JPH10108012A - Image area separating device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provided the image area separate device where a character area and a pattern area are separated with high precision from an original of lots of kinds including a pseudo medium tone processing original and a copy original. SOLUTION: At least outputs of two sections among three sections as an edge area extract section 101, a dot area extract section 102 and a white background area extract section 103 are selected by multiplexers 104-106 depending on the kind of an original and given to an AND section 107, where a character area and a pattern area are separated and discriminated. In the case of an original subject to pseudo-intermediate tone processing, outputs of all the area detection sections 101, 102, 103 are selected and ANDed, and areas of edge, non-dot and white background areas are discriminated to be character areas, and areas other than these are discriminated to be pattern areas.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the field of image processing, and more particularly to a technique for separating a character area and a picture area in an image read by a scanner or the like.

[0002]

2. Description of the Related Art For example, in a digital copying machine, when reading and reproducing an image of an original in which characters and patterns are mixed, a high-gradation process is applied to a pattern area in order to obtain a high-quality reproduced image. It is desirable to perform processing with emphasis on resolution for the character area. In order to realize such processing, it is necessary to separate a character area and a picture area in a document image with high accuracy.

Regarding such image area separation, Japanese Patent No.
As described in Japanese Patent Application Laid-Open No. 507948 (JP-A-2-193272), the background area and the edge area are used as character areas, and the other areas are used as halftone area candidates. There is a method that aims to But,
This method has a problem that it is easy to erroneously determine a character area in a highlighted portion of a halftone dot. As another conventional technique, there is a method disclosed in Japanese Patent Application Laid-Open No. 3-153167 in which a character area and a picture area are separated by using a halftone area determination and an edge area determination. According to this method, the accuracy of region determination in a highlight region of a halftone dot is improved.

In the prior art, Japanese Patent Application Laid-Open No. 3-15316 is disclosed.
As described in Japanese Patent Publication No. 7, the type of manuscript is
Limited to photos / dots. However, with the spread of low-cost color printers in recent years, the number of opportunities to copy a color original that has been subjected to binary pseudo halftone processing such as low line ruling dither and error diffusion has increased. In addition, there has been an increasing number of opportunities for generation copying in which a copy original reproduced by a color copying machine is reproduced again by a color copying machine. As the types of originals handled in this way increase, in the image area separation method that combines two limited feature amounts as in the related art, erroneous determination in the separation of the character area / picture area increases, resulting in high image quality. It is no longer possible to obtain a proper playback image.

[0005]

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to increase the height of the character area / picture area for a variety of originals such as an error diffusion original, a dither original, and a copy original, which have recently increased the number of copies. An object of the present invention is to provide an image area separating apparatus capable of performing accurate separation. It is another object of the present invention to provide an image area separating apparatus that can optimize a method for determining the separation of a character area / picture area according to the type of a document to be handled.
SUMMARY OF THE INVENTION It is an object of the present invention to provide an image area separating apparatus for separating a character area / picture area of a document subjected to pseudo halftone processing with high accuracy.

[0006]

SUMMARY OF THE INVENTION An image area separating apparatus according to the present invention includes a feature amount extracting means for extracting at least three kinds of feature amounts from an image signal of a document, and at least three kinds of feature amounts extracted by the feature amount extracting means. A feature amount selecting means for selecting at least two types of feature amounts from the feature amounts according to the type of the document, and a character on an image of the document using the feature amount selected by the feature amount selecting means. Means for making a separation determination between the area and the picture area. The feature amount extracting means includes at least means for extracting an edge area, means for extracting a halftone dot area, and means for extracting a white background area. Regarding the feature amount selection according to the manuscript type, when the manuscript type is a manuscript including a halftone area, the feature amount selecting means selects the extraction result of the edge area extracting means and the halftone area extracting means. (Claim 3). If the type of the original is an original including a region subjected to pseudo halftone processing, the extraction result of the edge region extraction unit, the extraction result of the halftone dot region extraction unit, and the extraction result of the white background region extraction unit are output by the feature amount selection unit. Selected (Claim 4). When the type of the original is a photographic paper or an original including a solid area, the extraction result of the edge area extracting means and the extraction result of the white background area extracting means are selected by the feature amount selecting means (claim 5). When the type of the original is a copy original, the feature amount selecting means selects the extraction result of the edge area extracting means and the extraction result of the white background area extracting means (claim 6).

