高斯混合模型的光学元件表面疵病边缘定位

Edge location of optical element surface defects based on gaussian mixture model

为抑制光照突变对边缘定位的影响,提升疵病边缘定位效果,提出高斯混合模型的光学元件表面疵病边缘定位方法。引入光照变化因子,通过颜色直方图匹配算法自适应更新高斯混合模型参数;利用参数更新后的高斯混合模型提取光学元件表面疵病背景图像,通过差运算原始光学元件表面疵病图像与疵病背景图像,得到疵病前景图像;利用Canny边缘算子提取光学元件表面疵病的边缘前景图像;加权融合边缘前景图像与前景图像,得到光学元件表面疵病边缘定位结果。实验结果表明,该方法可有效提取光学元件表面疵病前景图像与边缘前景图像,得到疵病边缘定位结果;在不同光照变化情况下,不同疵病宽度时,该方法疵病边缘定位的边缘连接完整度均较佳;该方法定位不同类型疵病边缘时的灰度差分乘积均较高,即边缘定位清晰度较佳,定位的边缘较明显。

In order to restrain the influence of light mutation on edge location and improve the effect of defect edge location,a method of defect edge location on optical element surface based on Gaussian mixture model is proposed.The illumination change factor is introduced,and the parameters of the Gaussian mixture model are updated adaptively through the color histogram matching algorithm;The Gaussian mixture model with updated parameters is used to extract the background image of optical element surface defects.The original optical element surface defect image and the de-fect background image are obtained by difference operation;Canny edge operator is used to extract the edge foreground image of optical element surface defects;Weighted fusion of edge foreground image and foreground image can obtain the edge location result of optical element surface defects.The experimental results show that this method can effective-ly extract the defect foreground image and edge foreground image of optical element surface,and obtain the defect edge location result;Under different lighting conditions and different defect widths,the edge connection integrity of the de-fect edge location method is better;The gray difference product of this method is higher when locating the edges of dif-ferent types of defects,that is,the edge location is clearer and the edge is more obvious.