KR100916996B1 - Image processing apparatus and method, lcd overdrive system using the same - Google Patents

Abstract

TECHNICAL FIELD The present invention relates to image processing technology, and more particularly, to image compression technology, and more particularly, to an image processing system requiring high quality image compression, such as an LCD overdrive system. Disclosure of Invention An object of the present invention is to provide an image processing apparatus and method capable of preventing deterioration of image quality even in a composite image having a strong boundary while securing a sufficient compression ratio, and an LCD device overdrive system using the same. Solid boundary images such as text, movie subtitles, and video game images have fundamentally different characteristics from natural images such as high spatial frequency components, low predictability from adjacent pixels, and low repeatability of color components in small areas. The present invention proposes a hybrid architecture that can improve the disadvantages of two digital images by focusing on the difference between the natural image and the composite image. That is, a DCT-based image compression algorithm having excellent compression ratio for natural images is used, while a non-DCT-based image compression algorithm is used for composite images. Also, in the present invention, as a DCT-based image compression algorithm, a compression scheme that analyzes the correlation of adjacent pixels of an image, arranges them into two groups, and gives a group representative value.

Image compression, DCT, composite image, DCT-based compression algorithm, boundary

Description

IMAGE PROCESSING APPARATUS AND METHOD, LCD OVERDRIVE SYSTEM USING THE SAME}

TECHNICAL FIELD The present invention relates to image processing technology, and more particularly, to image compression technology, and more particularly, to an image processing system requiring high quality image compression, such as an LCD overdrive system.

Recently, LCD panels are widely used in monitors, notebooks, TVs, and the like. However, as is well known, motion blur occurs when a fast video is reproduced due to a slow response speed and hold type display characteristics.

Meanwhile, technologies such as backlight scanning, overdriving, and the like have been introduced to overcome the drawbacks of the LCD panel. Overdriving is a technique for correcting the current image by detecting a difference from the previous image. In order to implement the overdriving, a memory capable of storing one frame of image is often required. frame buffer).

However, high resolution image display systems such as high definition (HD) TVs require frame buffers that support more capacity and higher operating speeds. However, as the size of the memory increases, the price becomes very expensive. Therefore, various techniques such as block truncation coding (BTC) -based algorithm, discrete cosine transform (DCT) -based algorithm, and color space conversion (CSC) technique can be used to reduce the frame buffer capacity. Image compression algorithms are used.

Here, the BTC-based image compression algorithm has an advantage that it can be implemented in a relatively simple circuit, but has a disadvantage that it is difficult to restore the original value when there is a low compression ratio and pixels having multiple values in the unit block.

DCT-based image compression algorithms are widely used because of their high compression ratio and good image quality in natural video. However, the DCT-based image compression algorithm has a problem that the variation of the compression ratio is large and the compression ratio is not guaranteed depending on the image, and the blurring phenomenon and ringing are performed in the composite image having a definite boundary such as a character, a movie subtitle, or a video game image. (ringing) phenomenon has a disadvantage that occurs.

The present invention has been proposed to solve the above problems of the prior art, an image processing apparatus and method capable of preventing image quality deterioration even in a composite image having a strong boundary while securing a sufficient compression ratio, and an LCD device overdrive system using the same The purpose is to provide.

According to an aspect of the present invention for solving the above technical problem, the step of aligning the pixel value in the input unit image block; Obtaining a difference between adjacent pixel values using aligned pixel values in the unit image block; Reordering difference values between the adjacent pixel values; Checking whether a two-level image exists based on the maximum difference; As a result of the check, dividing the pixel value in the unit image block into a high group and a row group according to the presence of a two-level image, and obtaining a pixel value deviation in each group; Determining whether or not the synthesized image is compared by comparing the pixel value deviation in the group with a set threshold value; And generating a bit pattern representing a pixel value in the unit image block as 0 and 1 and a representative value of each group as determined as a composite image.

According to another aspect of the present invention, a DCT-based compression transform unit for performing prediction, DCT transform, scan, quantization required for DCT-based compression on the current image frame; A composite image processor for detecting a composite image from a current image frame and compressing the composite image using a non-DCT based compression algorithm; And the compressed data output from the synthesized image processor for the image block including the synthesized image by comparing the sum of the errors of the general image output from the DCT based compression converter and the error of the synthesized image output from the synthesized image processor. There is provided an image processing apparatus having a ratio control unit for applying.

