KR102250252B1 - Method and apparatus of in-loop filtering by adaptive band offset bands, and appparatus for decoding and encoding by adaptive band offset bands - Google Patents

Abstract

The present invention discloses an in-loop filtering method and apparatus using an adaptive band offset period, and an encoding and decoding apparatus using an adaptive band offset period. The in-loop filtering method of the present invention includes a deblocking filtering step of removing block distortion of a reconstructed picture; And applying an adaptive sample offset (SAO) to the deblocking filtered image, wherein the applying of the adaptive sample offset is adaptive to a current block to which the sample offset is to be applied. And applying a band offset. According to the present invention, the possibility of selecting an appropriate offset according to the characteristics of a pixel is improved, an appropriate offset is selected according to the characteristics of each pixel, and a pixel value is more appropriately corrected, thereby improving video encoding efficiency.

Description

An in-loop filtering method and apparatus using an adaptive band offset section, and an encoding and decoding device using an adaptive band offset section (METHOD AND APPARATUS OF IN-LOOP FILTERING BY ADAPTIVE BAND OFFSET BANDS, AND APPPARATUS FOR DECODING AND ENCODING BY ADAPTIVE) BAND OFFSET BANDS}

The present invention relates to encoding and decoding of video data, and more particularly, to a method and apparatus for adaptively applying an offset for each pixel to a reconstructed image.

When an image is encoded at a low bit rate, the reconstructed image is stored in the frame memory while block artifacts have occurred, and the image including the block artifacts is stored in the motion compensation process of the next picture. Deterioration of image quality may propagate when coding is performed with reference. To solve this problem, in the H.264/AVC standard, block distortion (block distortion) before storing the reconstructed image in the frame memory without knowing the original image It uses adaptive deblocking filtering to remove block artifacts.

The deblocking filtering is a passive filter for removing only block distortion without knowing the original image, and a difference value between the original image and the reconstructed image ( offset) cannot be compensated.

[Prior technical literature]

(Patent Document 1) Korean Laid-Open Patent No. 10-2011-0083369 ("Video encoding method and device thereof using deblocking filtering, and video decoding method and device using deblocking filtering", Samsung Electronics Co., Ltd., 2011.07.20 open)

As a technique for compensating for an offset between an original image and a reconstructed image, there is an adaptive sample offset (SAO) method. The adaptive sample offset method is a method of obtaining a difference value between an original image and a reconstructed image, obtaining an offset based on the difference value between the original image and the reconstructed image, and adding it to the reconstructed image in order to correct errors in decimal arithmetic that may occur in the process of obtaining the reconstructed image.

The offset is applied separately for each pixel. The offset mode is largely divided into a band offset and an edge offset. There are two modes for the band offset by dividing the entire level of a pixel into 32 parts, dividing it into 16 central bands and 16 side bands. There are 4 modes of edge offset, 0 degrees, 45 degrees, 90 degrees, and 135 degrees, depending on the direction. The offset is obtained for each mode separately, and the mode that consumes the least cost is selected on a rate-distortion basis.

As for the offset mode, each frame included in the image is divided into blocks of various sizes using a quad-tree-type segmentation method, and a different offset mode is determined for each block. Among the two modes, in the case of the band offset method, a criterion for dividing into a center section and a side section is uniformly determined. The band offset section selected according to the above criteria does not effectively reflect the characteristics of the corresponding pixel and block.

In the present invention for solving the above-described problem, in performing an adaptive sample offset (SAO) for a reconstructed image, it is applicable in determining the band offset period among the two methods of SAO. We propose a method of adaptively classifying a center section and a side section by considering the characteristics of pixels and blocks.

An object of the present invention for solving the above-described problem is to provide an in-loop filtering method for adaptively differently applying a band offset type and a band offset value of SAO according to a current mode in performing a band offset method of SAO processing. will be.

Another object of the present invention is to provide an apparatus for encoding and decoding an image that adaptively divides a center section and a side section based on a pixel value distribution of a reference range associated with a corresponding block in performing SAO processing using a band band offset method. .

An in-loop filtering method using an adaptive band offset interval for achieving the object of the present invention described above includes a deblocking filtering step of removing block distortion of a reconstructed picture; And applying an adaptive sample offset (SAO) to the deblocking filtered image, wherein the applying of the sample offset includes an adaptive band according to a current block to which the sample offset is to be applied. It involves applying an offset.

In the present invention, the adaptive band offset is introduced as a concept to enable the SAO procedure more suitable for the current block by determining a band offset section or type differently according to the current block to which the SAO procedure is to be applied.

