摘要
用神经网络实现图像矢量量化是一种非常有效的方法。本文在分析自组织特征映射（ＳＯＦＭ）算法的基础上，提出了一种频率敏感自组织特征映射（ＦＳＯＦＭ）算法，并对网络学习训练参数的优化进行了探讨。实验表明，ＦＳＯＦＭ算法优于ＳＯＦＭ算法。
Neural network is a very efficient method for Vector Quantization(VQ).In this paper,Self-Organizing Feature Map(SOFM)algorithm has been analyzed ,Frequency Sensitive Self-Organizing Feature Maps(FSOFM)has been proposed,and the optimization of network learning parameters has been studied.Experimental results show that FSOFM algorithm is better than SOFM alogrithm.
出处
《通信学报》
EI
CSCD
北大核心
1995年第2期59-64,共6页
Journal on Communications
基金
"八五"攻关资助
关键词
矢量量化
图像编码
自组织特征映射算法
神经网络
vector quantization
image coding
self-organizing feature maps
neural network