According to another aspect of the present invention, a DCT-based compression transform unit for performing prediction, DCT transform, scan, quantization required for DCT-based compression on the current image frame; A composite image processor for detecting a composite image from a current image frame and compressing the composite image using a non-DCT based compression algorithm; The compressed data output from the synthesized image processor is included in the image block including the synthesized image by comparing the sum of the error of the general image output from the DCT-based compression converter and the synthesized image output from the synthesized image processor. A ratio controller for applying; A VLC encoder for compressing the output of the ratio controller by a variable length coding scheme; A frame buffer control unit for controlling data input / output of the frame buffer; The frame buffer controlled by the frame buffer controller to store one frame data output from the VLC encoder; A VLC decoder for decoding one frame data stored in the frame buffer; A current image restoration unit for restoring the output of the ratio control unit and outputting the current image frame as a current image frame; And a previous image restoration unit for restoring the output of the VLC decoder and outputting the output as the previous image frame.

Solid boundary images such as text, movie subtitles, and video game images have fundamentally different characteristics from natural images such as high spatial frequency components, low predictability from adjacent pixels, and low repeatability of color components in small areas. The present invention proposes a hybrid architecture that can improve the disadvantages of two digital images by focusing on the difference between the natural image and the composite image. That is, a DCT-based image compression algorithm having excellent compression ratio for natural images is used, while a non-DCT-based image compression algorithm is used for composite images. In addition, the present invention proposes a new image compression algorithm optimized for a composite image, which is a non-DCT based image compression algorithm that analyzes the correlation of adjacent pixels of an image, arranges them into two groups, and gives a group representative value.

The present invention has an effect of preventing image quality deterioration such as a blurring phenomenon and a ringing phenomenon while securing a constant compression ratio for a natural image as well as a composite image. On the other hand, since the compression ratio of 1/6 or more can be secured, the capacity of the frame memory can be minimized, thereby providing the possibility of using an embedded frame buffer without using an external frame buffer.

Hereinafter, preferred embodiments of the present invention will be introduced in order to enable those skilled in the art to more easily implement the present invention.

Typical DCT based image compression schemes include prediction, DCT transform, scan, quantize, and variable length coding (VLC) blocks. DCT based compression encoders essentially require a reconstruction block that includes inverse prediction, inverse DCT transform, inverse scan, and inverse quantization blocks to recover the original data for the prediction block. The recovery block can be treated as a decoder. Thus, the reconstruction block in this encoder can also act as a predictive data generator and can act as a separate decoder.

1 is a block diagram showing some components of an LCD overdrive system according to an embodiment of the present invention.

Referring to FIG. 1, in the LCD overdrive system according to the present embodiment, prediction, DCT conversion, scan, and quantization required for DCT-based compression on a current image frame (which is converted into a YCbCr region) DCT-based compression converter 110 for performing quantize, a composite image processor 120 for detecting a synthesized image from a current image frame, and compressing the same through a non-DCT-based compression algorithm, and a DCT-based compression converter ( Comparing the sum of the error (SAD) of the general image output from the 110 and the sum of the error (SAD) of the composite image output from the composite image processing unit 120 is compared to the composite image processing unit 120 for the image block including the composite image A ratio control unit 130 for applying the compressed data outputted from the reference signal), a VLC encoder 140 for compressing the output of the ratio control unit 130 by a variable length coding method, and data input / output of the SDRAM 160. Control SDRAM control unit 150 for controlling, SDRAM 160 for storing one frame data output from the VLC encoder 140 under the control of the SDRAM control unit 150, and for decoding one frame data stored in the SDRAM 160. A VLC decoder 170, a current image reconstructing unit 180 for reconstructing the output of the ratio controller 130 and outputting the current image frame, and a VLC decoder 170 for reconstructing the output of the VLC decoder 170 and outputting the previous image frame. The previous image restoration unit 190 is provided.

2 is a flowchart illustrating a process of the composite image processor 120 of FIG. 1.

Referring to FIG. 2, the composite image processing is performed in parallel with the natural image processing.

First, the pixel values in the input macro block are sorted (210). For example, when a data value of one pixel is 8 bits, the pixel value of the pixel may be 0 to 255. Table 1 below illustrates pixel values of a 4 × 4 unit image block.

240 0 10 7 235 6 8 0 250 One 9 0 250 235 248 252

Referring to Table 1, pixel values are aligned to be 0, 0, 0, 1, 6, 7, 8, 9, 10, 235, 235, 240, 248, 250, 250, and 252.

Subsequently, the difference between adjacent pixel values is calculated using the aligned pixel values in the macroblock (220). The difference value Diff. Is defined as in Equation 1 below.

Diff. = MB (n)-MB (n + 1)

Here, MB (n) is the pixel value of the corresponding pixel, and MB (n + 1) is the pixel value of the adjacent pixel. Referring to Table 1, the difference value (Diff.) Is obtained as 0, 0, 1, 5, 1, 1, 1, 1, 225, 0, 15, 8, 2, 0, 2.