The adaptive band offset procedure of the present invention is included in the adaptive sample offset step, and the adaptive band offset procedure adaptively determines a band offset section in consideration of a pixel value distribution of pixels belonging to a reference range related to the current block. Determining a sample offset type of the current block in consideration of the adaptively determined band offset period; And summing sample offsets for pixels belonging to the current block in consideration of the determined band offset period.

In the present invention, the reference range is an area specified by pixels belonging to an area adjacent to the current block according to a predetermined criterion, or an area specified by pixels belonging to a block having similarity to the current block according to a predetermined criterion. Can be In particular, an area specified by pixels belonging to an area adjacent to the current block according to a predetermined criterion may include at least one of an upper block, a left block, and an upper right block of the current block. An area specified by pixels belonging to an area adjacent to the current block according to a predetermined criterion may be set as a set of pixels surrounding the current block with a predetermined thickness.

In the present invention, the band offset period includes central bands and side bands, and adaptively determining the band offset period is within the reference range among all bands in which pixel values may exist. It may include determining a band offset interval according to the current block in consideration of the band to which the pixels belong most belong.

In the present invention, the center section may be determined adaptively in consideration of the pixel value distribution of pixels belonging to the reference range so that the band to which the pixels belonging to the reference range most belong is the center of the center section. It is possible. It is also preferable to determine the size of the center section so that the number of bands in which pixel values of pixels belonging to the reference range are distributed and the size of the center section have a negative correlation within a predetermined reference range.

In addition, central bands may be determined so that a pixel value distributed with the largest number of pixels in the reference range becomes the center, but the central band may be determined to be 1/2 of the total number of bands. In the adaptive sample offset processing step, when the reference range does not exist, the central bands are determined so that the center of the entire section becomes the center of the central bands, but the central bands are You can decide to occupy half of the entire range.

In addition, in the present invention, the band offset period is adapted to the central bands, the first side bands, and the second side bands in consideration of the pixel value distribution of pixels belonging to the reference range. It can be determined as an enemy.

A decoding apparatus using an adaptive band offset interval for achieving another object of the present invention includes: a deblocking filter for removing block distortion of a reconstructed image; And an SAO processing unit that processes the deblocking filtered image adaptive sample offset (SAO), wherein the SAO processing unit includes a band of the current block based on a pixel value distribution characteristic of a reference range associated with the current block. It may be characterized in that the offset section is determined. Here, the reference range may be a set of pixels of a block adjacent to the current block, and the reference range may be a set of pixels of an upper block and a left block of the current block.

In addition, the reference range may be a set of pixels surrounding the current block with a preset thickness, and the reference range may be a set of pixels of the current block. Here, the SAO processing unit may be characterized in that the section in which the pixel values of the reference range are distributed is determined as central bands, and the SAO processing unit is the pixel value in which the largest number of pixels of the reference range is distributed is a center. Central bands are determined to be, but the central bands may be characterized in that they occupy 1/2 of the entire range. In addition, when the reference range does not exist, the SAO processing unit determines the central bands so that the center of the entire section becomes the center of the central bands, but the central bands are the entire range. It can be characterized as occupying 1/2 of.

An encoding apparatus using an adaptive band offset interval to achieve another object of the present invention includes: a deblocking filter for removing block distortion of a reconstructed image; And an SAO processing unit that processes the deblocking filtered image adaptive sample offset (SAO), wherein the SAO processing unit includes a band of the current block based on a pixel value distribution characteristic of a reference range associated with the current block. It may be characterized in that the offset section is determined. Here, the reference range may be a set of pixels of a block adjacent to the current block, and the reference range may be a set of pixels of an upper block and a left block of the current block. In addition, the reference range may be a set of pixels surrounding the current block with a preset thickness, and the reference range may be a set of pixels of the current block.

Here, the SAO processing unit may be characterized in that the section in which the pixel values of the reference range are distributed is determined as central bands, and the SAO processing unit is the pixel value in which the largest number of pixels of the reference range is distributed is a center. Central bands are determined to be, but the central bands may be characterized in that they occupy 1/2 of the entire range. In addition, when the reference range does not exist, the SAO processing unit determines the central bands so that the center of the entire section becomes the center of the central bands, but the central bands are the entire range. It can be characterized as occupying 1/2 of.

