KR0185837B1 - An apparatus for determining texture region in image signal - Google Patents

Abstract

The present invention relates to a texture region discrimination apparatus, comprising: a low pass masking unit 13 for receiving a discrete cosine transformed DCT coefficient and extracting only a high frequency component by masking a low pass; An adder 14 for adding each DCT coefficient value at the output of the low pass masking unit 13; A first comparator (18) for comparing the output of the adder (14) with a first threshold set therein; A projection unit 16 which receives the discrete cosine transformed DCT coefficient and adds it in a predetermined direction to detect whether there is a strong edge in a specific direction; A dispersion calculation unit (17) for obtaining a dispersion value from the output of the projection unit (16); A second comparator 18 for comparing the dispersion value calculated by the dispersion calculation unit 17 with a second threshold value; The texture region discrimination signal generator 19 which generates a signal for discriminating a texture region when the high frequency component is large without a strong edge in a specific direction from the output of the first comparator 15 and the output of the second comparator 18. It consists of).

Description

Image texture area discrimination device

1 is a block diagram showing an apparatus for determining a texture area of an image according to the present invention.

2 is an example of an NxN discrete cosine transformed DCT coefficient block.

3 is a view for explaining x-axis projection according to the present invention.

4 is a diagram for explaining a y-axis projection according to the present invention.

* Explanation of symbols for main parts of the drawings

11: NxN block portion 12: NcN discrete cosine transform portion

13: low pass masking part 14: adding part

15: comparator 16: projection unit

17: dispersion calculation unit 18: comparison unit

19: texture area discrimination signal generator

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image signal processing apparatus, and more particularly, to a texture region discrimination apparatus for discriminating a texture region among video signals.

In general, there are contours and textures (also known as grains) as indicators for discriminating objects from the human eye. When the texture is taken from the plane or satellite in the air, it is a forest of trees, grass or sea and plains. The same thing is said, even though the visual characteristics of human beings are insensitive to the contours, their statistical characteristics are difficult to analyze.

Therefore, it is desirable to separate and process the texture areas. As a conventional method of recognizing and separating textures, there are methods using statistical analysis, methods using structural analysis, methods using specific point extraction, etc. There is a difficult problem.

Accordingly, an object of the present invention is to provide a texture region discrimination apparatus which can easily determine a texture region by processing a DCT coefficient of an input image in order to separate a texture portion in an image processing apparatus.

In order to achieve the above object, the apparatus of the present invention forms an image data as an NxN block to perform discrete cosine transform to determine a texture area, and receives the discrete cosine transformed DCT coefficient to mask low frequencies by inputting a high frequency mask. Low-pass masking means for extracting only components; An adder for adding each DCT coefficient value at an output of said low pass masking means; A first comparator comparing the output of the adder with a first threshold set therein; A projection unit which receives the discrete cosine transformed DCT coefficients and adds them in a predetermined direction to detect whether there is a strong edge in a specific direction; A dispersion calculation unit for obtaining a dispersion value from the output of the projection unit; A second comparator for comparing the dispersion value calculated by the dispersion calculation unit with a second threshold value; And a texture region discrimination signal generator for generating a signal for discriminating a texture region when the high frequency component is large without a strong edge in a specific direction from the output of the first comparator and the output of the second comparator.

In general, since the texture region has a high spatial frequency, when the discrete cosine transform is performed, many coefficients are distributed in the high frequency part. Accordingly, the present invention determines the texture region using the above properties of the texture region.

That is, among the DCT coefficients of the NxN block obtained by the discrete cosine transform, low frequency components are masked out, the coefficients of the high frequency parts are added, compared with a predetermined first threshold value, and the result obtained by adding the coefficients is larger than the first threshold value. In this case, it is regarded as a texture.

However, when there is a strong edge in a specific direction in the block, the sum of the high frequency components may exceed the first threshold, so that the DCT coefficient obtained by the discrete cosine transform is projected in four directions (that is, in a specific direction). After the dispersion value is obtained, the dispersion value is determined to be a strong edge when the dispersion value is larger than the second threshold.

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

A texture area determining apparatus according to the present invention, as shown in FIG. 1, includes: a block unit 11 for receiving image data to form an NxN (normally 8x8) block; A discrete cosine converter (12) for discrete cosine transforming the blocks formed in the block (11); A low pass masking unit 13 which receives the discrete cosine transformed DCT coefficients and extracts only high frequency components by masking low frequencies; An adder (14) for adding each DCT coefficient value at the output of the low-pass masking section (13); A first comparator 18 for comparing the output of the adder 14 with a first threshold provided therein; A projection unit (16) which receives the discrete cosine transformed DCT coefficients and adds them in a predetermined direction to detect whether there is a strong edge in a specific direction; A dispersion calculation unit (17) for obtaining a dispersion value from the output of the projection unit (16); A second comparator 18 for comparing the dispersion value calculated by the dispersion calculation unit 17 with a second threshold value; A texture region discrimination signal generator 19 for generating a signal for discriminating a texture region when a high frequency component is large without a strong edge in a specific direction from the output of the first comparator 15 and the output of the second comparator 18. It consists of).

Here, the first comparator 15 compares the output of the adder 14 with the first threshold and outputs 1 when the output of the adder 14 is greater than the first threshold, and 0 when the output is smaller. The second comparator 18 compares the output of the dispersion calculation unit 17 with the second threshold and outputs 0 when the output of the dispersion calculation unit 17 is larger than the second threshold. In this case, 1 is output.

The texture region determination signal generator 19 is implemented by an AND gate that logically multiplies the output of the first comparator 15 and the output of the second comparator 18, and outputs a high when determining that the texture region is a texture region. If it is not a texture area, a row is output.

Next, the operation of the apparatus of the present invention configured as described above will be described.

In general, since the texture region has a high spatial frequency, when the discrete cosine transform is performed, many coefficients are distributed in the high frequency part. Therefore, the present invention determines the texture region by using the above properties of the texture region.

That is, in the present invention, low frequency components are masked out of the DCT coefficients of the NxN block obtained by the discrete cosine transform, the coefficients of the high frequency parts are added, compared with a predetermined first threshold value, and the result of adding the coefficients is greater than the threshold value. If it is large, it is regarded as a texture. However, if there is a strong edge in a specific direction in the block, the sum of the high frequency components may exceed the first threshold, so that the DCT coefficients obtained by the discrete cosine transform are projected in four directions (that is, added in a specific direction). After the dispersion value is obtained, if the dispersion value is larger than the second threshold value, it is determined as a strong edge.

As shown in FIG. 2, the DCT coefficient obtained by the discrete cosine transform for an 8x8 DCT block has a low frequency component on the upper left side and a high frequency component toward the lower right side. The low-pass masking unit 13 cuts out the low frequency region of the discrete cosine transform (DCT) coefficients and outputs only high frequency components. For example, as shown in FIG. 2, scan numbers 0 to 9 are all set to 0, and the remaining values are output as they are. Let's do it.

The adder 14 adds the low-pass masked DCT coefficient to an absolute value as follows.

only,

In the first comparator 15, a reference value (first threshold value) of the high frequency region to be determined as a texture is determined, and when the output of the adder 14 is large compared with the output of the adder 14 with the first threshold, 1 is set. If it is small, 0 is output.

On the other hand, the DCT coefficients output by the discrete cosine converter (DCT) 12 are added in a specific direction by the projection unit 16 to obtain one-dimensional data.

That is, FIG. 3 obtains S1, S2, S3, S4, S5, S6, S7, and S8 by adding each column in the x-axis direction at the same 8 × 8 DCT coefficient as that of FIG. Find the variance of the sum. As shown in FIG. 3, as a result of the sum in the x-axis column direction, when a large value appears intensively in S3 and S4, it can be seen that there is a strong edge in this direction.

In addition, Fig. 4 obtains S1, S2, S3, S4, S5, S6, S7, and S8 by adding each row in the y-axis direction from the 8x8 DCT coefficient as shown in Fig. 2, and the dispersion calculation unit 17 calculates these agreements. Find the variance

As shown in Fig. 4, as a result of the sum in the y-axis row direction, when a large value appears intensively in S3 and S4, it can be seen that there is a strong edge in this direction. The second comparator 18 compares the output of the dispersion calculator 17 with a preset second threshold, and determines that there is a strong edge when the output of the dispersion calculator 17 is larger than the second threshold. If it is small, it outputs 1 to determine the texture area.

As described above, when the output of the first comparator 15 and the output of the second comparator 18 are both 1, the texture region discrimination signal generator 19 outputs 1, indicating that the block currently DCT is a texture region. If either one of the first comparator 15 and the second comparator 18 is 0, it indicates that the currently DCT block is not a texture area.

In this way, after removing the low frequency part of DCT coefficient obtained by the discrete cosine transform and calculating the sum of the high frequency parts, it is discriminated as the texture area when the first threshold value is over, but it is removed by the projection to remove the strong edge. It can be determined, and accordingly there is an effect that can be processed by separating the texture area.

Claims (4)

An apparatus for determining a texture region after forming discrete data by converting image data into an NxN block, the apparatus comprising: a low pass masking unit (13) for inputting the discrete cosine transformed DCT coefficients to mask low frequencies to extract only high frequency components; An adder 14 for adding each DCT coefficient value at the output of the low pass masking unit 13; A first comparator (18) for comparing the output of the adder (14) with a first threshold set therein; A projection unit 16 which receives the discrete cosine transformed DCT coefficients and adds each of them in a predetermined direction to detect whether there is a strong edge in a specific direction; A dispersion calculation unit (17) for obtaining a dispersion value from the output of the projection unit (16); A second comparator 18 for comparing the dispersion value calculated by the dispersion calculation unit 17 with a second threshold value; And a texture region discrimination signal generator for generating a signal for discriminating a texture region when the high frequency component is large without a strong edge in a specific direction from the output of the first comparator 15 and the output of the second comparator 18 ( 19) Texture area discrimination apparatus of the image. The method of claim 1, wherein the first comparator 15 compares the output of the adder 14 with the first threshold and outputs 1 when the output of the adder 14 is larger than the first threshold. In the case of the texture area discrimination apparatus of the image, characterized in that to output a zero. The method of claim 1, wherein the second comparator 18 compares the output of the dispersion calculation unit 17 with the second threshold and outputs 0 when the output of the dispersion calculation unit 17 is larger than the second threshold. An image texture area discrimination apparatus, characterized in that for outputting 1 when small. The image texture of claim 1, wherein the texture region determination signal generator 19 is implemented by an AND gate that logically multiplies the output of the first comparator 15 and the output of the second comparator 18. Area discrimination device.