the original image corresponding to the area is obtained, and a line in which no black pixel exists in the area of the original image and The area of the original image is divided by obtaining the line in which at least one black pixel exists, and the area of the reduced image corresponding to the divided area of the original image is obtained. After that, the length of the longest black run and the thickness of the area in the reduced image area are obtained, and a process of comparing with the corresponding threshold value is executed to determine the character (row) of the contact area and the field separator area. be able to. Therefore, the above problem can be solved.

[0012]

FIG. 5 shows an embodiment of the present invention and is a diagram showing the overall processing contents of image processing. First, as preprocessing, in simple binarization processing 1, an input document image is scanned by an image sensor or the like and binarized by a simple threshold value to generate a binarized image. The binarized image is stored in the original image memory 2 and is sent to the image reduction processing 3. In the image reduction processing 3, the original image is reduced to generate a reduced image, and the area division processing 5 is performed, and the image is stored in the reduced image memory 4.

In the area division processing 5, a reduced image is input,
The divided area 6 is generated by dividing it into constituent elements such as characters (rows), figures, tables, photographs, and field separators.
The data of the divided area 6 divided into these constituent elements is
It is sent to the attribute discrimination processing 7, which is a feature of this embodiment.

In the area division processing 5 of the pre-processing, the division method described in the above-mentioned JP-A-03-126181 by the applicant of the present application can be used. The division method repeats scanning in the horizontal and vertical directions twice, divides an image into rows and columns in which black pixels do not exist, and integrates those in the character area by combining them. The divided area 6 is obtained.

When such division processing is completed, attribute determination processing 7 is performed. In the attribute discrimination processing 7, the attribute discrimination is performed for the divided area 6 obtained in the pre-processing, and the attribute information 8 of each area is output. The processing content of this attribute discrimination processing 7 is shown in FIG.

1 to 3 are flow charts of the attribute discriminating process using the attribute discriminating method of the embodiment of the present invention. In addition, S10-S26, S30-S40, S50- in the figure
S58 indicates a step of each process.

In this attribute discrimination process, when the data of the divided area 6 is input, the number of black pixels in the area is counted and it is determined whether or not the number of black pixels is less than or equal to the threshold T1 (S10).
If the number of black pixels is less than or equal to the threshold T1, this area is "dust".
The area is determined (S11). "Trash" here means
It is a black pixel block that is not the content of the document, such as a stain attached to an actual document or a shadow caused by a stain attached to the scanner surface when a document image is input. When such a black pixel block exists,
This “dust” area is identified because it may cause an erroneous determination.

If the number of black pixels is larger than the threshold value T1 in step S10, the aspect ratio of the area circumscribed rectangle is obtained in step S12, and whether the aspect ratio is larger than the threshold value T2 or not is further determined. It is determined whether or not the thickness (that is, the smaller one of the height and the width of the region) is equal to or smaller than the threshold value T3. If the aspect ratio is greater than or equal to the threshold value T2 and the thickness is less than or equal to the threshold value T3, this area is determined to be the field separator area (S13).

If the determination result in step S12 is NO, it is determined whether or not the thickness of the area is less than or equal to the threshold value T4. If the thickness is less than or equal to the threshold value T4, it is determined as a character area (character string) (S15). In the case of NO in step S14, the rectangularity of the area is obtained from the following equation, and it is determined whether the rectangularity is equal to or greater than the threshold value T5, and whether the thickness of the area is equal to or greater than the threshold value T6 (S16).

Squareness = number of black pixels in area / area of circumscribed rectangle of area (number of vertical pixels × number of horizontal pixels) In step S16, the rectangularity of the area is equal to or greater than threshold value T5 and the thickness is equal to threshold value T.
If it is 6 or more, this area is determined to be a photograph area (S1
7) If NO, proceed to step S18.

FIG. 2 shows a detailed flowchart of the contact area separation and attribute discrimination processing in step S18. In FIG. 2, first, it is determined whether or not there is a continuous black run in which a black pixel is longer than the threshold value T11 in the area (S30).
~ S32). When the black run does not exist in both the vertical and horizontal directions or exists in both the vertical and horizontal directions, the inclination correction process of step S19 of FIG. 1 and subsequent steps are performed. When the black run exists only in the horizontal direction, the rectangular area of the original image corresponding to the circumscribed rectangle of the area is obtained (S33), and the values at both ends in the vertical direction of the rectangular area of the original image are set to ys and ye, respectively. In step S34, the horizontal division processing is performed on the rectangular area of the original image, and if there is post-execution division, the rectangular area of the reduced image corresponding to all the divided areas is obtained (S35, S3).
6). If the area is not divided in the horizontal division processing, the rectangular area of the reduced image corresponding to the rectangular area before the horizontal division processing is obtained, and the processing from step S19 onward in FIG. 1 is performed.