According to the above-described in-loop filtering method and image decoding apparatus, in band offset correction, by adaptively selecting the classification criterion for the center section and the side section in consideration of the characteristic of the pixel value distribution of the neighboring block, it is appropriate according to the characteristic of the pixel. Improves the possibility of selecting an offset. Accordingly, an appropriate offset is selected according to the characteristics of each pixel, and the pixel value is more appropriately corrected, thereby providing an effect of improving video encoding efficiency and reconstructed image quality.

In addition, since two sections are determined for each block by using the reference range associated with the current block without using additional information, there is no additional bit rate consumption, so the compression efficiency can be increased, and the image quality is improved due to the improvement of the offset efficiency, so subjective picture quality is improved. You can also expect.

1 is a flowchart of an in-loop filtering method using an adaptive band offset interval according to an embodiment of the present invention.
2 is a detailed flowchart of the adaptive sample offset processing step of FIG. 1.
3 is a conceptual diagram when a neighboring block is set as a reference range.
4 is a conceptual diagram illustrating a case in which pixels surrounding a current block are set as a reference range.
5 is a conceptual diagram illustrating a case of setting pixels of a current block as a reference range.
6 is a view in which a center section and a side section are divided according to a conventional standard.
7 shows an example in which the pixel value distribution range of the reference range is set as the center section.
8 is an example of a case in which a center section is set centering on the most distributed pixel value of a reference range pixel.
9 is an example of setting a center section, a first side section, and a second side section according to an embodiment of the present invention.
10 is a block diagram illustrating a configuration of an image decoding apparatus using an adaptive band offset interval according to an embodiment of the present invention.
11 is a block diagram illustrating a configuration of an image encoding apparatus using an adaptive band offset interval according to an embodiment of the present invention.
12 is a block diagram showing the configuration of the in-loop filter of FIGS. 10 to 11.
13 is a detailed block diagram of an adaptive sample offset processing unit in FIG. 12.

In the present invention, various modifications may be made and various embodiments may be provided, and specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to a specific embodiment, it should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the present invention.

In describing the embodiments of the present specification, when it is determined that a detailed description of a related known configuration or function may obscure the subject matter of the present specification, a detailed description thereof will be omitted. Terms such as first and second may be used to describe various elements, but the elements should not be limited by the terms. The above terms are used only for the purpose of distinguishing one component from another component. For example, without departing from the scope of the present invention, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element. The term and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

When a component is referred to as being "connected" or "connected" to another component, it is understood that it may be directly connected or connected to the other component, but other components may exist in the middle. It should be. On the other hand, when a component is referred to as being "directly connected" or "directly connected" to another component, it should be understood that there is no other component in the middle.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present application, terms such as "comprise" or "have" are intended to designate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, but one or more other features. It is to be understood that the presence or addition of elements or numbers, steps, actions, components, parts, or combinations thereof does not preclude in advance.

In addition, the constituent units shown in the embodiments of the present invention are shown independently to represent different characteristic functions, and does not mean that each constituent unit is formed of separate hardware or a single software constituent unit. That is, each constituent part is listed and included as a respective constituent part for convenience of explanation, and at least two of the constituent parts are combined to form one constituent part, or one constituent part is divided into a plurality of constituent parts to perform a function. Integrated embodiments and separate embodiments of the components are also included in the scope of the present invention as long as they do not depart from the essence of the present invention.

Unless otherwise defined, all terms used herein including technical or scientific terms have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms as defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in this application. Does not.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings. In describing the present invention, in order to facilitate an overall understanding, the same reference numerals are used for the same elements in the drawings, and duplicate descriptions for the same elements are omitted.

In order to apply the adaptive band offset interval, it is determined whether or not a neighboring block can be referenced to calculate a statistical characteristic of a pixel value of the neighboring block that can be referred to, and based on this, a section identification criterion is determined. According to the determined section discrimination criterion, the center section and the side section are divided, and offset correction is performed for each mode of each block. In the present invention, since information on a block that has already been decoded before the current block is used, additional information is not required.

In-loop filtering method

1 is a flowchart of an in-loop filtering method using an adaptive band offset interval according to an embodiment of the present invention. The in-loop filtering method shown in FIG. 1 includes the following steps performed in time series in the in-loop filter shown in FIG. 12.

As shown in FIG. 1, deblocking filtering to remove block distortion of the reconstructed image is first performed (S110). When an image is encoded at a low bit rate, the decoded image is stored in the frame memory while block distortion occurs, and the image including the block distortion is referenced and coded in the motion compensation processing of the next picture. To do this, adaptive deblocking filtering is used to remove block distortion before storing the reconstructed image in the frame memory.

