The Optimization Research Of Intra Prediction And Deblocking Filter Based On HEVC

With the growing demand for definition, compression and frame rate of video storage and transmission, the latest generation of video coding standard HEVC came into being. HEVC can compress the video to about half the bit rate of the H.264/AVC, while maintaining the same encoded image quality as H.264/AVC. While the cost of increasing the HEVC compression efficiency is a substantial increase in complexity.In order to reduce the complexity of HEVC, this paper aims at the intra prediction and deblocking filter module of HEVC encoder to optimize the study, and the improvement of the algorithm and framework is proposed and implemented. In terms of intra prediction, this paper presents a down-sampling fast CU partition algorithm. The fast intra prediction algorithm performs down-sampling on the original image and performs intra prediction on the down-sampled degraded image. Then,according to the optimal CU partitioning result of down-sampling, only intra mode prediction of the down-sampling optimal partition depth and the next depth is performed for the original image, which can simplify the intra-coding CU partitioning process, reduce the computational cost of mode prediction, and speed up the intra-frame coding. In terms of deblocking filter, a new filtering framework with low complexity is proposed. The low complexity framework divides the deblocking filter module into the boundary strength calculation part and the filter decision and operation part. Then, the boundary strength calculating module,which removes the logical redundancy and the calculate redundancy, is integrated into the prediction coding module, and the CU recursive partitioning logic of filter decision and operation is adjusted, so that the redundancy of the deblocking filter module is reduced and the parallelism is improved, which saves the processing time of the filtering module. The results of experiments show that the proposed down-sampling fast CU partitioning intra-coding algorithm and low-complexity deblocking filter framework can significantly reduce the coding time without degrading or losing image quality.