Method for 4*4 sub-macroblock parallel intraframe prediction in H.264/AVC

Abstract

The present invention discloses a method for 4*4 sub-macroblock parallel intraframe prediction in H.264 / AVC, comprising the steps of: a unified intraframe prediction formula, a reference value array, a reference position table and a concrete implementation step of parallel intraframe prediction. The unified intraframe prediction formula is improved based on the feature of a CUDA and an intraframe prediction formula, by the corresponding prediction formulas of 9 prediction modes to be transformed into one formula to meet the requirements of the CUDA multi-thread single instruction multiple data stream, to achieve the fine-grained parallel in the intraframe prediction sub-macroblock. The reference value array and the reference position table are designed to cooperate with the unified intraframe prediction formula and to completely eliminate a large number of branch statements that affects the parallel algorithm performance. The present invention in the process of the intraframe prediction achieves the pixel level parallel, can effectively use multi-core resources in a GPU to accelerate the process of intraframe prediction, and shorts a coding time.

Description

technical field [0001] The invention belongs to the field of video coding and the field of general computing of graphics cards, and more specifically relates to a parallel intra-frame prediction method of 4x4 sub-macroblocks in H.264 / AVC. Background technique [0002] The H.264 / AVC standard is the current mainstream video compression standard and is widely used in many fields. Different from the previous standards, H.264 / AVC proposes a new pixel-based intra-frame prediction method, which performs traversal calculations for 9 available prediction modes and various block sizes, and obtains it through the rate-distortion formula A prediction mode with good effects in both video quality and video compression rate, however, this full-mode prediction makes the calculation complexity of the whole process quite high. [0003] In order to reduce the encoding time of full-mode prediction, researchers have proposed many effective solutions, among which the method based on GPU hardware...

AI Technical Summary

Problems solved by technology

However, the existing algorithms only stay on the coarse-grained parallel method, and the fine-grained pixel-level parallelism is still not involved, and the performance of the GPU cannot be fully utilized.

The existing technical solutions always use a single thread to traverse and calculate the 9 prediction modes in the 4x4 sub-macroblock, and cannot use multiple threads to process all pixels in parallel, because the prediction formulas corresponding to each pixel in the coding block are different , and the reconstructed pixel value referenced by each pixel also changes with its own position, and further this situation will lead to a large number of branch statements, so it is not suitable to use CUDA for parallel processing

Images