Thereafter, the deblocking filtered image is processed with a Sample Adaptive Offset (SAO) (S120). In the present embodiment, the adaptive sample offset processing step S120 is characterized in that the sample offset is adaptively applied according to the current block to which the adaptive sample offset is to be applied. Existing adaptive sample offset divides the pixel values of pixels included in the current block according to a band offset category prepared in advance, and determines whether the pixel value is a central band or side bands to offset the band. It was a concept of adaptively determining the type of each pixel.

The adaptive sample offset according to an embodiment of the present invention is characterized in that the center section for determining the type of the band offset is adaptively determined according to the current block since the center section may be defined differently according to the current block. have. When the boundary region is determined, the center section and the side section are also determined. In this embodiment, the boundary region that distinguishes the section is a band offset of the current block in consideration of the pixel value distribution of pixels located in the reference range associated with the current block. There is a feature in determining the interval.

As described above, the adaptive sample offset processing step (S120) is to compensate for a DC offset for pixels subjected to a deblocking filter. Here, the DC offset refers to an average difference value between pixels subjected to a deblocking filter after transformation and quantization and original pixels, and may be regarded as an offset value due to transformation and quantization. After performing the deblocking filter, the current block (coding unit) may determine whether to selectively turn on/off the sample adaptive offset (SAO) in units of partitioned blocks of the current block.

2 is a detailed flowchart of the adaptive sample offset processing step of FIG. 1. Referring to FIG. 2, the adaptive sample offset processing step S120 will be described in more detail.

As shown in FIG. 2, the adaptive sample offset processing step (S120) is steps performed in a time series by the adaptive sample offset processing unit (1120 in FIG. 12), that is, the frame division step (S210), whether neighboring blocks can be referenced. A verification step S220, a pixel value distribution characteristic extraction step S230, a band offset section determination step S240, and an offset mode determination step S250 may be included. It does not require additional information compared to the existing method.

The adaptive sample offset processing unit 1120 performs block division on a frame or block to be subjected to adaptive sample offset processing from the deblocking filter-processed image. In this step (S210), the signal input to the frame division unit is a frame signal that is decoded after compression. For example, the adaptive sample offset processing unit may divide the entire frame into a quad-tree shape, and may divide up to four steps in detail. The size of the divided block and the offset mode per block are determined. The size of the block and the offset mode per block are determined according to the rate-distortion cost. In order to obtain an appropriate partitioning method, the rate-distortion cost is calculated for all partitioning steps and finally determined as the lowest cost partitioning type.

The adaptive sample offset processing unit 1120 searches for a reference range that can be referred to among neighboring blocks. Since the size of the current block may vary, the sizes of neighboring blocks also exist, and pixels within a certain range may be used instead of a block shape. If there is no referenceable block of the current block, the next step of extracting the pixel value distribution characteristic (S230) may not be performed, and the process may proceed to the step (S250) of determining a band offset section.

The adaptive sample offset processing unit 1120 extracts a pixel value distribution characteristic (S230). In this step, the adaptive sample offset processor determines a discrimination criterion for distinguishing the center section and the side section from the band offset by using the distribution of the pixel values of the reference range. The discrimination criterion may be the average of the pixel values of the reference range or the section on the band category to which the average of the pixel values belongs, and when the pixels of the reference range are displayed on the quantized band category, the section on the band category showing the highest frequency is used as the discrimination criterion. You can decide. In addition, the length of the distribution range is calculated through the histogram of the distribution of pixel values, and if the length of the distribution range is short, the size of the band offset section (number of bands) is adjusted, based on the location where the probability of the existence of pixels in the distribution is high. The narrow band offset section can be set with.

In the band offset section determining step S240, the entire band for the pixel value is divided into two sections. In the band offset, a center section and a side section are defined. Among them, a method of determining the center section may be determined based on a section in which pixel values of a reference range associated with the current block are distributed. And the side section is determined as the rest of the section excluding the determined central section. If there is no pixel value distribution characteristic information because the neighboring blocks cannot be referenced, the half of the center of the entire section where pixel values are possible is determined as the center section, and the rest of the other two sections can be determined as side sections. have.

In the offset mode determining step (S250), it is determined through a rate-distortion technique which section is more efficient for the current block based on the previously determined center section and the side section. In the offset mode determination step (S250), not only the band offset mode using the center section and the side section, but also the edge offset method, which is an existing method, can be evaluated. Determine the mode and size of the.

