用于分割双纹理图像的最佳单Gabor小波参数的反馈系统  

A Feedback System of an Optimal Single Gabor Wavelet Parameter for Bi-textured Image Segmentation

作　　者：朱程辉[1] 陈昕[1] 方敏[1] 

机构地区：[1]合肥工业大学电气与自动化工程学院,合肥230009

出　　处：《计算机工程》2003年第11期60-61,122,共3页Computer Engineering

基　　金：安徽省自然科学基金资助项目(01042310)

摘　　要：设计了一个对分割双纹理图像的最佳单Gabor小波参数进行自适应调整的反馈系 统?该系统利用Gabor小波的带通特性使某一频率的纹理通过,而另一频率的纹理受到抑制 ,由RBF神经网络对预处理后的图像参数进行估计,求出最优分割阈值?以分割阈值误差率 为反馈量对Gabor小波参数进行自适应调整,从而得到最佳Gabor小波参数?实验证明,该系 统对双纹理图像的分割达到了预期的结果?In this paper, a feedback system of automatically adjusting an optim al single Gabor wavelet parameter is designed, which aims to segment a bi-textur ed image. The system makes use of the band-pass character of Gabor wavelets, a f requency is passed and the other is inhibited. The parameters of the pretreated image are estimated by the RBF neural network and the optimum threshold is compu ted. The Gabor wavelet parameter is adjusted by the term of difference of each s egmentation threshold. As a result, the optimum Gabor wavelet parameter is achie ved. The system segments bi-textured image to achieve expectant results, which i s testified by the experiments.

关 键 词：纹理 分割 GABOR小波 RBF神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]