KR100598367B1 - Filtering method for block boundary region - Google Patents

Abstract

The method for filtering a block boundary region of a video decoding system according to the present invention is a method for filtering a block boundary region of an image when decoding an input image, the method including a difference between pixels adjacent to the block boundary region. Comparing the one function value and the threshold value; Based on the comparison result, the method includes filtering two pixels in each of two directions based on a block boundary of the image.

According to the present invention, in decoding the received image, there is an advantage that can provide a higher quality video by applying a post-processing method adaptive to the characteristics of the image.

Description

Filtering method for block boundary region}

1 is a diagram for explaining a smoothing process as one of a general deblocking method.

2 to 4 are flowcharts illustrating each embodiment of a process of processing an image by an image post-processing method of a video decoding system according to the present invention.

The present invention relates to a video decoding system, and more particularly, to a block boundary region filtering method in a video decoding system including a post-processing engine for improving image quality in a decoder which is a receiving means in a wired / wireless communication system supporting video.

Recently, many studies have been reported to provide high-quality moving images while providing moving image services to PDAs, hand-held PCs, and even mobile communication terminals. H.263, MPEG1 / 2, MPEG-4 and the like are used as moving picture standards used for moving picture services using wired and wireless communication. They differ slightly in compression, but they have a lot of similarities, so the low quality phenomenon that occurs when the network environment is poor is similar.

The degradation in image quality can be thought of as being caused by the fact that the network environment is deteriorated so that an image must be coded with a small amount of data and an error occurs due to the loss of information during transmission. The former is a more serious problem mainly in video communication, such as video that needs to be transmitted in real time, but even if it is not a real-time service, a large amount of data is required to transmit a large amount of data for good image quality.

The blocking phenomenon is mentioned as the case where the former is the cause of such quality deterioration. This phenomenon occurs in the compression method in order to compress to a low volume, and a pattern like a checkerboard is generated in the image. In order to solve the blocking phenomenon, a lot of researches have been made, and studies for solving problems occurring in still images and studies for solving problems occurring in moving images are being actively conducted.

However, if you try to completely eliminate the blocking phenomenon, there is a problem that it may have an adverse effect by blurring an important edge area, and even if the resolution is solved to some extent, the degradation of the image quality caused by the low capacity compression is basically prevented. There is no problem.

The present invention provides a block boundary region filtering method of a video decoding system capable of providing a higher quality video by applying an post-processing method adaptive to a characteristic of an image in decoding a received image. There is a purpose.

In order to achieve the above object, a block boundary region filtering method of a video decoding system according to the present invention is a method of filtering a block boundary region of an image when decoding an input image, the pixel adjacent to the block boundary region. Comparing a threshold with a first function value that includes a difference between them; Based on the comparison result, the method includes filtering two pixels in each of two directions based on a block boundary of the image.

According to the present invention, in decoding the received image, there is an advantage that can provide a higher quality video by applying a post-processing method adaptive to the characteristics of the image.

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

First, a general video compression method will be briefly described for better understanding of the present invention, and the post-processing process applied to the video post-processing method of the video decoding system according to the present invention will be described sequentially.

<Common Compression Method of Video>

In general, H.263, MPEG1 / 2, and 4, which are standard in wired and wireless video, compress video using a similar method. The biggest feature is that the video is first compressed using motion estimation. . All frames constituting the video may be classified into I frames, P frames, and B frames.

Here, the I frame is a case where all parts of the frame are independently compressed without referring to a previous or subsequent frame. After performing DCT transformation of an image by 8 * 8 units, the transformed value is quantized using a specific quantization coefficient. After that, the resulting values are coded using Variable Length Coding (VLC). In this case, if the result values after DCT conversion are large on the basis of variable length coding (VLC) characteristics, the data may be coded with a small data amount. Therefore, if a specific quantization coefficient is set large, the values smaller than the quantization coefficient are all zero and the probability of the same value appearing in succession can be coded with a small amount of data, while the image quality is so large. Therefore, it will fall. The quantization value is given in macroblock units, which are basic units divided by 16 * 16 pixels.

