Coding mode determination method and device, equipment and storage medium

Abstract

The embodiment of the invention discloses a coding mode determination method and device, equipment and a storage medium, and the method comprises the steps: dividing a coding unit under each coding specification into conversion units under a next coding specification adjacent to the coding specification, and calculating the coding cost of each conversion unit; and sequentially performing cost optimization integration on the coding unit and the conversion unit under the same coding specification to obtain an overall coding mode under the optimal coding cost. According to the technical scheme provided by the embodiment of the invention, the coding cost can be minimized when the overall coding mode is adopted for video coding, the adaptability of the video coding modes adopted under differentvideos is ensured, the calculation overhead of video coding is reduced, and the video coding rate is further improved.

Description

technical field [0001] The embodiments of the present invention relate to the technical field of video coding, and in particular, to a method, device, device and storage medium for determining a coding mode. Background technique [0002] At present, video coding is usually coded with the coding unit (Coding Unit, CU) as the basic unit. At this time, the coding unit CU has four sizes: 64*64, 32*32, 16*16, and 8*8. All of them need to recursively try the quadtree from the 64*64 size CU to the 8*8 size CU in order to select the coding size with the best efficiency. At this time, the recursion depth of the 64*64 size CU is 0 , adding 1 to the depth of each downward recursion, and the recursion depth of the 8*8 size CU is 3. At the same time, at each recursion depth, the encoding cost of the current non-recursive CU and the current CU are divided after recursion Compare the sum of the coding costs of the 4 small CUs to determine whether recursive partitioning is required at this...

AI Technical Summary

Problems solved by technology

[0003] Each CU usually includes two parts, the Transform Unit (TU) and the Prediction Unit (PU). The TU mainly performs the basic rate-distortion optimization (RDO) in consideration of the impact of the video coding bit rate and distortion. ) process, the PU mainly tries to calculate the encoding cost under the RDO process after the CU is recursively divided into TUs in different prediction modes; 2NxN / Nx2N / 2NxnD / 2NxnU / nRx2N / nLx2N / Intra2Nx2N / IntraNxN12 types), the prediction data during the TU execution process is obtained, and then the calculated residual data is transformed, quantized and entropy encoded to obtain the current prediction mode Encoding bit rate, at the same time, dequantize and inverse transform the quantized residual data to obtain the corresponding reconstructed data, and then compare it with the original residual data to obtain the corresponding encoding distortion, and then according to the encoding bit rate and encoding distortion Calculate the encoding cost in this prediction mode, and then calculate the encoding cost of the current non-recursive CU at each recursion depth and the sum of the encoding costs of the 4 small CUs divided by the current CU after recursion, so as to judge the current CU Whether recursive division is required; at this time, each recursive situation from a 64*64 size CU to an 8*8 size CU needs to calculate the corresponding encoding cost one by one based on the TUs divided under the CU, resulting in a large amount of calculation burden. Increased resource overhead for video encoding

[0004] Therefore, in order to reduce the calculation overhead of TU when CU is recursively divided, there are usually the following two methods: 1) Use the coefficients in the quantization domain to predict the coding distortion, and establish a code rate model to predict the coding code rate, so that there is no need for residual Data dequantization and inverse transformation process, but the accuracy of encoding distortion and encoding rate estimation is high, once the estimation is inaccurate, it will cause increased video encoding loss; 2) According to the existing encoding information or calculation Analyze the video texture under each TU to determine whether recursive division is required. This will increase the amount of additional calculations. At the same time, the method of uniformly calculating the encoding cost will lose the encoding quality under different components.

Images