Further, according to the present invention, there is provided an image area separation apparatus suitable for image area separation of a document subjected to pseudo halftone processing,
In the apparatus, an edge area, a halftone area, and a white background area are respectively extracted from an image signal of a document, and an area excluding the edge area, the white background area, and the halftone area extracted by the means are all overlapped. Means for determining the area as a character area and determining the other area as a picture area.

[0008]

FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG. 1, an image signal 100 obtained by reading and quantizing an original using an image input device such as a scanner (not shown) is input to an edge region extraction unit 101, a halftone dot region extraction unit 102, and a white background region extraction unit 103. .
The edge area extraction unit 101 is a part that extracts a feature amount appearing at an edge portion of a character area or performs area determination based on such a feature amount. The dot area extraction unit 102
This is a part for extracting a feature amount appearing in a halftone dot region or determining an area based on such a feature amount. The white background region extraction unit 103 is a part that extracts a feature amount appearing in a white background region or performs a region determination based on such a feature amount.

The output of each of the area extraction units 101, 102, 103 is given to one input of a corresponding two-input, one-output multiplexer (MPX) 104, 105, 106. The other input of each of the multiplexers 104, 105, 106 is always supplied with "1". Each of the multiplexers 104, 105, and 106 has an input selection signal
A signal corresponding to the type of the document to be processed is sent to a document type selection unit 10.
8 The document type selection unit 108 and the multiplexers 104, 105, and 106 constitute a feature value selection unit that selects a feature value used for character / picture area determination according to the type of document. Document type selection unit 10
Reference numeral 8 denotes, for example, a signal for giving a signal corresponding to the document type manually designated by an operation unit (not shown), or a signal for automatically recognizing the document type by pre-scanning the document and outputting a signal corresponding to the recognition result. Is to give.
The outputs of the multiplexers 104, 105, and 106 are input to the AND unit 107. The output of the AND unit 107 is
It represents the final result of character / picture area separation determination, and is "1" for pixels in the character area and "0" for pixels in the picture area.

As the area determination method in the edge area extraction unit 101 and the halftone area extraction unit 102, for example, “Ouchi, Imao, Yamada,“ Image area separation method for character / picture (halftone, photograph) mixed image) ”, IEICE Transactions Vol.J7
5_D_2, No. 1, 1992_01 ".

The character area has many high-level density pixels and low-level density pixels (hereinafter referred to as black pixels and white pixels).
At the edge portion, these black pixels and white pixels are continuous. The edge area extraction unit 101 detects a character edge based on the continuity of each of the black pixels and the white pixels. A specific example of the character area extraction unit 101 is shown in FIG. 2 and described.

In FIG. 2, the ternarization processing section 201
Binarization of the input image signal 100 using the threshold values TH1 and TH2 (white pixels <TH1, TH1 ≦ halftone pixel <T
H2, TH2 ≦ black pixel). Threshold values TH1, TH2
Is, for example, 256 when the input image signal is from 0 to 255.
TH1 = 20, TH2 when expressed by gradation (0 = white)
= 80.

For the image signal after the ternarization, the black pixel continuity detecting unit 202 detects a portion where black pixels are continuous and the white pixel continuity detecting unit 203 detects a portion where white pixels are continuous by pattern matching. I do. In this embodiment, a pattern of 3 × 3 pixels shown in FIG. 3 is used for this pattern matching. The black pixel continuity detecting unit 202 sets a pixel (the center pixel of 3 × 3 pixels in this example) matching any one of the patterns shown in the upper part of FIG. Reference numeral 203 denotes a pixel (3 × 3) that matches one of the patterns shown in the lower part of FIG.
The central pixel of the three pixels) is a white continuous pixel.

