Method and apparatus for video coding using versatile information-based context model
A method and an apparatus for video coding using a video-characteristic-based context model are disclosed. The video coding method and the video coding apparatus determine a context model optimized for an individual video on the basis of information in which various characteristics of the individual video are reflected in entropy encoding and decoding using context adaptive binary arithmetic coding (CABAC).
1. An entropy decoding method, performed by a video decoding apparatus, for decoding a syntax element of a block level related to a current block, the entropy decoding method comprising:
acquiring encoded data related to the syntax element of the block level;
deriving a context model based on encoding information of at least one neighboring block of the current block and high level video information of a higher level than the block level related to the current block, wherein the context model designates an initial value of a symbol value occurrence probability;
generating a bin string by decoding the encoded data using the context model; and
obtaining a value of the syntax element of the block level by inversely binarizing the bin string,
wherein generating the bin string comprises:
generating at least one bin of the bin string using the symbol value occurrence probability; and
updating, for use in generating a next bin, the symbol value occurrence probability according to a decoding result for the at least one bin.
2. The entropy decoding method of claim 1 , wherein the encoding information of at least one neighboring block is information of syntax elements of a left block and a top block, which are adjacent to the current block.
3. The entropy decoding method of claim 1 , wherein the high level video information includes all or some of resolution information, encoding parameter information, or chrominance channel information as 2 D video information.
4. The entropy decoding method of claim 1 , wherein the high level video information includes all or some of XR (extended Reality) content information, light field video information, or point cloud video information as immersive video information.
5. An entropy encoding method, performed by a video encoding apparatus, for encoding a syntax element of a block level related to a current block, the entropy encoding method comprising:
acquiring a bin string for the syntax element of the block level by binarizing a value of the syntax element of the block level;
deriving a context model based on encoding information of at least one neighboring block of the current block and high level video information of a higher level than the block level related to the current block, wherein the context model designates an initial value of a probability of a symbol value; and
generating encoded data related to the syntax element of the block level by encoding the bin string using the context model,
wherein generating the encoded data related to the syntax element of the block level comprises:
encoding at least one bin of the bin string by using a symbol value occurrence probability; and
updating, for use in encoding a next bin, the probability of the symbol value according to an encoding result for the at least one bin.
6. The entropy encoding method of claim 5 , wherein the encoding information of at least one neighboring block is information of syntax elements of a left block and a top block, which are adjacent to the current block.
7. The entropy encoding method of claim 5 , wherein the high level video information includes all or some of resolution information, encoding parameter information, or chrominance channel information as 2D video information.
8. The entropy encoding method of claim 5 , wherein the high level video information includes all or some of XR (extended Reality) content information, light field video information, or point cloud video information as immersive video information.
9. A method for providing a video decoding apparatus with video data, the method comprising:
encoding a syntax element of a block level related to a current block; and
transmitting the syntax element to the video decoding apparatus,
wherein encoding of the syntax element of the block level comprises:
acquiring a bin string for the syntax element of the block level by binarizing a value of the syntax element of the block level;
deriving a context model based on encoding information of at least one neighboring block of the current block and high level video information of a higher level than the block level related to the current block, wherein the context model designates an initial value of a probability of a symbol value; and
generating encoded data related to the syntax element of the block level by encoding the bin string using the context model,
wherein generating the encoded data related to the syntax element of the block level comprises:
encoding at least one bin of the bin string by using a symbol value occurrence probability; and
updating, for use in encoding a next bin, the probability of the symbol value according to an encoding result for the at least one bin.