静态图象的一种轮廓—纹理编码方法

A controur-texture Coding method for still image

本文提出了一种新的静态图象高压缩比编码方法——轮廓-纹理编码方法。首先利用模糊(Fuzzy)集合理论,对图象数据在频域与空域上作预处理,去掉部分冗余信息,提取并细化轮廓。然后利用FREEMAN链码对轮廓进行编码。设计了一种区域标号化的方法,对纹理图象进行编码,从而提高了压缩比。为验证算法的有效性,对USC-GIRL标准图象在IBM PC/AT机上进行了模拟实验,从恢复图象的质量与压缩比综合衡量,结果是令人满意的。

A new contour-texture coding method is introduced. The image is preprocessed using fuzzy sets on the spatial and frequency domain to cut out a part of redundance information and to abstract and fine the contour. The contour is encoded by employing Freeman codes. A region lablized method is designed to encode the texture so as to get a higher impression ratio. A standard image of USC-GIRL is simulated on the IBM PC/ AT computer to illustrade the effectiveness of the algorithm, which, in a good way, results in trading off the image quality and the high compression ratio.