The information generated as a result of the offset mode determination step S250 includes, for example, offset information, frame division information, and mode information, and the information is included in the slide header and can be used by the decoder. Since the adaptive band offset period generated by the present invention is determined using surrounding information without the need to add additional information to the bit stream, there is no bit rate loss.

In the adaptive sample offset processing step S120, a reference range may be set in various ways in order to adaptively determine a band offset period. The embodiment described below assumes an RGB image, which is a general format of an image acquisition device, as an input signal, and is described based on the HEVC image compression standard including adaptive sample offset (SAO) technology, but the format or size of the input image is It is not limited to the description of this embodiment.

In the conventional band offset performed by the adaptive sample offset (SAO), the entire band was divided into a fixed center section and a side section and used. This conventional method was not efficient because it did not consider adaptability to the corresponding pixel. However, in the present invention, since the center section of the band offset is determined through the reference range pixel value distribution characteristic associated with the current block, it is possible to correct the offset more efficiently for the corresponding block.

In order to determine an adaptive band interval, it is necessary to obtain a pixel value distribution characteristic of a reference range associated with a current block. Before describing an embodiment of a method of setting a reference range for this, the concept will be described in more detail.

In this embodiment, the reference range is an area specified by pixels belonging to an area adjacent to the current block according to a predetermined criterion, or an area specified by pixels belonging to a block having similarity to the current block according to a predetermined criterion. Used in meaning. Here, the region adjacent to the current block may include one or more of a left block (or region), an upper block, and an upper right block of the current block. A block having similarity according to a predetermined criterion is a block showing a distribution of pixel values similar to the current block, and is similar in consideration of the similarity value quantified by a function to find the most similar block such as cosine distance and Euclidean distance. It may mean a block determined to be.

The size of the current block for SAO processing varies from 8*8 to 64*64. If the size of the current block is small, the distribution of pixel values may be biased. To determine the sample offset mode or offset value for the current block more appropriately, consider the pixel values of pixels belonging to the reference range related to the current block. It is desirable to do it. It is a feature of the present invention to predetermine an adjacent region having similarity to the current block as a reference region, and to determine an offset band interval of the current block in consideration of a pixel value distribution of the reference region.

3 is a conceptual diagram when a neighboring block is set as a reference range. As shown in FIG. 3, there is a method of analyzing the distribution of pixel values of the corresponding block 310 by referring to neighboring left and upper blocks. This method considers the encoding order, and can be used if decoding has not yet been performed below and to the right of the current block. That is, the reference range 320 may be a set of pixels of an upper block and a left block of the current block 310.

Meanwhile, the referenced neighboring block is not limited to the left block and the upper block. In some cases, all blocks adjacent to the current block may be referred to, or a left block, an upper left block, an upper block, an upper right block, and a lower left block may be referred to. That is, the reference range may be a set of pixels of a neighboring block of the current block.

4 is a conceptual diagram illustrating a case in which pixels surrounding a current block are set as a reference range.

As shown in FIG. 4, pixel value distribution characteristics may be obtained by using pixels 420 having a predetermined thickness surrounding the current block 410. If the step of applying the adaptive sample offset is performed after all of the frames have been decoded, since pixel values exist in all parts surrounding the corresponding block, there is a method of using all of them. That is, the reference range 420 may be characterized in that it is a set of pixels surrounding the current block with a preset thickness.

5 is a conceptual diagram illustrating a case of setting pixels of a current block as a reference range.

As shown in FIG. 5, a pixel value distribution may be obtained based on the pixel value information of the current block 510. Since the pixel value of the current block has already been decoded, the pixel value can also be used to obtain the distribution. That is, the reference range 520 may be a set of pixels of the current block.

When a distribution of pixel values of pixels belonging to the reference range is obtained, the band offset section determining unit 1122 of the adaptive sample offset processing unit determines a center section based on this.

6 is a view in which a center section and a side section are divided according to a conventional standard. As shown in FIG. 6, the conventional criterion does not refer to the reference range associated with the current block, and a portion corresponding to half of the center of the entire section in which pixel values are possible is determined as the center section 610, and Both of the remaining portions may be determined as side sections 620. Embodiments presented to determine a central section according to an embodiment of the present invention are as shown in FIGS. 7 to 9.

7 is a diagram illustrating a case in which a pixel value distribution range of a reference range is set as a center section. As shown in FIG. 7, the band offset section determiner 1122 may determine a section in which pixel values of the reference range are distributed as central bands. When the center section is determined, both ends of the center section become a boundary for dividing the center section and the side section.

