JPH02193272A - Image area identification device - Google Patents

Abstract

PURPOSE:To surely identify a binary area and a half tone area by joining together a background area and an area whose edge is detected and setting its area as a character image area candidate, and setting other area as a half tone area candidate. CONSTITUTION:As for an input image, an edge area is extracted by an edge detecting part 1, and a background area is extracted from a background detecting part 2. Two extracting signals become a character area candidate synthesized by a synthesizing part 7 by an OR processing, and other area becomes a half tone area candidate, and a signal for showing them is inputted to an area deciding part 6. In the area deciding part 4, the number of picture element decided to be a character area candidate in MXN picture elements for the area deciding part is counted, and when a counted value is larger than threshold determined in advance, it is decided to be a character area, and when the counted value is smaller than its threshold, it is decided to be a half tone area. In such a manner, the background area is identified as the character area, and the binary area and the half tone area can be surely identified.

Description

Detailed Description of the Invention (Industrial Field of Application) This invention is a method for detecting two images from a document image containing a mixture of binary images represented by character line images, etc., and halftone images represented by photographic images, etc.
The present invention relates to a document image area identification device for identifying value areas and halftone areas.

(Prior Art) In order to satisfactorily reproduce a binary image and a halftone image included in a document as a digital image, it is necessary to perform processing according to the characteristics of each of the binary region and the halftone region. Therefore, it is necessary to distinguish between binary areas such as character line images and halftone areas such as photographic images. Furthermore, when considering the improvement of image compression in communication, such identification of image areas becomes essential.

As a conventional method, there is a method for identifying image areas based on density differences within blocks (for example, Japanese Patent Laid-Open No. 58-20537
(See Publication No. 6). In this method, the density characteristics of text and line images have a large difference in density between adjacent pixels, as shown in the density characteristics in Figure 2, while the density characteristics of photographic images have characteristics as shown in the density characteristics in Figure 3. In addition, images are identified by utilizing the property that the difference in density is gradual. That is, the maximum density difference within the block is determined for each block of MXN pixels, and when the density difference is larger than a threshold value, it is identified as a binary area, and when it is smaller, it is identified as a halftone area.

(Problem to be solved by the invention) However, in this conventional method, small characters are
The resolution is lower due to the F characteristic, and the density characteristics are similar to the halftone image (so it is mistakenly determined to be a halftone region, and therefore halftone processing is performed, so the image quality of the reproduced image deteriorates. Ta.

Furthermore, in halftone areas, edge components are misidentified as text areas and character processing is performed, resulting in deterioration of the image quality of halftone images.

The present invention eliminates the drawbacks of the prior art, and can accurately identify binary regions such as character line images in document images and halftone regions such as photographic images, and is particularly superior to conventional methods. An object of the present invention is to provide an image area identification device that can improve the reproducibility of small characters, prevent misidentification of edges in halftone areas, and obtain a good reproduced image. It is something.

(Means for Solving the Problems) As shown in FIG. 1, the present invention includes an edge detection unit 1 that detects edge areas in an image of a document containing a mixture of character line images and photographic images; The background detection unit 2 detects the background area, and the edge area detected by the edge detection unit 1 and the background area detected by the background detection processing unit 2 are synthesized by a logical sum operation to form a character area candidate, and the other intermediate For each pixel of the character area candidate and halftone area candidate obtained by the synthesis unit 3 as a tone area candidate, and the character area candidate and halftone area candidate obtained by the synthesis unit 3, This is an image area identification device characterized by comprising an area determination unit 4 that determines whether the image area is a character area or a halftone area.

(Operation) The edge detection section 1 extracts edge regions from the input image, and the background detection section 2 extracts background regions. The two extracted signals are combined by a combining section 3 through OR processing to form a character area candidate, and the other areas are used as halftone area candidates, and signals representing these are input to an area determining section 4. The area determination section 4 counts the number of pixels determined to be character area candidates within MXN pixels for area determination, and when the counted value is larger than a predetermined threshold value, it is determined to be a character area, and the threshold value is determined. When it is smaller than the value, it is determined to be in a halftone area.