In the horizontal division processing, when the area is divided, the length of the longest black run in each area is obtained, and the length of the longest black run is equal to or larger than the threshold value T12 and the thickness of the area is equal to or smaller than the threshold value T3. If so (S37), this area is determined to be a field separator (S38), and otherwise it is determined to be a character area (S39).

When the black run exists only in the vertical direction in step S32 of FIG. 1, the rectangular area of the original image corresponding to the circumscribed rectangle of the area is obtained (S40), and the rectangular areas of the rectangular area of the original image are aligned in the horizontal direction. Let the values be xs and xe, respectively. Step S
In step 41, vertical division processing is performed on the rectangular area of the original image, and if there is post-execution division, the rectangular area of the reduced image corresponding to all the divided areas is obtained (S36). If the area is not divided in the vertical division processing, the rectangular area of the reduced image corresponding to the rectangular area before the vertical division processing is obtained, and the processing from step S19 onward in FIG. 1 is performed. After that, the same processing as when the black run exists only in the horizontal direction is performed.

FIG. 3 is a flow chart showing the horizontal division processing in detail. First, in step S50, the number of divisions is initialized to 0, and the original image is horizontally scanned inside the circumscribed rectangle.
The scanning line is performed from y values ys to ye (S51).
If there is no black pixel in the scanning line in step S53, it is determined whether or not there is a black pixel in the previous line (S54), and if it exists, the scanning line is newly set to yes. Set and increase the number of divisions by one (S55, S5
6). Next, if there is at least one black pixel in the scanning line in step S53, it is determined whether or not there is a black pixel in the previous line (S57), and there is even one black pixel. If not, the scanning line is newly set to ys (S58). The above scanning is performed up to the final line of the circumscribed rectangle. Note that the vertical division processing is similar to the horizontal division processing,
The same processing is performed by changing ye and ys to xe and xs and changing the horizontal scanning to the vertical scanning.

In the inclination correction process of step S19 of FIG. 1, the length of the longest black run in the row direction and the length of the longest black run in the column direction are obtained, and the obtained length of the longest black run in the row direction and the width of the circumscribed rectangle are obtained. It is determined whether the ratio is greater than or equal to the threshold value T7 (S20).
When the ratio is equal to or greater than the threshold T7, it is determined whether or not the ratio of the length of the longest black run in the column direction to the height of the circumscribed rectangle is equal to or greater than the threshold T8 (S21). A determination is made (S22). In the case of NO in steps S20 and S21, the process proceeds to step S23.

In step S23, the ratio of the number of black / white inversions in the area (the sum of the number of black / white inversions in the vertical and horizontal directions) to the circumscribed rectangular area is obtained, and it is determined whether or not the area ratio is smaller than the threshold value T9. If it is less than the threshold value T9, it is determined to be a graphic area (S24). If step S23 is NO, the rectangularity of the area is equal to the threshold value T in step S25.
It is determined whether it is less than 10, and if it is the threshold value T10 or more, in step S26, another character area (independent character)
To determine.

Thresholds T1 to T13 in the attribute discrimination processing of FIGS. 1 and 2 are set as follows. For example,
For each region obtained by dividing the 50 dpi document image obtained by reducing the result of the OR operation of 8 × 8 pixels of the 400 dpi original document image into one pixel, , The size of “dust” is 4 pixels or less (T1 = 4), and the aspect ratio of the field separator is 40.
The above is defined (T2 = 40) and the thickness is 4 mm or less (T3 = 4). Further, the thickness of the character string is 5 mm or less (T in consideration of the size of characters used in the body of a general document).
4 = 5) is suitable.

Further, in order to obtain other threshold values, the degree of rectangularity in the area of independent characters, figures, photographs, and tables in the actual reduced document image, the height of the circumscribing rectangle of the longest vertical black and horizontal run, the ratio to the width, and The results of investigating the distribution of black and white inversion density are shown in FIGS.

FIG. 6 is a diagram showing the distribution of the rectangularity of the area for each attribute. Since most photographs have a rectangularity of 0.6 or more, it is appropriate to set T5 = 0.6. .. Also,
It is suitable that T6 is about 1 cm.

FIG. 7 shows a rectangle circumscribing a region having the longest black run length in the x-direction (horizontal direction) and the y-direction (longitudinal direction) obtained by the inclination correction process described later in each region. It is the figure which represented distribution of the ratio with respect to a size for every attribute. The X coordinate is the ratio of the length of the longest black run in the row direction to the width of the circumscribed rectangle, and the Y coordinate is the ratio of the length of the longest black run in the column direction to the height of the circumscribed rectangle. In the figure, t is a table area, c is a character area, and f is a graphic area. From FIG. 7, in the table area, both the row direction and the column direction have large values,
It is appropriate to set the threshold values T7 = 0.95 and T8 = 0.95.