According to another embodiment related to this, the band offset section determining unit 1122 is a section in which pixel values in a reference range are distributed when a section in which pixel values are distributed is distributed only in a section less than 1/2 of the entire range of the pixel values. Can also be determined as the central section. Also, the band offset section determiner 1122 may determine a region in which the pixel values of the reference range are distributed above a predetermined reference value, that is, a main distribution region of the pixel values, as the center section. Here, the predetermined reference value may be, for example, a ratio of the number of pixels present in the corresponding region to the number of all pixels belonging to the reference range.

FIG. 8 is a diagram illustrating a case in which a center section is set around a maximum distributed pixel value of a reference range pixel. As illustrated in FIG. 8, the band offset section determiner 1122 may determine central bands such that a pixel value distributed with the largest number of pixels in the reference range is a center. In particular, the band offset section determiner may select a band having the largest number of pixels based on the band category as the center of the center section, and determine bands located at a predetermined distance from the center as the center section.

The band offset section determiner may determine that the central bands occupy 1/2 of the entire range. If the section in which the pixel values of the reference range are distributed is greater than 1/2 of the total range that the pixel values can have, the range that becomes 1/2 of the entire range may be determined based on the pixel values in which the largest number exists. have.

Here, the band offset section determining unit determines the central bands so that the center of the entire section becomes the center of the central bands when the reference range does not exist, but the central bands are You can decide to occupy half of the entire range.

9 shows an example in which the number of band offset sections is adjusted when the distribution range of the reference range pixels is very short and the existence probability value is large in a narrow band section. As shown in FIG. 9, the band offset period may include a center period BO_0, a first side period BO_1, and a second side period BO_2.

In particular, the band offset section determiner may adjust the number of band offset sections to set a narrow band offset section based on the highest probability of existence in the distribution. The band offset section determiner determines central bands so that the pixel value distributed with the largest number of pixels in the reference range is the center, but increases the number of band offset sections because a large probability value is distributed in a narrow range around the center section. So that a narrow band offset section can be set. Here, the number of band offset intervals is inversely proportional to the distribution range of the reference range pixels, and the width of the band offset interval is inversely proportional to the size of the probability value of the corresponding range. In this case, three band sections are generated differently from the conventional method of FIG. 6.

Here, the band offset section determination unit determines the central bands so that the center of the entire section becomes the center of the central bands when the reference range does not exist, but the central bands are You can decide to occupy half of the entire range.

Referring back to FIG. 1, after the adaptive sample offset (SAO) processing step S120, an adaptive loop filter (ALF) may be selectively applied (S130). Step S130 is a step that can be selected as needed. In this step, the adaptive loop filter (ALF) may compensate for errors when encoding more precisely than the deblocking filter by using a Wiener filter. Specifically, the adaptive loop filter (ALF) uses a deblocking filter and an adaptive offset (ALF) using a Wiener filter that minimizes the sum of square errors of differences between original pixels and decoded pixels. The error can be compensated for by more precisely encoding the signal restored after SAO). The adaptive loop filter (ALF) information may be included in a slice header and transmitted to a decoder. The adaptive loop filter (ALF) information may include a Wiener filter coefficient, adaptive loop filter (ALF) on/off information, and filter shape information. On/off information of the adaptive loop filter (ALF) may be included in a slice header in units of coding units and transmitted to a decoder. By designing the filter shape to have a symmetrical shape, it is possible to reduce the complexity of an encoder and a decoder.

Decryption device

10 is a block diagram illustrating a configuration of an image decoding apparatus using an adaptive band offset interval according to an embodiment of the present invention.

Referring to FIG. 10, a decoding apparatus according to an embodiment of the present invention includes an entropy decoding unit 731, an inverse quantization unit 733, an inverse transform unit 735, a motion compensation unit 737, and an intra prediction unit 739. , A frame buffer 741, an adder 743, and an in-loop filter 745.

The entropy decoding unit 731 receives the compressed bit stream and performs entropy decoding to generate quantized coefficients. The inverse quantization unit 733 and the inverse transform unit 735 restore residual values by performing inverse quantization and inverse transformation on the quantized coefficients.

The motion compensation unit 737 performs motion compensation using information decoded from the bit stream by the entropy decoding unit 731 to generate a predicted prediction unit.

The intra prediction unit 739 performs intra prediction on the current prediction unit according to an intra mode from previously coded pixel values around the current prediction unit in the current picture.

