基于改进PCA算法的工件表面纹理缺陷分割
首发时间:2013-10-25
摘要:光照不均匀在机械加工零件表面缺陷视觉检测中是一大问题。为有效地提取出其表面的纹理刮伤等缺陷,本文通过对零件表面纹理分类与分割,综合比较PCA算法在纹理图像缺陷分割中的应用,在不增加算法复杂度和计算量的基础上,通过对经典PCA算法改进,减小PCA算法对光照局部不均匀的敏感性。图像处理结果表明,经过改进后的PCA算法,改善了基于直方图的阈值缺陷分割效果,有效提取出了零件表面的纹理缺陷;并且通过均方值方法、峰值信噪比、熵方法对处理过后的图像进行验证,验证结果表明改进的PCA图像的熵值增大,图像的能量分布均,同时信号量并未衰减;由此可知,通过改进PCA算法能够实现对光照不均匀零件表面质量检测的目的。
关键词: 图像处理;表面纹理;主成分分析;阈值处理;缺陷分割
For information in English, please click here
Texture image defection segmentation in the part surface based on improved PCA algorithm
Abstract:In order to realize the quality detection of part surface,the classify and segmentation of part surface texture is proposed. By means of multiple compare the apply of the PCA algorithm in texture detection segmentation, on the base of do not increase the algorithm complexity and count volume,to improve the typical PCA algorithm and decrease the sensibility of PCA algorithm to locality illumination, better the threshold defection segmentation which based on the histogram.
Keywords: image processing surface texture PCA(Principal Component Analysis) threshold processing defect segmentation
基金:
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
基于改进PCA算法的工件表面纹理缺陷分割
评论
全部评论0/1000