基于FLICM与几何特征组合的缺陷识别算法  

Defect Target Recognition Algorithm Based on FLICM and Geometric Features

作　　者：王忠华[1,2] 黄发亮[2] 池桂英 WANG Zhong-hua HUANG Fa-liang CHI Gui-ying(Key Laboratory of Nondestructive Testing （Miniztry of Education）, Nanehang Hangkong University, Nanchang 330063, China School of lnformation Engineering, Nanehang Hangkong University, Nanehang 330063, China)

机构地区：[1]无损检测技术教育部重点实验室(南昌航空大学),南昌330063 [2]南昌航空大学信息工程学院,南昌330063

出　　处：《南昌航空大学学报（自然科学版）》2017年第2期75-83,共9页Journal of Nanchang Hangkong University(Natural Sciences)

基　　金：国家自然科学基金(61362036);无损检测技术教育部重点实验室开放基金(ZD201429005);江西省自然科学基金(20132BAB201024)

摘　　要：针对边缘模糊、对比度低的缺陷图像,采用传统模糊聚类方法容易引起目标聚类错误,阻碍了缺陷特征参数的提取精度,从而引发了缺陷错误分类和降低了缺陷识别率。为此,本文提出了一种新颖的基于FLICM与几何特征组合的图像缺陷识别算法。首先采用FLICM模型分割缺陷图像,以获取图像缺陷区域;其次提取图像缺陷的多类几何特征;最后采用几何特征的参数变量组合识别缺陷目标。实验结果表明该模型提高了疏松、夹杂、裂纹、分层、窜层、气孔缺陷识别率。In view of defect image of edge blurring and low contrast,the traditional fuzzy clustering method is easy to cause the target clustering errors and hinder the extraction accuracies of defect feature parameters,which lead to the defect misclassification and reduce the defect recognition rate. Therefore,a novel image defect recognition algorithm based on FLICM and geometric feature is presented in this paper. Firstly,the FLICM model is used to segment the defect image to acquire the image defect areas. Secondly,the multiple geometric features of image defect are extracted. Finally,the parameter variable combinations of geometric features are designed to recognize the defect target. The experimental results show that our model improves the recognition rates of the shrinkage,inclusion,crack,delamination,channel and blowhole defects.

关 键 词：模糊C均值 几何特征 特征组合 缺陷分类 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]