JPH0214392A - Document area analyzing device - Google Patents

Abstract

PURPOSE:To automatically and promptly discriminate where a character area is, where a graphic area is, etc., by composing the title document area analyzing device of a compressed data creating and storing means and a projection discriminating means. CONSTITUTION:Compressed picture data 11 are once stored in a compressed picture memory 36 and read as compressed picture data 41 whenever a projection operation is performed by changing a projection area and a projection direction. An area extracting circuit 14 designates the projection area and the projection direction by a control signal 15. When this control signal 15 is inputted to a projection area designating circuit 37, the circuit 37 gives area and direction data 38 to indicate the projection area and the projection direction to an address for projection generating circuit 39, generates a projection address 40, and successively reads the compressed picture data 41 of the projection area from the compressed picture memory 36. In a counter circuit 42, the number of black picture elements are counted up, projection data 13 are obtained, and the data 13 are outputted. The counter circuit 42 is initialized at the first part of a projection line.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is applicable to character areas, graphic areas, table areas, photographic areas, etc.
It automatically analyzes document images obtained by scanning and reading documents that contain various areas, determines which areas are text areas and which areas are graphic areas, and selects only text areas if they are text areas. The present invention relates to such a document area analysis device for automatically reading characters.

[Conventional technology]

Conventionally, there have been no document area analysis devices that automatically analyze document images to determine which areas are text areas and which are graphic areas, and there are no reading devices that can read documents with only text areas printed. For documents containing a mixture of graphic areas, table areas, photographic areas, etc. in addition to character areas, there were only reading devices that manually specified the character area and read the characters accordingly.

[Problem to be solved by the invention]

Therefore, when reading a document in which graphic areas and the like coexist in addition to character areas, it is necessary to manually specify the character area, which is troublesome.

The present invention provides a document that enables automatic and high-speed designation of text areas without the need for manual designation even when reading a document in which graphic areas, etc. coexist in addition to text areas. It is an object of the present invention to provide an area analysis device, that is, a document area analysis device that can automatically and quickly determine which areas are character areas and which areas are graphic areas from a read image of a document.

[Means to solve the problem]

In order to achieve the above object, in the present invention, a document area analysis device is configured by a compressed data creation storage means and a projection determination means.

[Operation] The compressed data creation storage means divides the read pixel data obtained by scanning and reading the document into areas (referred to as unit areas) consisting of vertical A dots and horizontal B dots in the document. However, a.

b represents an integer of 2 or more), each unit area is given a binary value depending on whether the number of specific dots of logical 1 or 0 in each unit area is within a certain threshold range. to create compressed data of the read pixel data and store it as a compressed image.

Further, the projection determining means projects the compressed image read from the compressed data creation storage means while controlling the projection area and projection direction, and uses the obtained projection data to project the compressed image read from the compressed data creation storage means. The area type, such as a character area, a graphic area, or a photo area, is determined within the area.

The reason why a compressed image is created and then projected is that the area type can be determined more quickly. If an uncompressed original image is projected, it will take a long time to perform the discrimination, making it impractical.

Here, the operation of projection calculation will be explained with reference to FIG.

In FIG. 6, P represents a pattern P consisting of 6 dots horizontally and 6 dots vertically. Each hatched dot is a black dot, and each unhatched dot is a white dot.

It is assumed that such a pattern P is projected from the right side in the X direction. The number of black dots in each row viewed from the X direction is @AI
As described in 2.0.1, 4°, 2.0.0, and when this is displayed in levels according to the number of black dots, a pattern like B1 is obtained.

Obtaining such data as seen in +1jlA1 and pattern B1 is called projection calculation.

Therefore, it will be understood that if projection is performed in the Y direction, projection data as seen in column A2 and pattern B2 can be obtained for each column in the same manner.

〔Example〕

Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 7 is an explanatory diagram showing an example of a document to be analyzed. In the figure, 1 is a document, 2 is a diagram area therein, 3a, 3b, and 3c are character areas, and 4 is a photo area. One stage is composed of the diagram area 2 and the character area 3a, and the character area 3b, the photo area 4, and the character area 3
Another column is constructed by C, and after all, this document 1 shows an example of a two-column structure.

Each area, such as a character area or a diagram area, has its own characteristics in the way black pixels are distributed, so it is possible to determine from these characteristics whether the area is a character area or a photo area.

If document 1 has a height of HO and a width of WO mm, and the image size when document 1 is viewed as an image is a height of No dot and a width of MO dot, then the conversion coefficient between dots and mm can be obtained. However, this conversion factor is determined by the resolution of the image scanner used as a document reading device; for example, for a scanner with a resolution of 200 dpi, the conversion factor is B dot/proof.

