基于主成分分析的模糊支持向量机焊接图像分割
|
摘要
焊接是钢结构件生产的主要方法,射线检测法是焊接缺陷的重要检查方法之一,针对焊缝缺陷图像目标边界模糊、灰度不均匀及强噪声的特征,笔者提出了基于主成分分析的模糊支持向量机方法分割焊接缺陷图像。首先,利用主成分分析法降低模糊支持向量机特征向量的维数,去除次要特征分量对支持向量的影响,提高支持向量机的分类速度和精度;然后针对焊接缺陷图像的特征,提出了以3×3窗口为单元的分割算法,将模糊支持向量机引入该系统,进一步降低了噪声对构建最优分类器的不良影响。试验结果表明,对于焊缝缺陷图像,基于主成分分析的模糊支持向量机可以取得较好的分割效果。
Welding is a main way of producing steel structural parts, and X-ray examination is one of the major techniques for detection of welding defects. In view of characters of the welding defect image such as dim target boundary, uneven gray scale and strong noise, a method of segmenting the welding defect image by fuzzy support vector machine based on principal component analysis is put forward. Firstly, the principal component analysis method is used to lessen the dimension of the characteristic vector of the fuzzy support vector machine, and eliminate the influence of subordinate characteristic vector on support components, so as to enhance the classification velocity and precision of the support vector machine. Secondly, according to characters of the welding defect image, a kind of segmentation algorithm based on window of 3×3 serving as the unit is proposed. Finally, the fuzzy support vector machine is introduced into the system, so as to weaken the ill effect of noise on constitution of optimal classifier. Experimental results show that employment of the fuzzy support vector machine obtains excellent results in segmenting the welding defect image based on principal component analysis.
Welding is a main way of producing steel structural parts, and X-ray examination is one of the major techniques for detection of welding defects. In view of characters of the welding defect image such as dim target boundary, uneven gray scale and strong noise, a method of segmenting the welding defect image by fuzzy support vector machine based on principal component analysis is put forward. Firstly, the principal component analysis method is used to lessen the dimension of the characteristic vector of the fuzzy support vector machine, and eliminate the influence of subordinate characteristic vector on support components, so as to enhance the classification velocity and precision of the support vector machine. Secondly, according to characters of the welding defect image, a kind of segmentation algorithm based on window of 3×3 serving as the unit is proposed. Finally, the fuzzy support vector machine is introduced into the system, so as to weaken the ill effect of noise on constitution of optimal classifier. Experimental results show that employment of the fuzzy support vector machine obtains excellent results in segmenting the welding defect image based on principal component analysis.
引文
[1]张晓光,高顶.射线检测焊接缺陷的提取和自动识别[M].北京:国防工业出版社,2004:152-224.
[2]CortesC.VapnikV.Support vector networks[J].MachineLearn-ing,1995,20(3):273-297
[3]Scholkopf B,Smola A,Wlilliamson R:C.New support vector algorithms[J].Neural Computations,2000,12(5):1207-1245.
[4]徐海祥,夏学知,陈 炜,等.基于支持向量机的噪声图像分割[J].微电子学与计算机,2007,24(11):14-20.
[5]徐海祥,喻莉,朱光喜,等.基于支持向量机的磁共振脑组织图像分割[J].中国图像图形学报,2005,10(10):1275-1280.
[6]朱艳伟.基于主成分分析的支持向量机划分地震相[D].西安:西安科技大学,2006.
[7]李金皎,王爱侠,王娇.模式识别[M].北京:电子工业出版社,2010:104-146.
[8]杨志名.模糊支持向量机及其应用研究[D].山东:中国农业大学,2005.
[9]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-42.
[10]Yan C,Sang N,Zhang T.Local entropy-based transition region ectraction and thresholding[J].Patter Recognition Letters,2003,24:2935-2941.
[2]CortesC.VapnikV.Support vector networks[J].MachineLearn-ing,1995,20(3):273-297
[3]Scholkopf B,Smola A,Wlilliamson R:C.New support vector algorithms[J].Neural Computations,2000,12(5):1207-1245.
[4]徐海祥,夏学知,陈 炜,等.基于支持向量机的噪声图像分割[J].微电子学与计算机,2007,24(11):14-20.
[5]徐海祥,喻莉,朱光喜,等.基于支持向量机的磁共振脑组织图像分割[J].中国图像图形学报,2005,10(10):1275-1280.
[6]朱艳伟.基于主成分分析的支持向量机划分地震相[D].西安:西安科技大学,2006.
[7]李金皎,王爱侠,王娇.模式识别[M].北京:电子工业出版社,2010:104-146.
[8]杨志名.模糊支持向量机及其应用研究[D].山东:中国农业大学,2005.
[9]张学工.关于统计学习理论与支持向量机[J].自动化学报,2000,26(1):32-42.
[10]Yan C,Sang N,Zhang T.Local entropy-based transition region ectraction and thresholding[J].Patter Recognition Letters,2003,24:2935-2941.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心