Subsequently, the difference value Diff. Is rearranged (230). Reordering the difference values (Diff.) With reference to Table 1 results in 225, 15, 8, 5, 2, 2, 1, 1, 1, 1, 1, 0, 0, 0.

Subsequently, it is checked whether there is a two-level image divided into a background and a foreground based on the maximum difference value (Diff.) In the macro block (240). Quit. The presence or absence of a two-level image is determined by whether the gap between groups is large or small. Referring to Table 1, since the maximum rearranged value is 225 and the next difference is 15, the gap between groups is very large. It can be seen that a two-level image exists.

On the other hand, if there is a 2-level image, the pixel value is divided into a high (H) group and a low (L) group, and a noise range within the group is obtained using the two groups (250). Referring to Table 1, pixels having pixel values of 240, 235, 250, 250, 235, 248, and 252 are high (H) groups, and 0, 10, 7, 6, 8, 0, 1, 9, Pixels having pixel values such as 0 are classified into a row (L) group.

Next, it is determined whether or not the composite image is compared by comparing the pixel value deviation in the group with the set threshold (260). If the pixel value deviation in the group is larger than the threshold, the process is ended as if it is not a composite image.

As a result of the determination, in the case of the synthesized image, a pattern in which values in the macroblock are represented by H (1) and L (0) and representative values of two groups are calculated (270). In addition, the sum of the errors of the obtained composite image and the input original image is also calculated. For example, Table 2 shows the H / L patterns of the unit image blocks shown in Table 1.

One 0 0 0 One 0 0 0 One 0 0 0 One One One One

In addition, the representative value (average) of the H group is 244, and the representative value of the L group is 5. Therefore, considering the 16-bit bit pattern, the 8-bit representative value of the H / L group, and the RGB pixel, the total bit value is (16 + 8 + 8) × 3 = 96 bits. This is compressed to 1/4 compared to the bit value 384 bits (= 16 (number of pixels) x 8 (data value of 1 pixel) x 3 (RGB)) before compression.

Finally, the sum of the synthesized image determination flag, the pattern, the representative value, and the error generated above is sent to the ratio controller 130 to compare the error with the general image and apply it when the error of the synthesized image is small.

This method can restore compression without noise. However, there is a disadvantage that the compression ratio is slightly lower than that of DCT, and the following process can be additionally applied to compensate for this.

In the region converted from RGB to YCbCr, Y (luminance) and Cb / Cr have a similar pattern. Therefore, if the bit pattern of 0/1 is obtained for Y only, and Cb / Cr has the same pattern as Y, Cb / Cr uses only the representative value of the H / L group except for the bit pattern. As the unit image block increases, the problem that the data value of the 0/1 pattern also increases can be solved in this manner.

In a special case, Y and Cb / Cr have similar patterns but the patterns are reversed. For this purpose, the correlation between Y and Cb / Cr patterns is analyzed without adding data, and the L / H representative values of Cb / Cr are exchanged with each other to reduce unnecessary codes to increase the compression ratio.

This is shown below.

<Code when adding Y, Cb / Cr relationship>

Y pattern + Y (H) + Y (L) + Cb (Rev) + Cb (H) + Cb (L) + Cr (Rev) + Cr (H) + Cr (L)

Here, the Y pattern represents a bit pattern of Y, Y (H) represents a representative value of the H group of Y, Y (L) represents a representative value of the L group of Y, and Cb (Rev) represents whether the reverse pattern of Cb.

<Code for changing H / L value by calculating Y, Cb / Cr relationship>

Y pattern + Y (H) + Y (L) + Cb (H / L) + Cb (L / H) + Cr (H / L) + Cr (L / H)

As described above, the present invention can fundamentally prevent deterioration of image quality such as blurring and ringing while securing a constant compression ratio for both natural and synthetic images.

Lena and Baboon images, which are commonly used for testing in image processing, were used for testing. The master image is also used to test the performance of the algorithm on the composite image. Table 3 below compares the image quality of each compressor method.

3A to 3D are taken from a master image to explain the quality of the synthesized image. FIG. 3A is a master image of an original image, FIG. 3B is a master image during YCbCr420 color space conversion considered to have a compression ratio 1/2, FIG. 3C is a master image of a BTC based algorithm having a compression ratio 1/3, and FIG. 3D is compression ratio 1 / 6 is a master image when applying the algorithm of the present invention.

BTC-based compression algorithms exhibit blocky effects and color leakage near the edges. YCbCr420 has a yellow fade. The algorithm (hardware) proposed in the present invention well represents the left edge of the character by silhouette mode compression (synthetic image processing). In contrast, the shadow region maintains smoothness by DCT mode compression.

4A is an original Lena image, and FIG. 4B is an image reconstructed after compression according to the present invention. The high quality image of FIG. 4B highlights the advantages of the DCT-based compression method in the natural image portion.