FIG. 8 is a distribution diagram of the values of the rectangularity and the black and white inversion density in the independent character area and the graphic area. The X coordinate is the rectangularity of the area, and the Y coordinate is the black and white inversion density. In the figure, f is a graphic area and c is a character area. From this FIG.
It is appropriate that 9 = 0.2 and T10 = 0.3.

The black run length threshold value and the longest black run length threshold value T11, T12 in FIG. 2 are appropriately 32 pixels and 40 pixels, respectively, in consideration of the character size. .. Further, the threshold value T13 of the thickness of the region is defined as a thickness of 4 mm or less (T13 = 4) as in T3.

9A and 9B are views for explaining the concept of the inclination correction process (S19) in FIG. In a document image read in a horizontal state, the number of black pixels representing a horizontal line in the document is lined up on the same line, but in a document image read in a tilted state, the number of black pixels in the horizontal line depends on the degree of inclination. It moves to the adjacent line. Therefore, in the inclination correction process, since the numbers of black pixels thus shifted are arranged on the same line, several consecutive lines in the image area are
A reduction operation is performed to replace each row with one row obtained by taking the logical sum. In this process, the number of lines that can be replaced with one line differs depending on the tilt angle to be corrected and the size of the area. For example, assuming that the tilt allowable angle is about 1 degree, it is appropriate to be calculated by the following formula. is there. Number of lines to be replaced with one line = (width of area circumscribed rectangle) × sin
1 ° In the image obtained by this processing, the length of the longest black run in the row direction is the length of the longest black run in the row direction in this region.

Similar processing is performed in the column direction. 1 at this time
It is appropriate that the number of columns to be replaced with columns is calculated by the following equation, as described above. Number of rows replaced in one column = (height of rectangle circumscribing area) × si
n1 ° In the image obtained by this processing, the length of the longest black run in the column direction is the length of the longest black run in the column direction in this region. These row-direction and column-direction reduction operations can be performed simultaneously.

The present invention is not limited to the above embodiment,
For example, in the attribute determination process of FIG.
Various modifications are possible, such as omitting S11 or executing contact area separation and attribute determination processing S18 after inclination correction processing S19.

[0036]

As described above in detail, according to the present invention, an area in which a character (column) generated when a document image is reduced and a field separator are in contact with each other is returned to the original image and the area is redivided and the attribute is discriminated. Therefore, it can be easily and accurately determined.

[Brief description of drawings]

FIG. 1 is a content diagram of attribute discrimination processing showing an attribute discrimination method according to an embodiment of the present invention.

FIG. 2 is a diagram showing the processing content of step S18 in FIG.

FIG. 3 is a diagram showing the processing contents of step S34 in FIG.

FIG. 4 is a diagram showing an example of a reduced image.

FIG. 5 is an overall processing content diagram of image processing showing an embodiment of the present invention.

FIG. 6 is a rectangularity distribution diagram.

FIG. 7 is a circumscribed rectangle ratio distribution diagram.

FIG. 8 is a distribution diagram of rectangularity and black / white inversion density.

9 is an explanatory diagram of a tilt correction process in FIG.

[Explanation of symbols]

 1 simple binarization processing 2 original image memory 3 image reduction processing 4 reduced image memory 5 area division processing 6 area division data 7 attribute discrimination processing 8 attribute information

Claims (1)

[Claims] 1. A document image is simply binarized to create a binarized image (original image), the entire binarized image is further reduced, and then the reduced image is segmented into constituent elements, and the segmentation is performed. In the method of discriminating the attribute of the document image for discriminating the attribute of the region, the number of black pixels in the reduced image region, the aspect ratio of the rectangle circumscribing the region, the thickness of the region, and the rectangularity of the region are respectively obtained. A process of comparing with a threshold value for them, a process of determining whether or not there is a black run longer than a predetermined threshold value in the reduced image area, an area of the original image corresponding to the reduced image area, and the original image After dividing the area of the original image by obtaining a line in which no black pixel exists and a line in which at least one black pixel exists, the area of the reduced image corresponding to the divided original image area is obtained. Processing and the By calculating the length of the longest black run in the small image area and the thickness of the area and comparing it with the corresponding threshold value, and the inclination correction processing, the inside of the reduced image area with respect to the size of the circumscribed rectangle of the reduced image area The process of obtaining the ratio of the length of the longest black run in step S1, and comparing it with the corresponding threshold value, and the process of obtaining the black-white inversion density in the reduced image area and comparing it with the corresponding threshold value are performed. Characters (rows), figures, photographs,
A method of discriminating an attribute of a document image, characterized by discriminating each area of a table and a field separator.