Document image 1 generally has a large amount of data, and if it is projected as is, the time required for projection calculation will be long, so the document image is compressed prior to projection.

FIG. 8 is an explanatory diagram of this compression method. In the figure, document image 1 is shown as having an image size of N0 dots in height and MO dots in width. Normally, the unit is 1 dot, but due to compression it becomes KXK dot (
A square area with dots on both the vertical and horizontal sides is used as a unit area, and this becomes 1-bit data. In the figure = 8 dots
The case is shown below.

The compressed unit area 5, which became a unit area as a result of compression, is
This is an area where the X coordinate is defined as (81-7)≦X≦8■, and the Y coordinate is defined as (8J-7)≦Y≦8J.

However, 1≦1≦(MO/K), 1≦J≦[NO/K]
There is a relationship between [] is a Gaussian symbol (that is, (x)
=y means that y represents an integer not exceeding X).

FIG. 1 is a block diagram showing one embodiment of the present invention. In the figure, 6 is a document, 7 is an image scanner (optical character reading device), 9 is a document image memory, 10 is an image compression circuit, 12 is a projection calculation circuit, 14 is an area extraction circuit, 17
2 is a character area memory, and 20 is a character reading circuit.

Explain the operation. A document 6 is read by an image scanner 7 to output a digital image signal 8 and stored in a document image memory 9. On the other hand, the same digital image signal 8 is input to an image compression circuit 10, which compresses the KXK dot area into data having a unit area (1 bit data) and outputs compressed image data 11.

Next, the compressed image data 11 is input to the projection calculation circuit 12, and the area extraction circuit 14 uses the projection data 13 output from here to extract character areas and other areas. 17 and recorded there.

In order to extract a character area, the area extraction circuit 14 sends a control signal 15 for controlling the projection area several times to the projection calculation circuit 12, executes the projection calculation according to the control signal 15, and extracts the data from the data obtained in this way. Extract each area such as character area.

The character area data read from the memory 17 is sent to the document image memory 9 as a character area address 18, and if the character image data 19 is read from the memory 9, and is read by the character reading circuit 20 and the result is output, This means that only the character area in document 6 has been extracted and character reading has been performed.

FIG. 2 is a block diagram showing details of the image compression circuit 10 in FIG. 1. In the figure, 21 is a bit size register, and 23 is a count calculation circuit that counts the number of black dots (pixels) in a unit area of 1 line, adds the previous count value to it, and outputs the result. , 27 is a 1-line compressed image memory, 2
5 is an X-direction address generation circuit, 28 is a Y-direction address generation circuit, 30 is a first line initialization circuit, 33 is an upper/lower limit determination circuit, and 35 is a gate circuit.

Explain the operation. The digital image signal 8 is input to the bit size register 21, converted into parallel data in units of K bits, and input to the count calculation circuit 23 as parallel bit data. The count calculation circuit 23 regards the bit data as 1-line unit area data and counts the number of black dots (pixels) therein, and furthermore, the count value up to the previous time is written into the 1-line compressed image memory 27. Therefore, it is read out, added up, and sent to the l-line compressed image memory 27 as the number of black pixels in the unit area up to that point, 24, to be stored for each unit area.

The X-direction address I (1 described earlier with reference to FIG. 8) of the unit area of the 1-line compressed image memory 27 is generated by the X-direction address generation circuit 25 as an address signal 26, and is sent to the 1-line compressed image memory 27. Given. Further, the Y-direction address generation circuit 28 generates and sends out a line number signal 29 within the unit area, and when this is the first line, the first line initialization circuit 30 receives this and sends out an initial value signal 31. is output to initialize (zero clear) the count circuit 23, and then the count circuit 23 counts the number of black pixels in the bit data as described above and sends it to the 1-line compressed image memory 27 for storage.

In the second line, the counting circuit 23 counts the number of black pixels in the bit data for the second line, and then reads out the number of black pixels stored in the one-line compressed image memory 27 as that for the first line. , take the sum, and calculate the sum up to the second line (the sum of the number of black pixels in the bit data for the first line and the number of black pixels in the bit data for the second line) for l lines. Each unit area is written and stored in the compressed image memory 27.

Thereafter, in the same way, when the last line of one unit area is reached, the l-line compressed image memory 27 stores the black pixels included in one unit area (K lines). Since the sum total of the numbers is to be stored, this is taken into the upper and lower limit determination circuit 33 for each unit area as the black pixel number signal 32 in the unit area, and if it is within the predetermined upper and lower limit ranges (BL, B11). A binary signal 34 of the unit area is created by binarizing the unit area depending on whether it exists or not, and outputting it as compressed image data 11 through the gate circuit 35 which is open at this point.

