JPEG2000码率控制截断预测算法及VLSI设计

Rate Control Truncation Algorithm Based on Prediction and VLSI Architecture for JPEG2000

针对JPEG2000低码率下的码率控制算法在运算量和存储量上冗余较大的问题,论文提出一种基于预测的码率控制截断算法。论文算法根据目标码率对图像数据样本采样,进行率失真斜率的预测,提前截断块编码过程,并提出一种JPEG2000系统两级码率控制的VLSI结构,第一级采用论文提出的预测算法截断码流,第二级确定精确的分层截断点。测试结果表明论文算法在码率低于0.5比特/像素时,可节省运算量和存储量70%以上,而峰值信噪比基本没有降低。

A rate control truncation algorithm based on prediction is proposed to resolve the problem of computation complexity and large memory usage in the traditional JPEG2000 rate control algorithm when the bit-rate is low.By taking samples of image data for target bit-rate and predicting the rate-distortion slope,the algorithm truncates the block coding in advance,and the VLSI architecture containing two stages rate control for JPEG2000 system is proposed.The code-stream is truncated by the prediction algorithm in the first stage,and the hierarchical truncation points are determined in the second stage.Experiment indicates that the proposed algorithm has efficiently reduced above 70% computations and memory usage with no reduction in the peak signal to noise ratio.