A Video-Coding Quantization Scheme for an Embedded System

Abstract

The traditional MPEG quantization process leads to a low efficiency for the video compression, owing to lacking of a division component inside the general embedded processors. In order to solve this problem, a new scheme of quantization is proposed in this paper. According to the traits of the general embedded processors, an optimal quantization matrix is constructed, which realizes one-step quantization by substituting shift-operations for division-operations. Meanwhile, this scheme uses different quantization strategies in terms of the differences between the video blocks to be coded, and reduces the image quality loss caused by the unified quantization. Experiments show that this scheme can not only obviously improve the efficiency of quantization, but subjective and objective quality of coded images as well.