According to the present invention, as described above, the background area and the edge-detected area are combined as character image area candidates, and the other areas are used as halftone area candidates, so that the background area is identified as the character area. Therefore, the binary area and the halftone area can be reliably distinguished. Further, since region determination is performed for character image region candidates and halftone region candidates based on the number of pixels of the character region candidates within MXN pixels around the pixel of interest, -layer reliable region identification is possible. However, since the region determination is performed pixel by pixel, there are fewer erroneous region determinations (and the reproducibility of lowercase letters is high), making it possible to reproduce high image quality.

(Example) Next, the content of the present invention will be explained in detail based on the example shown in FIG. FIG. 1 shows an embodiment in which the image identification device of the present invention is applied to an image processing device, which consists of an image identification processing section according to the invention and an image processing section. The image identification processing section includes an edge detection section 1 that detects edge components in the input image signal, a background detection section 2 that detects the background part in the input image signal, and OR processing of the outputs of these detection sections 1 and 2. It consists of a combining section 3 that performs the compositing and an area determining section 4 that determines the area. The image processing section includes a character processing section 5 that performs binary processing on the input image signal, a halftone processing section 6 that performs halftone processing on the input image signal, and a character processing section 5 according to the judgment of the area judgment section 4. ) consists of an image composition section 7 which selects and outputs the output of the halftone processing section 6.

The input image signal has density characteristics depending on the attributes of each image. First, the edge detection unit l
This performs contour extraction, so-called edge detection. FIG. 2 is a schematic diagram showing the density characteristics of a character image and the corresponding edge output.

Since the character image area has a large density difference, almost the area corresponding to the outline of the character is extracted. As a result, the outline of a normal-sized character is extracted. However, when the character size becomes small (as the resolution decreases due to the MTF characteristics of the input system and the density gradient becomes gentler, contour areas that are not detected as edges occur. Figure 3 shows the density characteristics of halftone images and their corresponding responses). 2 is a schematic diagram showing an edge output. Since a halftone image has a gentle density gradient, edge components are not extracted for the most part, except for a few.

Edge detection is usually a method used in multivalued images, and can be configured with a 3×3 or 5×5 digital filter as shown in FIG. Also 3X
It can also be configured by a method of determining the density difference within a 3.5×5 pixel area.

Next, the background detection unit 2 extracts the background area component from the input image signal. The background area refers to the background density level of the paper surface of a document image. Therefore, in order to extract the background area, we
Rather than using it for value conversion, as shown in FIG. 4 or FIG. 5, a lower value of the density level is used as a threshold value, and a portion below that value is extracted.

Background extraction is based on the fact that the character area is structurally a binary image and is composed of black pixels and white pixels. The identification rate can be increased by treating the background area as a character area.

In the character area, most of the areas other than the edge area are background areas. Therefore, the edge region output detected by the edge detection section 1 and the background region output detected by the background detection section 2 are combined into a synthesis section 3.
When combined by performing OR processing, most of the area becomes a character area. On the other hand, in the halftone region (~), there is almost no low density level that corresponds to the background level. Therefore, there are almost no areas that are determined to be halftone areas but are changed to text areas by background output. Therefore, the identification rate can be increased by treating the background area as a character area.

An edge area output indicating whether each pixel detected by the edge detection unit 1 is an edge component, and a background area output indicating whether each pixel detected by the background detection unit 2 is a background component. By performing an OR operation and synthesizing them by the synthesizing section 3, a signal consisting of a character area candidate and a halftone area candidate is formed. However, if this continues, there is a risk that misidentifications may still occur, such as in the case of small characters. Therefore, the above-mentioned misidentification is eliminated by block identification using MXN pixels, and accurate judgment identification is performed by the area determination unit 4.

