增强的Fuzzy ART在苹果图像分割中的应用  

Application of improved Fuzzy ART neural network for segmentation of apple image

作　　者：武鹏[1] 

机构地区：[1]重庆大学计算机学院,重庆400044

出　　处：《微型机与应用》2010年第14期36-39,共4页Microcomputer & Its Applications

基　　金：重庆市自然基金(CSTC#2006BB2229;2008BB2194)

摘　　要：分析了Fuzzy ART神经网络的工作过程和参数对网络聚类结果的影响,提出了一种自适应确定警戒线的方法。采用基于灰度和纹理的特征提取方法,抽取组合特征向量,使用动态调整参数的Fuzzy ART网络分割图像,从而提取苹果的伤疤部分。实验结果证明,灰度特征和纹理特征相结合能有效地完成苹果表面伤疤的分割,不仅节约了时间,而且提高了网络的精度。The process and the parameters of Fuzzy ART neural network are discussed in this paper. A method of dynamically controlling vigilance parameters is presented. Anglicizing the feature of apple image, eigenvectors are extracted form the gray feature and texture feature of image. The image is segmented based on improved Fuzzy ART neural network. Finally the defect of the apple image is segmented. The results show the method can implement texture image segmentation of defective apple effectively.

关 键 词：图像分割 模糊ART Yager交集 纹理特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]