Entropy-coding for video encoding and decoding
A method is executed to efficiently operate a bin buffer to limit a bin-to-bit ratio in entropy encoding and decoding related to bitstream generation and parsing. In addition, a method of configuring a list includes various entropy encoding/decoding methods and adaptively uses the entropy encoding/decoding methods for each basic unit of entropy encoding/decoding.
1. A method for entropy decoding executed by a video decoding apparatus, the method comprising:
receiving a bitstream that has been generated by encoding a picture;
performing an arithmetic decoding process to generate at least one bin string each including at least one bin by decoding the bitstream; and
generating a syntax element through inverse binarization of the bin string,
wherein the number of bins generated by decoding the bitstream satisfies a constraint that the number does not exceed a threshold,
wherein the threshold is a value calculated based on a size of the picture, a sampling ratio between a luma component and a chroma component, and a bit depth, and
wherein the threshold is variably set according to a tier of a video.
2. The method of claim 1 , wherein in order to satisfy a constraint that the number of bins does not exceed the threshold, the performing of the arithmetic decoding process comprises:
increasing a counter by N per bit read and decrementing the counter by M according to occurrence of one bin, wherein M and N are natural numbers and are determined based on the threshold; and
selectively performing context-based adaptive binary arithmetic decoding or uniform probability-based binary arithmetic decoding depending on whether the counter is greater than N.
3. The method of claim 1 , wherein the arithmetic decoding process comprises:
generating a detailed section from a current section for decoding;
obtaining a current probability value for a most probable symbol (MPS) using a first context model and a second context model for each of the syntax element, and generating an MPS value;
calculating a length of a section of a least probable symbol (LPS) using the detailed section and the MPS value, and updating the current section using the length of the section of the LPS;
generating one bin using an offset of a preset length acquired from the bitstream and the updated current section, and re-adjusting, when a value of the LPS is determined as the one bin, the offset and the current section, wherein the value of the LPS is generated from a value of the MPS;
updating the first context model and the second context model; and
when the readjusted current section is less than a preset threshold, renormalizing the current section and the offset.
4. The method of claim 3 , wherein the updating of the first context model and the second context model comprises:
calculating a first adaptation rate and a second adaptation rate for each of the syntax element; and
updating the first context model using the first adaptation rate and updating the second context model using the second adaptation rate,
wherein the first context model is clipped to a maximum possible value of a probability of the first context model, and the second context model is clipped to a maximum possible value of a probability of the second context model.
5. A method for entropy encoding syntax elements generated according to predictive encoding of each block constituting a picture, the method comprising:
binarizing each of the syntax elements and generating at least one bin string each including at least one bin;
performing an arithmetic coding process to generate encoded data from the bin string; and
generating a bitstream comprised of one or more network abstraction layer (NAL) units from the encoded data,
wherein the number of bins relative to a length of the one or more NAL units is constrained not to exceed a threshold,
wherein the threshold is a value calculated based on a size of the picture, a sampling ratio between a luma component and a chroma component, and a bit depth,
wherein the threshold is variably set according to a tier of an image.
6. The method of claim 5 , wherein when the number of bins relative to the length of the one or more NAL units exceeds the threshold, each of the one or more NAL units is configured by inserting one or more zero words of a predefined length.
7. The method of claim 5 , wherein in order to satisfy a constraint that the number of bins does not exceed the threshold, the performing of the arithmetic coding process comprises:
increasing a counter by N per bit and decrementing the counter by M per bin, wherein M and N are natural numbers and are determined based on the threshold; and
selectively performing context-based adaptive binary arithmetic coding or uniform probability-based binary arithmetic coding depending on whether the counter is greater than N.
8. A method for transmitting a bitstream containing encoded video data, the method comprising:
generating the bitstream by encoding syntax elements generated according to predictive encoding for each block constituting a picture; and
transmitting the bitstream to a video decoding apparatus,
wherein generating the bitstream comprises:
binarizing each of the syntax elements to generate at least one bin string each including at least one bin;
performing an arithmetic coding process to generate encoded data from the bin string; and
generating a bitstream comprised of one or more network abstraction layer (NAL) units from the encoded data,
wherein the number of bins relative to a length of the one or more NAL units is constrained not to exceed a threshold,
wherein the threshold is a value calculated based on a size of the picture, a sampling ratio between a luma component and a chroma component, and a bit depth,
wherein the threshold is variably set according to a tier of an image.