In addition, a P frame compresses an image by referring to a previous frame, and extracts motion information by searching for the most similar macro block in the previous frame in the above-described macro block unit. After extracting the motion information, the difference value between the macroblock in the previous previous frame and the macroblock to be coded in the current frame is obtained, and the quantization and VLC (Variable) are performed through DCT transformation in 8 * 8 units as in the I frame. Length Coding) is compressed along with the motion information. In the case of a P frame, the amount of coded data is much smaller than that of an I frame because the probability that the values are smaller than the pixel values of a general image is large enough to code a difference from similar macro blocks in the previous frame. However, the picture quality is not as good as the I frame. In general, the first frame is coded as an I frame and the remaining frames are coded as P frames, but they are also coded periodically as I frames to ensure image quality. And, the B frame is similar to the P frame except that it refers to the next frame as well as the previous frame.

<Image quality deterioration due to quantization>

1) loss of color

In the above-described quantization compression method, since information is lost by the quantization coefficient, accurate color information cannot be represented. The larger the quantization coefficient is, the more the original color information is distorted. When the image is played after the decoder decompresses the distorted image, the image quality is degraded. Distortion of the image appears as a phenomenon in which the original color is different or the whole image is blurred.

2) blocking artifact

As described above, the quantization coefficient is determined for each macroblock, and the unit in which the quantization is performed is an 8 * 8 DCT block unit. Therefore, when quantization is performed for each DCT block, a phenomenon occurs that the vicinity of a boundary between adjacent DCT blocks is shifted. This occurs because the color is distorted due to quantization in each block, which is called a blocking phenomenon. Therefore, if the quantization coefficient is large, the image may be distorted.

< Image quality improvement through post processing>

1) Optional Deblocking

Low quantization coefficients may be used to prevent the degradation of the image quality described above. However, if the network environment is bad, the amount of compressed data must be small, and thus high quantization coefficients cannot be used. Therefore, a disadvantage arises in that the above described image quality deterioration cannot be avoided. Instead, after the video is received, the decoder can compensate for the degraded picture quality through post-processing. A typical post-processing process currently used is to apply a deblocking method that compensates for a blocking phenomenon. The deblocking method is to smooth the DCT block boundary region. In general, a deblocking method is applied by smoothing two pixels in both directions around the boundary region. In this case, referring to FIG. 1, Bx, Cx, By, and Cy will perform smoothing more severely than Ax, Dx, Ay, and Dy.

Here, smoothing is a method in which the difference between two pixels is less than the original difference, and the most serious smoothing method is a method of modifying two pixels to an average value of two pixels. However, when the boundary area is smoothed in a batch, there is a risk that the boundary area may be smoothed to the original edge. Therefore, it is necessary to determine whether the boundary region of the DCT block is an original edge region or an error that appears in the process of quantizing the DCT block.

In general, the error that occurs during the quantization of the DCT block does not exceed twice the quantization coefficient value. Therefore, when the difference between two pixels adjacent to the boundary region exceeds the size of the quantization coefficient value twice, it may be determined that the region is the original edge, otherwise the error due to quantization may be determined. In the present invention, a case of using a Robert edge filter as a method of determining an edge through a difference between adjacent pixels will be described.

Rebert edge filter

if (abs (D1) + abs (D2)> th1) then edge

else blocking

Therefore, the deblocking edge filter is applied only when it is determined that the error is not an edge by the above equation. The deblocking method described above is illustrated in FIG. 2. First, when an image comes in, it is determined whether it is a DCT block boundary region, and if it is a DCT block boundary region, it is determined whether the boundary region is an edge region. If it is not an edge area, a deblocking edge filter is performed through a smoothing process.

