一种方向提升小波变换和网格编码量化的图像编码算法

由于JPEG2000图像压缩标准不能有效表示图像边缘的缺陷,提出了一种结合方向提升小波变换和网格编码量化(TCQ)的图像编码算法.该算法先通过基于率失真优化的四叉树分割算法得到最优变换方向,再沿最优变换方向做方向提升小波变换,然后对得到的小波子带系数进行TCQ量化,最后将量化系数编码输出.实验结果表明,对所给定的测试图像,该算法比JPEG2000峰值信噪比最多提高了1.49 dB,比JPEG2000的结构相似度最多提高了3.98%,解码图像的主观视觉质量更好.

An Image Coding Algorithm Combining Directional Lifting Wavelet Transform and Trellis Coded Quantization

An image coding algorithm combining the directional lifting wavelet transform (DL-WT) and the trellis coded quantization (TCQ) is presented to overcome the drawback of JPEG2000 image compression standard in representing image edges. This algorithm firstly uses a rate-distortion optimization-based quadtree segmentation method to get the optimal transform di-rections, performs DLWT along the optimal transform directions, then quantizes the wavelet subband coefficients by TCQ, and finally encodes the quantization coefficients outputting the bit-stream. Experimental results show that the proposed image coding algorithm outperforms JPEG2000 for the given test images up to 1.49 dB in peak signal-to-noise ratio and 3.98% rela-tively in multi-scale structural similarity index, and that significant improvement in the subjective visual quality of decoded images is also observed.