图像的视觉加权子带编码

Human Visual Weighted Sub-band Coding of Images

本文给出了一种考虑人类视觉特性的图像子带编码方法。首先通过一个子带分析系统将图像分割成处于不同频带范围的子带图像,然后基于人类视觉系统对不同频带视觉信号的感受特性分别对不同频带的子带图像进行视觉加权,进而采用PCM及DPCM方法分别对于带图像量化及编码。同时本文还给出了各向异性的视觉系统模型,并据此对子带图像的编码数目进行选择,在很大程度上降低了图像的数据量。计算机模拟实验表明,在高压缩比情况下,该方法仍能获得主观满意的解码图像质量。

A subband coding method of images incorporating properties of Human Vision is presented in this paper. An image is first decomposed into several subband images localized in different frequency ranges, and then on the basis of the selectivity of Human Visual System the subband images are weighted respectively. The PCM and DPCM are used to quantize and code the subband images. An anisotropical Human Visual System model is also given in this paper. The numbers of subband images are selected in accordance with the anisotropical model, resulting in the reduction of image data to a great extent. The computer simulation results shove that a subjectively more pleasing quality of decoded image is obtained at realative high data compression rates.