嵌入式平台上JPEG2000简化算法研究

Simplification algorithms for JPEG2000 on embedded platform

本文在深入分析JPEG2000图像压缩算法的基础上,针对嵌入式平台数据处理能力较弱的缺点,提出了量化步长选择、改进的率失真处理等多种优化方案,并对离散小波变换在XScale上的数值计算问题进行了讨论,从而显著降低了算法实现的复杂度。实际结果证明,在基本不牺牲编码效率的前提下,本文提出的优化方法大幅度提高了编、解码速度——其中编码器速度提高90%左右,解码器速度提高30%左右。

Based on the analysis of the JPEG2000 image coding algorithm and focusing to the weak data processing ability of embedded platforms, a set of simplification and optimization algorithms are proposed in this paper to reduce the coding complexity of JPEG2000. The quantization parameter is chosen adaptively and the rate-distortion optimization is improved. The numeric calculation for wavelet transform is also discussed for XScale platform. It can be proved by the experiments that the proposed algorithms can speed up the coding and decoding process of JPEG2000 while the coding efficiency is maintained. Exactly, the speed is improved about 90% and 30% for encoder and decoder respectively.