The neighborhood detection unit 204 checks the detection results of the black pixel continuity detection unit 202 and the white pixel continuity detection unit 203 to determine whether the black continuity pixel and the white continuity pixel are in the vicinity. And a non-edge area are determined. More specifically, in this embodiment, 5 ×
When one or more black continuous pixels and one or more white continuous pixels exist inside each block of a size of 5 pixels,
The block is determined as an edge area, and if not, the block is determined as a non-edge area. And
Corresponding to the pixel in the block determined as the edge area "
1 is output, and "0" is output corresponding to the pixel in the block determined to be the non-edge area.

In the halftone dot region, pixels having a high density value and pixels having a low density value appear alternately and periodically. The halftone dot area extraction unit 102 identifies a halftone dot area by detecting the extreme pixel having the high density value or the low density value.
A specific example of the halftone dot region extraction unit 102 will be described with reference to FIG.

In FIG. 4, an extreme pixel detection unit 301 includes:
The extreme pixel is detected by the calculation. In this embodiment, as shown in FIG. 5, when the following conditions A and B are simultaneously satisfied in a block of 3 × 3 pixels, the center pixel is detected as an extreme pixel. Condition A: The density level (L) of the center pixel is higher or lower than that of any surrounding pixels. Condition B: the density level (L) of the center pixel and the density levels (a, b) of the paired pixels diagonally across the center pixel
Are in the relationship of | 2 × Lab−> TH for all four pairs. Here, TH is a fixed threshold.

If one or more extremal pixels detected by the extremal pixel detecting unit 301 are present in a block of 4 × 4 pixels, the halftone dot area detecting unit 302 regards the block as a dot candidate area. If it is determined that there is no extremal pixel, the block is determined as a non-dot halftone dot candidate area. In response to this determination result, the halftone area correction unit 303 makes a final halftone / non-halftone determination. In the present embodiment, as shown in FIG. 6, 3 × 3 blocks centered on the block of interest (one block is 3 × 3
Pixel), if four or more blocks are halftone dot candidate areas, the block of interest is set to a halftone dot area; otherwise, the block of interest is set to a non-halftone dot area. Then, “0” is output corresponding to the pixel in the block which is regarded as the halftone area, and “1” is output corresponding to the pixel in the block which is regarded as the non-halftone area.

The white background area extraction unit 103 determines whether the background is white. A specific example of the white background area extraction unit 103 will be described with reference to FIG. In FIG. 7, the binarization processing unit 404 converts the input image signal 100 into a binary signal using a threshold value THW.
Value. That is, a pixel having a value equal to or less than THW is defined as a white pixel, and a pixel having a value exceeding THW is defined as a black pixel.

The pattern matching unit 402 converts the image signal after binarization into blocks of 4 × 4 pixels.
A white pixel block (a block of four continuous white pixels in the vertical or horizontal direction) in units of × 1 pixels or 1 × 4 pixels is detected. And
A block in which a white pixel block is detected is defined as a white candidate block.

As shown in FIG. 8, the white correction unit 403 includes a 9 × 9 block (1
When one or more white candidate blocks exist in each of the four 4 × 4 block areas shaded in a 4 × 4 pixel size block, the focused white candidate block (the center block in FIG. 8) is used. It is determined as a white background area. Otherwise, the target white candidate block is determined to be a non-white background area. Then, “1” is output corresponding to the pixels in the white background area, and “0” is output corresponding to the pixels in the non-white background area.