Although the technical idea of the present invention has been described in detail according to the above preferred embodiment, it should be noted that the above-described embodiment is for the purpose of description and not of limitation. In addition, those skilled in the art will understand that various embodiments are possible within the scope of the technical idea of the present invention.

1 is a block diagram showing some components of an LCD overdrive system according to an embodiment of the present invention.

2 is a flowchart illustrating a process of the composite image processor 120 of FIG. 1.

3A to 3D are taken from a master image to explain the quality of the synthesized image.

4A is an original Lena image, and FIG. 4B is an image reconstructed after compression according to the present invention.

Claims (12)

Arranging pixel values in an input unit image block; Obtaining a difference between adjacent pixel values using aligned pixel values in the unit image block; Reordering difference values between the adjacent pixel values; Checking whether a two-level image exists based on the maximum difference; As a result of the check, dividing the pixel value in the unit image block into a high group and a row group according to the presence of a two-level image, and obtaining a pixel value deviation in each group; Determining whether or not the synthesized image is compared by comparing the pixel value deviation in the group with a set threshold value; And Generating a bit pattern representing a pixel value in the unit image block as 0 and 1 and a representative value of each group according to the determination result as a composite image Image processing method comprising a. The method of claim 1, After checking whether the two level image exists, And ending the process if the two-level image does not exist as a result of the check. The method according to claim 1 or 2, After determining whether or not the composite image, And ending the process as it is determined that the image is not a synthesized image. The method of claim 1, And the unit image block is converted into a YCbCr region. The method of claim 4, wherein In generating the bit pattern and the representative value of each group, The bit pattern is generated only for Y, and Cb / Cr is regarded as the same bit pattern as Y. The method of claim 5, In generating the bit pattern and the representative value of each group, And analyzing the correlation between Y and Cb / Cr to change the high group representative value and the low group representative value of Cb / Cr. A DCT-based compression converter configured to perform prediction, DCT transform, scan, and quantization for DCT-based compression on a current image frame; A composite image processor for detecting a composite image from a current image frame and compressing the composite image using a non-DCT based compression algorithm; And The compressed data output from the synthesized image processor is included in the image block including the synthesized image by comparing the sum of the error of the general image output from the DCT-based compression converter and the synthesized image output from the synthesized image processor. Rate control section to apply An image processing apparatus having a. The method of claim 7, wherein And the image frame is converted into a YCbCr region. The method of claim 8, The composite image processing unit, Arrange the pixel values in the input unit image block, obtain the difference between adjacent pixel values using the aligned pixel values in the unit image block, rearrange the difference values between the adjacent pixel values, and based on the maximum difference value. It is checked whether a two level image exists, and as a result of the check, the pixel value in the unit image block is divided into a high group and a row group according to the presence of the two level image, the pixel value deviation in each group is obtained, and the pixel value deviation in the group. It is determined whether the synthesized image is compared with the set threshold value, and as a result of the determination, the bit pattern expressing pixel values in the unit image block as 0 and 1 and the representative value of each group are generated. Image processing device. A DCT-based compression converter configured to perform prediction, DCT transform, scan, and quantization for DCT-based compression on a current image frame; A composite image processor for detecting a composite image from a current image frame and compressing the composite image using a non-DCT based compression algorithm; Compressed data output from the composite image processing unit for an image block including a composite image by comparing the sum of the errors of the general image output from the DCT-based compression conversion unit with the sum of the errors of the composite image output from the synthesis image processing unit. A ratio control unit for applying; A VLC encoder for compressing the output of the ratio controller by a variable length coding scheme; A frame buffer control unit for controlling data input / output of the frame buffer; The frame buffer controlled by the frame buffer controller to store one frame data output from the VLC encoder; A VLC decoder for decoding one frame data stored in the frame buffer; A current image restoration unit for restoring the output of the ratio control unit and outputting the current image frame as a current image frame; And And a previous image restoration unit for restoring the output of the VLC decoder and outputting the output as the previous image frame. The method of claim 10, And the image frame is converted into a YCbCr region. The method of claim 11, The composite image processing unit, Arrange the pixel values in the input unit image block, obtain the difference between adjacent pixel values using the aligned pixel values in the unit image block, rearrange the difference values between the adjacent pixel values, and based on the maximum difference value. It is checked whether a two level image exists, and as a result of the check, the pixel value in the unit image block is divided into a high group and a row group according to the presence of the two level image, the pixel value deviation in each group is obtained, and the pixel value deviation in the group. It is determined whether the synthesized image is compared with the set threshold value, and as a result of the determination, the bit pattern expressing pixel values in the unit image block as 0 and 1 and the representative value of each group are generated. LCD overdrive system.

Images