2) Edge Hardening Method

As mentioned earlier, the use of high quantization coefficients causes the entire image to be blurred. To solve this problem, an image enhancement filter may be used. An image enhancement filter adjusts a value so that a difference is greater than a difference between two original adjacent pixels, and may be regarded as an opposite concept of smoothing. The following method is mentioned as an example of a reinforcement filter.

Using the Robert Filter results described above,

A ₀₀ = A ₀₀ + D1 * a

A ₁₁ = A ₁₁ -D1 * a

A ₁₀ = A ₁₀ + D2 * a

A ₀₁ = A ₀₁ -D2 * a

The difference between the two pixels is increased to enhance the image at the edge part.

However, when the images are strengthened collectively for all the images, the following two problems may occur.

-Blocking phenomenon occurring at the DCT block boundary can be enhanced.

Noise that is not edges can be enhanced.

In order to prevent the above two problems, the present invention enhances an image, but performs enhancement only when it is determined that the edge is not a DCT block boundary. Therefore, the enhancement method applied in the present invention may be referred to as edge enhancement rather than image enhancement. Edge determination may be determined when the result of the above-described Robert filter is more than a predetermined threshold. The edge enhancement method described above is described in FIG. First, when an image is input, it is determined whether an edge is detected when the image is not a DCT block. If it is an edge, a reinforcement filter is performed.

In this case, in the present invention, in reinforcing the corresponding region, the brightness difference between the two pixels that are diagonally connected to the four pixels input to the Robert edge filter is further adjusted. In adjusting the brightness difference between two diagonally connected pixels, the adjustment width is proportional to the original brightness difference between the two pixels.

4 illustrates a case in which the deblocking method and the edge enhancement method described above are performed together. First, when an image is input, it is determined whether it is a DCT boundary region. If it is a DCT boundary region, it is determined whether the region is an edge region, and if it is not an edge, smoothing processing is performed using a deblocking edge filter only for pixels around the boundary region. do. If it is not the DCT boundary, it is determined whether the edge is an edge, and if it is an edge, the enhancement is performed using an edge enhancement filter.

As described above, in the block boundary region filtering method of the video coding system according to the present invention, in decoding a received image, a higher quality video may be generated by applying a post-processing method adaptive to the characteristics of the image. There is an advantage that can be provided.

In addition, the block boundary region filtering method of the video coding system according to the present invention can be usefully used to restore low quality data, which is generally generated in a low network environment, such as a mobile communication system, and the restoration is performed at the decoder. There is an advantage that can be applied irrespective of errors occurring during transmission.

Claims (9)

A method of filtering block boundary pixels according to a difference value of pixels adjacent to a block boundary region of an image when decoding an input image, Comparing a first function value including a difference between pixels adjacent to a block boundary region with a threshold for determining whether to filter according to the first function value; And deblocking filtering two pixels, each of which is bidirectional, based on a block boundary of the image through a smoothing process based on the comparison result. The method of claim 1, And filtering the block boundary region when the first function value is less than or equal to a threshold value as a result of the comparing step. The method of claim 2, And if the first function value is greater than the threshold as a result of the comparing step, filtering the block boundary region is not performed. 4. The method of claim 3, wherein the block boundary area is determined as an edge when the first function value is greater than a threshold as a result of the comparing step. delete A method of filtering pixels of a block boundary region of an image when decoding an input image, Determining whether a block boundary region requires filtering through a smoothing process in consideration of a quantization coefficient; And performing filtering by smoothing two pixels in each of two directions around the block boundary of the image according to the determination result. 7. The method of claim 6, wherein the smoothing process adjusts the pixel value so that the difference between adjacent pixels is less. 8. The method of claim 7, wherein the smoothing process comprises an averaging method using pixels of the block boundary area. 8. The method of claim 7, wherein the performing of the filtering comprises averaging arithmetic for one pixel each around the block boundary.

Images