FIG. 3 is a block diagram showing details of the projection calculation circuit 12 in FIG. 1. In the figure, 36 is a compressed image memory, 37 is a projection area designation circuit, 39 is a projection address generation circuit, and 42 is a counter circuit.

The compressed image data 11 is temporarily stored in the compressed image memory 36, and is read out as compressed image data 41 each time a projection calculation is performed by changing the projection area and projection direction. A region extraction circuit 14 specifies a projection region and a projection direction (X direction or Y direction as shown in FIG. 6) using a control signal 15. When this control signal 15 is input, the projection area specifying circuit 37 supplies the area/direction data 38 representing the projection area and the projection direction to the projection address generation circuit 39 to generate a projection address 40, and the compressed image memory 36 The compressed image data 41 of the projection area is sequentially read out from the projection area.

The counter circuit 42 counts up the number of black pixels in the manner described above with reference to FIG. 6, obtains and outputs the projection data 13. Counter circuit 42 is initialized at the beginning of the projection line.

FIG. 4 shows an example of the X-direction projection pattern 44 and the Y-direction projection pattern 45 of the document image 1. 46a.

46b is a threshold value for area extraction.

The projection value of compressed image data is whether it is a character area or not.
It takes various values depending on whether it is a graphic area, a table area, a photo area, etc. Therefore, this fact is used to perform area discrimination, but for now, in the example of Fig. 4,
By analyzing the projection patterns 44 and 45, it can be seen that the document image 1 has a two-column configuration.

Therefore, when the control signal 15 is applied to limit the projection area to the left column of the document image 1 and the image is projected in the X direction, projection patterns 47a and 47b as shown in FIG. 5 are obtained. When this is binarized using an appropriate threshold value 46c, character lines can be cut out as a series of small mountains in pattern 47b corresponding to character area 3, and pattern 4 corresponding to diagram area 2 can be cut out as a series of small mountains.
In 7a, the pattern is a group, so it can be seen that it is not a character area.

Similarly, by projecting the columns on the right, text areas and photo areas can be extracted.

As explained above, since projection calculations are performed by reading compressed image data from compressed image memory, the time required for projection calculations is shorter than when compressed data is not used. Since each area is binarized based on the number of black pixels within it, it is possible to determine whether the area is text, graphics, tables, photographs, etc.
Binarization (compression) is performed without impairing the characteristics such as, and each area can be easily distinguished.

〔Effect of the invention〕

As explained above, according to the present invention, not only document images but also compressed images are calculated and stored.
Projection calculations can be repeated at high speed until correct area discrimination is performed. Also, when calculating a compressed image,
Since the upper and lower limits of the number of black pixels, which serve as a criterion for binarizing a unit area, can be arbitrarily determined, it is possible to appropriately obtain a compressed image for extracting a character area in the case of a character area.

Furthermore, since the projection calculation is performed by specifying not only the projection area but also the projection direction, it is possible to discriminate between various areas.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a block diagram showing details of the image compression circuit 10 in FIG. 1, and FIG. 3 is a block diagram showing details of the projection calculation circuit 12 in FIG. 1. The block diagram, FIGS. 4 and 5 are explanatory diagrams each showing an example of a projection pattern of a document image, FIG. 6 is an explanatory diagram of projection calculation, and FIG. 7 is an explanatory diagram showing an example of a document to be analyzed. FIG. 8 is an explanatory diagram of how to compress a document image. Explanation of symbols 6... Document, 7... Image scanner (optical character reading device), 9... Document image memory, 10... Image compression circuit, 12... Projection calculation circuit, 14...・Area extraction circuit, 17...Character area memory, 20...Character reading circuit, Agent Patent attorney Akio Namiki Agent Patent attorney Kiyota Matsuzaki 1 Illustration 2 Illustration 3 +3 1g5 Illustration $b Ij

Claims (1)

[Claims] 1) The read pixel data obtained by scanning and reading a document is divided into regions (referred to as unit regions) each consisting of a dot in the vertical direction and b dots in the horizontal direction in the document (however, (a and b each represent an integer of 2 or more), each unit area is determined depending on whether the number of specific dots of logical 1 or 0 in each unit area is within a certain threshold range. compressed data creation storage means for binarizing and creating compressed data of the read pixel data and storing it as a compressed image; and when projecting the compressed image read from the compressed data creation storage means, the projection area and projection The compressed image is projected while controlling the direction, and the compressed image is projected from the obtained projection data to determine the type of area, such as a character area, a graphic area, or a photographic area, in the document constituting the compressed image. document area analysis device.