A document type to be processed and a document type selection unit 108
Multiplexers 104, 105, 10 based on signals from
The relationship with the input selection (feature amount selection) of No. 6 is as shown in FIG. First, when the target document is a dot document, a sufficiently accurate character / text is determined based on the dot region determination and the edge region determination.
Image area separation of the picture is possible. Therefore, the multiplexers 104 and 105 are controlled so as to select the outputs of the edge area extracting unit 101 and the halftone dot area extracting unit 102 as inputs, and the multiplexer 106 is controlled so as to select "1" as an input. Therefore, in the case of this original, only when the output of the edge area extracting unit 101 is “1” (edge area) and the output of the halftone area extracting unit 102 is “1” (non-dot area). , AND unit 107
Is "1" (character area) and "0" (picture area) under other conditions.

If the target document is a document subjected to pseudo-halftone processing such as error diffusion or dither, it is difficult to obtain sufficient image area separation accuracy only by judging a single dot area, edge area, and white background area. Therefore, the multiplexers 104 and 10
5 and 106 are controlled so that the outputs of the edge region extraction unit 101, the halftone dot region extraction unit 102, and the white background region extraction unit 103 are all selected as inputs. Therefore,
The output of the edge area extraction unit 101 is “1” (edge area), the output of the halftone dot area extraction unit 102 is “1” (non-dot area), and the output of the white background area extraction unit 103 is “1”.
Only in the case of “1” (white background area), the output of the AND unit 107 becomes “1” (character area), and under other conditions, the output becomes “1” (character area).
0 "(pattern area). That is, the edge area extracting unit 1
01 and the halftone dot area extraction unit 10
2, a region excluding the halftone dot region (non-dot region), a region where all the white background regions extracted by the white background region extraction unit 103 overlap each other are defined as a character region, and the other regions are defined as a picture region. Image area separation.

When the target document is a document containing photographic paper or a solid area, the halftone dot area determination has no effect on image area separation. Therefore, the multiplexers 104 and 106 are controlled so as to select the outputs of the edge region extraction unit 101 and the white background region extraction unit 103 as inputs, and the multiplexer 105 is set to “1”.
Controlled to select input. Therefore, the reason why the output of the logical product unit 107 is “1” (character area) is that the output of the edge area extraction unit 101 is “1” (edge area) and the output of the white background area extraction unit 103 is Only when it is "1" (white background area).

When the target document is a copy document, the accuracy of the halftone dot region determination is generally insufficient, so that it is possible to perform more reliable region separation without using the result of the determination. Therefore, in this case, the multiplexers 104 and 106 are controlled to select the outputs of the edge area extraction unit 101 and the white background area extraction unit 103, respectively, and the multiplexer 105
Is controlled to select the "1" input. Therefore, the output of the edge area extraction unit 101 is “1” (edge area), and the output of the white background area extraction unit 103 is “1”.
Only under the condition of (white background area), the output of the AND unit 107 becomes “1” (character area).

When only the four types of originals shown in FIG. 9 are to be used, the edge area determination is always used for character / picture area determination. Therefore, the multiplexer 104 is omitted and the edge area extraction unit 101 May be directly input to the logical product unit 107. However, if the case of dealing with more various types of originals is considered, the multiplexer 10
It is better to interpose 4.

The image area separating apparatus of the embodiment described above can be widely used as an image area separating apparatus of a digital copying machine, a facsimile, a scanner, and various other image processing apparatuses. In this embodiment, the edge area determination, the halftone area determination, and the white background area determination are performed simultaneously and in parallel. However, these area determinations are sequentially performed one by one for each unit. May be stored, and the final character / picture area determination may be performed when all the area determination results are obtained for a certain unit. In addition, although three types of feature amounts, such as an edge, a halftone dot, and a white background, are used for image area separation, another feature amount can be added. It is advantageous in terms of processing speed if each section of the image area separating device is realized by hardware, but if speed does not matter, each section may be realized by software on a computer system. Also, a computer storage medium storing such software of the present invention is prepared, and the software of the present invention is loaded into a general-purpose computer system using the computer storage medium, so that the program is stored on a general-purpose computer system. It is also possible to carry out the image area separation according to the invention.

