基于小波神经网络的图像表述  

mage Representation Based on a Wavelet Neural Network

作　　者：蔡念[1] 杨杰[1] 

机构地区：[1]上海交通大学图像处理与模式识别研究所,上海200030

出　　处：《影像技术》2006年第1期31-33,共3页Image Technology

摘　　要：小波神经网络有机地融合了小波分析的时频特性和神经网络自适应优点。本文将小波神经网络应用于图像表述,提出相应的图像表述算法。分别采用两种小波函数作为网络激励函数,以验证图像表述效果。实验结果表明,小波神经网络能够有效地表述图像,其算法具有较强的鲁棒性。Wavelet neural network organically combines the good localization characteristics of the wavelet transform theory and the adaptive learning virtue of neural networks.In this paper,a wavelet neural network is applied to image representation and the correlated algorithms are also proposed.Two kinds of wavelet functions are respectively selected as the excitation function of wavelet neural network to testify the performance of image representation.The experiment results show that the wavelet neural network can effectively represent the image and the proposed algorithms have good robustness.

关 键 词：小波神经网络 图像表述 小波函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]