The addition unit 743 restores an image by adding the residual value provided by the inverse transform unit 735 to the predicted prediction unit provided by the motion compensation unit 737 and provides the reconstructed image to the in-loop filter 745.

12 is a block diagram showing the configuration of the in-loop filter 745 of FIG. 10.

As shown in FIG. 12, the in-loop filter 745 includes a deblocking filter 1110 that removes block distortion of a reconstructed image, and an adaptive sample offset for the deblocking filtered image. , SAO) processing may include a SAO processing unit 1120. The in-loop filter 745 is deblocking filtering to remove block distortion of the reconstructed image provided by the addition unit 743, and an adaptive sample offset of the deblocking filtered image. , SAO) processing, and finally providing the reconstructed image to the frame buffer 741.

The frame buffer 741 stores the reconstructed image. That is, the decoder performs a decoding operation by adding the compressed prediction error (residual value provided by the inverse transform unit) to the prediction unit.

As shown in FIG. 13, the SAO processing unit 1120 of the present embodiment includes a band offset section determining unit 1122, a sample offset type determining unit 1124, and a sample offset summing unit 1126. The SAO processor may determine a band offset section of the current block based on a pixel value distribution characteristic of a reference range associated with the current block.

The band offset interval determiner may determine the reference range as a set of pixels of a block adjacent to the current block, and in particular, may determine a set of pixels of an upper block and a left block of the current block as the reference range. In addition, a set of pixels surrounding the current block with a preset thickness may be determined as a reference range, and may be determined to include a set of pixels of the current block.

The band offset section determining unit may determine a section in which the pixel values of the reference range are distributed as central bands, and determine the central bands so that the pixel value in which the most pixels of the reference range are distributed becomes the center. However, the section boundary may be determined so that the central bands occupy 1/2 of the entire range. In addition, when the reference range does not exist, the central bands are determined so that the center of the entire section becomes the center of the central bands, but the central bands are 1/2 of the entire range. You can decide to occupy it.

The sample offset type determiner 1124 determines whether the band offset type of the current block is a center interval offset or a side interval band offset.

The sample offset summing unit 1126 adds the offset values calculated for each band with respect to pixels belonging to a section determined by the type according to the determined sample offset type.

A specific operation aspect of the in-loop filter 745 included in the decoding apparatus is according to the in-loop filtering method using the adaptive band offset interval according to an embodiment of the present invention.

Encoding device

11 is a block diagram illustrating a configuration of an image encoding apparatus using an adaptive band offset interval according to an embodiment of the present invention.

Referring to FIG. 11, an image encoding apparatus includes an encoder 630, and the encoder 630 includes an inter prediction unit 632, an intra prediction unit 635, a subtractor 637, a transform unit 639, and a quantization unit. A unit 641, an entropy encoding unit 643, an inverse quantization unit 645, an inverse transform unit 647, an adder 649, a frame buffer 651, and an in-loop filter 653 may be included. The inter prediction unit 632 includes a motion prediction unit 631 and a motion compensation unit 633.

The input image is divided into a recursive coding unit (CU), and the coding unit performs inter prediction by the inter prediction unit 632 or intra prediction by the intra prediction unit 635. It is divided into a prediction unit (PU) unit. The coding unit CU is partitioned hierarchically. A smallest CU (SCU) may have a size of at least 8x8 pixels, and a large CU (LCU) may have a size of 64x64 pixels. Information related to the SCU and the LCU may be signaled to the decoder through a sequence parameter set (SPS).

The inter prediction unit 632 generates a motion vector by estimating a motion of the provided prediction unit to be currently encoded.

The motion prediction unit 631 searches for a region similar to the current block in a previous N-1 reference picture and/or a subsequent N+1 reference picture of the currently-coded N-th picture using the provided current prediction unit, and a motion vector in block units. Create

The motion compensation unit 633 generates a prediction block (or a predicted prediction unit) obtained by performing motion compensation using a motion vector and a reference picture generated from the motion prediction unit 631.

The intra prediction unit 635 performs intra prediction on the current prediction unit according to an intra mode from previously coded pixel values around the current prediction unit in the current picture.

The subtractor 637 generates a residual value by subtracting the prediction block (or predicted prediction unit) provided by the motion compensation unit 633 and the current block (or current prediction unit), and the transform unit 639 and the quantization unit 641 ) Transforms the residual value (residue) into a DCT (Discrete Cosine Transform) and quantizes it.

The entropy encoding unit 643 generates a bit stream by entropy encoding information such as quantized DCT coefficients, motion parameters, and information related to SCU and LCU.