[0027]

As described above, according to the present invention, the following effects can be obtained. According to the first aspect of the present invention, compared to a method using two limited types of feature values,
An image area separating apparatus capable of separating character areas / picture areas with high accuracy from various types of originals can be realized. According to the second aspect of the present invention, the separation of the character area / picture area is performed for any of an original including a halftone image, an original including photographic paper or a solid area, an original subjected to pseudo halftone processing, and a copied original. An image area separation device that can be performed with high accuracy can be provided. Claim 3
According to the invention according to the invention described in claim 5, the invention including the halftone image is subjected to pseudo halftone processing such as error diffusion or dither according to the invention according to claim 4, For example, for a document including a photographic paper or a solid area, and according to the invention of claim 6, for a copy document, a character area /
High precision separation of the picture area is possible. According to the seventh aspect of the present invention, it is possible to realize an image area separating apparatus capable of separating a character area / a picture area of a document subjected to pseudo halftone processing with high accuracy.

[Brief description of the drawings]

FIG. 1 is a block diagram of an image area separating apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating an example of an edge area extraction unit.

FIG. 3 is a diagram illustrating an example of a pattern used for pattern matching for detecting a portion where black pixels or white pixels continue.

FIG. 4 is a block diagram illustrating an example of a halftone dot area extraction unit.

FIG. 5 is a diagram for explaining extreme value pixel detection.

FIG. 6 is a diagram for explaining halftone dot area correction;

FIG. 7 is a block diagram illustrating an example of a white background region extraction unit.

FIG. 8 is a diagram for explaining white correction.

FIG. 9 is a diagram illustrating a relationship between a document type and a selected feature amount.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 input image signal 101 edge region extracting unit 102 halftone region extracting unit 103 white background region extracting unit 104, 105, 106 multiplexer 107 logical product unit 108 document type selecting unit 201 ternarization processing unit 202 black pixel continuity detecting unit 203 White pixel continuity detection unit 204 Neighborhood detection unit 301 Extreme pixel detection unit 302 Halftone dot detection unit 303 Halftone dot correction unit 401 Binarization processing unit 402 Pattern matching unit 403 White correction unit

Claims (7)

[Claims] 1. A feature amount extracting unit for extracting at least three types of feature amounts from an image signal of a document, and at least three types of feature amounts extracted by the feature amount extracting unit.
A feature amount selecting unit for selecting at least two types of feature amounts in accordance with the type of the document; and separating a character region and a picture region on the image of the document using the feature amounts selected by the feature amount selecting unit. An image area separating apparatus comprising: means for performing determination. 2. The image area separating apparatus according to claim 1, wherein
An image area separating apparatus characterized in that the feature amount extracting means includes at least means for extracting an edge area, means for extracting a halftone dot area, and means for extracting a white background area. 3. The image area separating apparatus according to claim 2, wherein
An image area separating apparatus, wherein, when the type of a document is a document including a halftone area, an extraction result of the edge area extraction means and an extraction result of the halftone area extraction means are selected by the feature amount selection means. 4. The image area separating apparatus according to claim 2, wherein
When the type of the original includes a pseudo-halftone-processed area, the feature amount selection unit selects the extraction result of the edge region extraction unit, the extraction result of the halftone dot region extraction unit, and the extraction result of the white background region extraction unit. An image area separation device. 5. The image area separating apparatus according to claim 2, wherein
When the type of original is photographic paper or an original that includes a solid area,
An image area separating apparatus, wherein an extraction result of an edge area extracting means and an extraction result of a white background area extracting means are selected by a feature amount selecting means. 6. The image area separating apparatus according to claim 2, wherein
An image area separating apparatus, wherein when a type of a document is a copy document, an extraction result of an edge region extraction unit and an extraction result of a white background region extraction unit are selected by a feature amount selection unit. 7. A means for extracting an edge area, a halftone area, and a white background area from an image signal of a document, respectively, and an area excluding the edge area, the white background area, and the halftone area extracted by the means overlap each other. An image area separating device, comprising: means for determining an area that has been rendered as a character area and determining the other area as a picture area.