Area separation system for document picture by discrete cosine transformation

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Industrial application field]

The present invention relates to a method of separating character and photo areas from a document in which characters and photographs are mixed, and more particularly to a method of separating character and photo areas by using a discrete cosine transform (hereinafter referred to as DCT). ..

【0002】

[Conventional technology]

In the conventional area separation technology, a full-color image is converted into a binary image to identify the character and the photographic area from the connected state of pixels, and the texture of the image is obtained in the original image state of the full-color image to obtain the texture of the character and the photograph. There was a method to determine the area. Typical of the latter are those using FFT, fractal dimension, wavelet transform, etc.

【0003】

[Problems to be Solved by the Invention]

In the conventional technology, the image captured from the image scanner is once binarized and the characters and the photo are discriminated by the connected state of the pixels, and the text and the photo are discriminated from the texture of the document image while remaining in full color. Has been done. As a method commonly used for texture analysis, there is a region separation method that applies Fourier transform, fractal dimension, KL transform, wavelet transform, etc., but all of them have a problem that conversion takes time.

【0004】

The present invention has been made in view of the problems of the prior art as described above, and an object of the present invention is to realize high-precision and high-speed separation of characters and photographic areas by using DCT.

【0005】

[Means for solving problems]

The area separation method for a document image by the discrete cosine transform of the present invention is obtained from an input device in an area separation method for separating a character area and a photographic area for a printed document input to an input device such as an image scanner. It is characterized in that a discrete cosine transform process is performed after performing image reduction conversion on a full-color document image, and characters and photographs are discriminated from the discrete cosine transform coefficient obtained by the discrete cosine transform process.

【0006】

[Action]

The present invention first adopts a given document image reduction means in order to adapt DCT, which has recently attracted attention in the color image coding method, to texture analysis and realize region separation, and further, DCT coefficient. Focusing on the coefficient that characterizes the character and the photo, the character and the photo area are discriminated for each block based on the evaluation function calculated from the statistical properties of the focused feature coefficient, and the photo area is extracted in a rectangular shape. The most important feature is to do.

【0007】

For the image that has undergone the image reduction conversion, a clear difference is observed in the DCT coefficient characteristics of the character image and the photographic image. In the present invention, since the reduced image is subjected to DCT processing, region separation is satisfactorily performed.

【0008】

[Example]

Hereinafter, examples of the present invention will be described with reference to the drawings.

【0009】

FIG. 1 is a partially rested configuration diagram of the present invention, and FIGS. 2 and 3 are diagrams showing a detailed configuration of the present invention.

【0010】

As shown in each figure, the printed document 201 in which characters and photographs input to the image input unit 101 (input device 202) are mixed is stored as an RGB 24-bit (8 bits each) full-color image 203.

【0011】

After that, the stored full-color image 203 is subjected to 1/2 or 1/3 reduction conversion by the reduction conversion process 102 to be a reduced image 204, and further, the YIQ conversion process 103 by the luminance image creation unit 104. Y component 205 is obtained. In this embodiment, the Y component is used as the information representing the luminance component as described above, but in addition to this, Y, Cr, Cb, CIE (1976) L^*a^*b^*The information representing the luminance component may be adopted with the color space as.

【0012】

In the present invention, the Y component 205 is subjected to DCT processing 105 (the DCT block is a square with 8 pixels on each side), and the counting analysis unit 106 performs DCT processing 105 out of 64 DCT coefficients as shown in FIG. The coefficient value for discriminating between characters and photographs is obtained from the coefficient group. In this embodiment, the absolute sum of the DC coefficient and the high frequency coefficient shown in FIG. 3 is used as the feature coefficient according to the coefficient characteristics of the reduced image. The DC coefficient is particularly affected by the background color (especially white) and takes a larger value than that of the photographic image. In addition, the high-frequency component shown by the diagonal line in Fig. 4 has a large dispersion in the cover character part of small changes, which is an important part in reproduction.

【0013】

In this embodiment, by defining the feature coefficient as described above, it is possible to separate the character area including the background image and the photographic area with high quality.

【0014】

Next, the character and the photographic area are discriminated based on the discriminant evaluation function that has been analytically derived in advance by the region discriminating unit 107.

【0015】

In the present invention, a two-dimensional (linear) discriminant function is applied, but it is also possible to extend it to n dimensions by combining a coefficient considered to be an important coefficient and then obtain an evaluation function. The (linear) discriminant function used in this example was obtained according to the following procedure.

【0016】

[Number 1]

<img file="JPH06339019A_D0001.tif" />Equation (1) shows an evaluation function that discriminates between a character image and a photographic image. Therefore, for character and photo area discrimination, the DCT coefficient of each block is obtained, and the feature coefficient values are substituted for x and y in Eq. (1). For example, it is determined as a character area. This determination process is executed for all blocks.

【0017】

Next, a method of extracting a rectangular area will be described. The extraction process includes area correction (image correction) after discrimination, and a detailed configuration is shown in FIG.

【0018】

As shown in FIG. 3, when it is determined as a "photograph area" by the evaluation function, there are blocks determined as a character area around and inside the photo area, and in some cases, it is not rectangular. Therefore, the image correction unit 108 corrects the inside of the region by a correction process using the adjacent region as shown in FIG.

【0019】

Next, in order to extract each region into a rectangular shape, labeling is performed, and the maximum and minimum values of the vertical and horizontal coordinates of each region as shown in FIG. 6 are detected. Finally, the area is rectangularly extracted according to these coordinates and output from the image output unit 109.

【0020】

The reason why the region is rectangularly extracted as shown in FIG. 6 is to facilitate the structural coding performed after that, and if the subsequent processing is different, it depends on the processing content. It may be changed, and it is not particularly important.

【0021】

[Effect of the invention]

As described above, according to the wood invention, it is possible to perform high-quality area separation between the character area (the area including the background area) and the photographic area at high speed for a document image in which characters and photographs are mixed. It becomes.

[Simple explanation of drawings]

[Figure 1]

It is a figure which shows the schematic operation of this invention.

[Figure 2]

It is a block diagram up to the discrimination process of this invention.

[Fig. 3]

It is a block diagram after the discrimination process of this invention.

[Fig. 4]

It is a diagram showing the important coefficient used to distinguish characters and photographs from the 64 D mark coefficients obtained in each block, and each grid corresponds to the DCT coefficient F (m, n). There is.

[Fig. 5]

It is a figure which shows the hole filling process used for image correction.

[Fig. 6]

It is explanatory drawing of the process which extracts the corrected photographic area rectangularly.

[Explanation of symbols]

101 Image input section 102 Reduction conversion processing 103 YIQ conversion process 104 Luminance image creation unit 105 DCT conversion process 106 Coefficient analysis unit 107 Area discriminator 108 Image correction unit 109 Image output 201 characters, mixed photo document 202 Input device 203 full color image 204 Reduced image 205 Y component 206 Evaluation function