The area determination unit 4-1 identifies a character area and a halftone area based on the number of pixels of a character area candidate within the surrounding M×N pixels for each pixel. A schematic diagram of area determination is shown in FIG. As a feature, the fact that misidentification occurs locally due to the processing described above is utilized.

That is, if the number of pixels determined to be character area candidates within an area of MXN pixels is larger than the threshold value, the area is determined as a character area, and if it is less, it is determined as a halftone area. Also, the size of the area is smaller than that of the conventional T
The guideline is the minimum character size that cannot be reproduced using the /I separation method. As shown in FIG. 7, the conventional method treats the background area as a halftone area, so small characters may be recognized as a halftone area, but the method of the present invention treats the background area as a character area, so it is not accurate. Identification can be made.

In normal region determination, such processing is performed in units of blocks of MXN pixels, but in this embodiment, it is performed in units of pixels. That is, in order to determine the image area of the pixel of interest, MXN pixels, which are neighboring areas, are referred to, and it is determined whether the image of interest is a character area or a halftone area based on the number of black pixels. By doing so, the identification accuracy is improved and the hardware configuration is also simplified.

FIG. 9 shows an example of a 71-code configuration of processing for determining an image area based on the number of pixels in an area within MXN (5×5 in this example). A counter 11 counts the number of pixels in the sub-scanning direction for five lines of data. The output 3-bit data is delayed by latches 12 to 15, and a counter 16 counts the number of pixels for 5 main scanning dots. With the above configuration, it is possible to count the number of black pixels within 5×5 pixels with respect to the center pixel. Finally, this number of pixels is compared by a comparator 17, and it is determined pixel by pixel that if it is higher than the threshold set in the threshold setting section 18, it is a character area, and if it is low, it is a halftone area. be able to.

An image obtained by binary processing the input image by the character processing section 6 and an image obtained by halftone processing by the halftone processing section 6 are input to the image synthesis section 7 . Therefore, by switching the output based on the above-mentioned determination signal, it is possible to select an image that has undergone processing suitable for a binary image and a halftone image, and output a high-quality image.

(Effects of the Invention) According to the present invention, the background area and the edge-detected area are combined as character image area candidates, and the other areas are used as halftone area candidates, so that the background area can be identified as a character area. Therefore, the binary area and the halftone area can be reliably distinguished.

Further, since region determination is performed for character image region candidates and halftone region candidates based on the number of pixels of the character region candidates within MXN pixels around the pixel of interest, -layer reliable region identification is possible. Moreover, since region determination is performed pixel by pixel, erroneous determinations in region determination are reduced and the reproducibility of lowercase letters is increased, making it possible to reproduce high image quality.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the overall configuration of an embodiment of the present invention. FIG. 2 is a schematic diagram showing the characteristics of a character image, and FIG. 3 is a schematic diagram showing the characteristics of a photographic image. Figure 4 is a schematic diagram showing the characteristics of the background output of the character area, Figure 5
The figure is a schematic diagram showing the characteristics of the background output of the photo area. FIG. 6 is a diagram for explaining area identification, in which (a) shows an original document and (b) shows an image input to the area determination section. FIG. 7 is a diagram illustrating the size of blocks for area identification. FIG. 8 is a diagram showing an example of an edge detection filter. FIG. 9 is a diagram showing an example of a circuit for determining the number of pixels in a block. 1... Edge detection section, 2... Background detection section, 3...
Synthesis unit, 4... Area determination unit, 5... Character processing unit, 6
Halftone processing section, 7... image composition section. Figure Figure Figure C Figure Figure (Kyu) (i))

Claims (1)

[Claims] An edge detection unit that detects an edge area in an image of a document containing a mixture of character line images and photographic images; a background detection unit that detects a background area in the image; A compositing unit synthesizes the edge area detected by the background detection unit and the background area detected by the background detection unit to form a character area candidate, and the other area is a halftone area; an area determination unit that determines whether each pixel in the halftone area is a character area or a halftone area based on the number of pixels of character area candidates existing within M×N pixels surrounding the pixel; An image area identification device characterized by: