摘要
图像矢量量化(VQ)是图像压缩算法中的重要环节,在VQ中起决定性因素的是构造出性能优异的码书。为改善矢量量化码书的性能,文中在分析Kohonen自组织特征映射(SOFM)的基础上,提出一种识别距离SOFM的算法,同时将矢量量化应用于图像的小波变换域。测试结果表明,改进的算法使码书设计的计算量得到明显的降低,而且码书的性能得到了提高。
Image Vector Quantization (VQ) plays an important role in image compression algorithm. A decisive factor in VQ is to construct the codebook of outstanding performance. In order to promote the codebook performance of vector quantization, this paper proposes an algorithm of Recognition Distance SOFM based on the basic self-organizing feature mapping (SOFM). At the same time, the algorithm is applicable in wavelet transform of origin images. Simulation shows that the computation is substantially reduced and the codebook performance obviously improved.
出处
《通信技术》
2009年第3期233-235,共3页
Communications Technology
关键词
自组织特征映射
神经网络
矢量量化
self-organizing feature maDDing
neural network
vector quantization