一种基于小波零树的彩色图像编码算法

Algorithm for color image coding based on wavelet zerotree

为提高基于分层树的集合分割算法(SPIHT)的编解码速度,对算法结构重新设计,提出了一种简单的无链表小波零树编码算法.通过预处理和主处理过程实现了流水线编码结构,减少了存储器的访问频率.预处理完成对所有节点显著性的标注,主处理完成无链表的零树编码.对彩色图像编码时首先进行RGB空间到YUV空间的变换,然后在比特平面扫描过程中,依次对YUV三个平面进行编码,这样输出的码流是完全嵌入式的.一般情况下,亮度分量的最高显著平面都高于色度分量.因此通过色度分量输出控制策略,可以进一步提高算法对彩色图像的编码性能.硬件实现结果表明,改进后算法的编解码速度明显提高,而率失真性能并没有下降,所以新算法特别适合高分辨率或不规则图像的编解码芯片设计.

To improve the coding speed of set partitioning in hierarchical trees (SPIHT) algorithm, the structure of SPIHT was re-designed, a simple listless coding algorithm based on wavelet zerotree was developed. Pipeline coding structure was realized by pre-processing and main-processing processes and access frequency was reduced. Pre-processing was designed to mark the significance of all nodes and main-processing to finish listless zerotree coding. In color image coding, the transform from RGB (red-green-blue) space to YUV (luminance-chrominance) space was performed, YUV color planes was correspondingly coded on bit plane sorting. So, the output coding stream was fully embedded. Generally, the first significant bit plane of luminance component is higher than chrominance component's. The performance of color image coding is further improved by output-controlled policy of chrominance component. Hardware implementation results show that the speed of codec is improved significantly, but rate-distortion performance is not reduced. The new algorithm is especially suitable for high resolution or irregular image codec chip design.