首先根据当前运动图象压缩的国际标准,讨论了极低码率下的编码问题,然后介绍了Matching Pursuit在该领域中的应用,其具体过程是利用二维Gabor函数字典对帧间预测差值进行Matching Pursuit[1],这样能获得低码率下更高的压缩比和更好的效果,而且由于函数字典不是基于块的,因此消除了在低码率下DCT变换的方块效应.此外,在进行帧间预测差值编码时,是从能量最集中的区域开始,逐渐分配比特,从而使得解码后的图象能体现运动变化显著的区域.

After reviewing some international recommendations for video compression, the application of Matching Pursuit on video compression at very low bit rate is introduced. This algorithm decomposes the signal into linear expansions of waveforms using an overcomplete basis of Gabor functions. With adapting Matching Pursuit to the motion residual, a substantial improvement in both PSNR and perceived visual quality can be gotten. The dictionary which is not block-based can reduce block distortions compared with the low bit rate DCT systems. When coding motion residual, bits are assigned progressively to high energy areas in the motion residual. Thus the decoded image can show the areas where motion changes quickly. Another advantage is the flexible choice of dictionary.