The inverse quantization unit 645 and the inverse transform unit 647 inverse quantize and inversely transform the quantized data through the quantization unit 641. The adder 649 restores the image by adding the inverse transformed data and the predicted prediction unit provided by the motion compensation unit 633 and provides the reconstructed image to the in-loop filter 653.

12 is a block diagram showing the configuration of the in-loop filter 653 of FIG. 11.

As shown in FIG. 12, the in-loop filter 653 includes a deblocking filter 1110 for removing block distortion of a reconstructed image, and a sample adaptive offset for the deblocking filtered image. , SAO) processing may include a SAO processing unit 1120. The in-loop filter 653 includes deblocking filtering to remove block distortion of the reconstructed image provided by the adder 649, and adaptive sample offset (Sample Adaptive Offset) for the deblocking filtered image. SAO) processing, and finally providing the reconstructed image to the frame buffer 651.

The frame buffer 651 stores the reconstructed image.

Here, the SAO processing unit 1120 may be characterized in determining a band offset section of the current block based on a pixel value distribution characteristic of a reference range associated with the current block.

The band offset section determiner 1122 may adaptively determine the boundary between the center section and the side section by considering the band offset section according to the current block or in consideration of the distribution of pixel values of the reference range related to the current block. Since the method of determining the band offset period has been described above, a common description will be omitted. A common description of the sample offset type determination unit and the sample offset summing unit is omitted. A detailed operation of the in-loop filter 653 included in the encoding apparatus is according to the in-loop filtering method using the adaptive band offset section according to the embodiment of the present invention, and thus a description thereof will be omitted.

Claims (10)

In the in-loop filtering method,
Performing deblocking filtering on the reconstructed picture; And
Including the step of applying a sample offset (Sample Adaptive Offset, SAO) to the current block belonging to the deblocking filtered image,
Applying the sample offset,
Adaptively determining a band offset period of the current block in consideration of the reference area of the current block; Here, the band offset period includes a plurality of bands,
Calculating the sample offset for each of the bands; And
Including the step of adding the sample offset to the current pixel belonging to the current block,
In-loop filtering method, the total number of bands belonging to the entire range of values that a pixel can have is 32. The method of claim 1, wherein the reference area,
An area specified by pixels belonging to an area adjacent to the current block according to a predetermined criterion, or
The in-loop filtering method, which is an area specified by pixels belonging to a block having similarity to the current block according to a predetermined criterion. The method of claim 2, wherein an area specified by pixels belonging to an area adjacent to the current block according to the predetermined criterion is
The in-loop filtering method comprising at least one of an upper block, a left block, and an upper right block of the current block. The method of claim 1,
32 bands belonging to the entire range of values that the pixel can have are divided into a plurality of band sections,
The number of bands included in at least one of the plurality of band sections is 4, 8 or 12,
The band offset interval is any one of the plurality of band intervals, in-loop filtering method. The method of claim 1,
The sample offset added to the current pixel is an offset calculated for a band to which the current pixel value belongs among the plurality of bands. In the video decoding apparatus,
A deblocking filter that performs deblocking filtering on the reconstructed picture; And
Including a SAO processing unit for applying a sample offset (Sample Adaptive Offset, SAO) to the current block belonging to the deblocking filtered image,
The SAO processing unit,
In consideration of the reference region of the current block, a band offset period of the current block is adaptively determined, wherein the band offset period includes a plurality of bands,
For each of the bands, calculating the sample offset, and
Add the sample offset to the current pixel belonging to the current block,
The image decoding apparatus, wherein the total number of bands belonging to the entire range of values that a pixel can have is 32. The method of claim 6, wherein the reference area,
An area specified by pixels belonging to an area adjacent to the current block according to a predetermined criterion, or
An image decoding apparatus, which is an area specified by pixels belonging to a block having similarity to the current block according to a predetermined criterion. The method of claim 7, wherein an area specified by pixels belonging to an area adjacent to the current block according to the predetermined criterion is
An image decoding apparatus comprising at least one region of an upper block, a left block, and an upper right block of the current block. The method of claim 6,
32 bands belonging to the entire range of values that the pixel can have are divided into a plurality of band sections,
The number of bands included in at least one of the plurality of band sections is 4, 8 or 12,
The band offset period is any one of the plurality of band periods, the image decoding apparatus. The method of claim 6,
The sample offset added to the current pixel is an offset calculated for a band to which the value of the current pixel belongs among the plurality of bands.

Images