计算机工程 ›› 2012, Vol. 38 ›› Issue (7): 196-197,200.doi: 10.3969/j.issn.1000-3428.2012.07.065

• 图形图像处理 • 上一篇    下一篇

一种新的金属腐蚀缺陷图像识别算法

王 彦,谢晓方,王 波   

  1. (海军航空工程学院兵器科学与技术系,山东 烟台 264001)
  • 收稿日期:2011-06-29 出版日期:2012-04-05 发布日期:2012-04-05
  • 作者简介:王 彦(1982-),男,硕士研究生,主研方向:图像处理,模式识别;谢晓方,教授、博士生导师;王 波,工程师

Novel Recognition Algorithm of Metal Corrosion Defect Image

WANG Yan, XIE Xiao-fang, WANG Bo   

  1. (Department of Ordnance Science and Technology, Naval Aeronautical Engineering University, Yantai 264001, China)
  • Received:2011-06-29 Online:2012-04-05 Published:2012-04-05

摘要: 为从大量X光图像中找出含有金属腐蚀缺陷的图像,提出一种新的X光图像识别算法。对待测的X光图像建立灰度梯度共生矩阵(GLCM),依据人们对腐蚀图像的直观印象,从GLCM中提取出5种特征参数构建特征向量。将特征向量转化为特征Vague集,并将其与目标Vague集进行相似性度量,从而实现金属腐蚀缺陷图像的识别。实验结果证明,相比BP神经网络和单Bayes分类器,该算法具有更高的识别准确率。

关键词: 金属腐蚀图像, 灰度梯度共生矩阵, 特征向量, 相似性, 模式识别

Abstract: In order to find the image containing the corrosion from lots of X-ray image, a new method of model recognition is offered in this paper. It constructes the Gray Level-gradient Co-occurrence Matrix(GLCM) for each X-ray image and the eigenvector by 5 characters of the matrix according to the feeling of the people to the corrosion. It transforms the eigenvector to the Vague sets and recognizes the corrosion by calculating the similarity between the character Vague sets and the target Vague sets. Experimental result shows that this algorithm has higher identification accuracy than BP neural network and single Bayes classifier.

Key words: metal corrosion image, Gray Level-gradient Co-occurrence Matrix(GLCM), eigenvector, similarity, pattern recognition

中图分类号: