基于三维小波变换的视频图像编码算法

An Image Sequence Compression Coding Scheme Based on 3-D Wavelet Transform

针对现有图像压缩标准（ＤＣＴ变换＋运动补偿）在高压缩比时，方块效应严重影响主观质量这一问题，提出了一种基于三维小波变换的图像序列压缩算法。该算法在量化编码时，突破原有算法对系数逐帧处理的框架，将着眼点由“面”转向“体”；采用平面与纵向相结合的形式组织系数，并对运动程度不同的区域加以区分。以ＱＣＩＦ格式的标准序列图像Ｍｉｓｓ Ａｍｅｒｉｃａｎ作为测试图像进行实验，实验结果表明，该算法可将测试图像序列数据流控制在 ３２ｋｂ／ｓ（１０帧／ｓ）以内，主观质量较好，无方块效应。

A new scheme based on 3-D wavelet transform for image sequence compression coding is proposed. The coefficients in the same sub-image are first organized as a unit, and so the correlation, including spatial-temporal correlation and father-children correlation from hierarchical blocks can be explored more efficiently. Ac- cording to the motion character, the coefficients in the same sub-image are treated separately. Computer simulation for the standard test image 'Miss America' shows that the reconstructed image with the proposed coding scheme can hold acceptable subject quality with a rate below 32kb/s (10 frame/s).