摘要： 基于小波零树编码的相似结构,给出了一种均值去除的嵌入式DCT(MSEDCT)编码方法。该方法将子块DCT后的系数按一定方式重新组合,并针对重组后图像DC系数块存在的大量冗余信息,运用简单的均值去除方法进一步降低冗余信息。在此基础上,运用小波零树量化器及算术熵编码器实现对图像的有效压缩。实验结果表明,该方法在获得较高压缩比的同时,呈现出较好的图像质量, 并具有比特率可分级特性,适合于无线网络或Internet上图像的累进传输或鲁棒传输。

关键词: 嵌入式信源编码, 小波零树, 比特率可分级

Abstract: A mean-subtract embedded DCT coding method(MSEDCT) based on the similar wavelet zerotree structure is proposed. The method rearranges coefficients of block DCT in some ways and reduces high correlation between DC coefficients by a simple mean-subtract method. Then based on the wavelet zerotree quantizer and its arithmetic entropy encoder, an image is compressed effectively with high PSNR at high compression ratios. The bitstream by this method is rate-scalable, suitable for image progressive transmission or robust transmission over wireless channels or Internet.

中图分类号: 

• TN919.81