基于小波变换的分类量化图像编码算法

Wavelet Image Coding Based Scalar-Vector Quantization

针对标量量化压缩比小而向量量化压缩速度慢、图像复原效果不理想等弱点 ,提出了基于小波变换的分类量化图像编码算法 (简称“分类量化编码”)。该算法基本思想为 :首先将小波变换后的图像高频子带划分为局部块 ;然后利用文中给出的相对距离最近之阈值选择方法 ,依据纹理复杂度和重要性程度将这些局部块划分为4类 (平坦、过渡、弱纹理和强纹理 ) ;最后对平坦局部块进行向量量化编码 ,对强纹理局部块进行标量量化编码。实验结果表明 :该图像压缩算法在压缩速度、图像复原效果、压缩比等方面明显优于零树小波编码和 JPEG方法。

To improve the compression ratio, image quality and coding efficiency, an image coding algorithm based on wavelet transform and scalar vector quantization is presented. Firstly, the image is decomposed into a series of different frequency sub images by wavelet transform. Secondly, the high frequency sub images are divided into local blocks. Thirdly, the local blocks are classified into four classes(flat, transition, weak vein and strong vein) according to the textural intensity of every local block in high frequency subimage by using the new threshold selection method. Finally, the flat blocks are quantized by using scalar and the strong vein blocks are quantized by using vector. It is shown experimentally that the new image compression scheme performs better than that of EZW and JPEG in the aspects of compression ratio, image quality and coding efficiency.