光学元件损伤暗场图像中的目标自动提取研究

基于光学元件损伤暗场成像图像的特点以及每个像素与周边的梯度关系,选择灰度——梯度对共生矩阵为研究模型,该模型集中反映了损伤图像中两种最基本的要素——灰度信息和梯度信息(边缘)。利用该模型产生准则函数,在信息论中最大熵原理基础上,应用文中改进的二维最大熵算法能自动确定最佳分割阈值,实现了对暗场图像的分割。该算法能够准确的从光学元件损伤暗场图像中提取损伤目标,为目标识别、精确定位和损伤大小的精确计算等后续处理提供依据。为了验证该算法,试验中对多个光学元件损伤暗场图像进行了处理分析,试验表明,该算法分割得到的损伤目标与实际相符,误差在检测的范围之内,为大型光学系统中光学元件损伤的在线、自动化检测提供了一种有效的技术途径。

A study on target extraction automatically from the dark-field image of optics damage

Based on the characteristics of dark-field image,the gray information and gradient information of image are calculated for the entire image,which is collected by a digital device.So the gray level-gradient co-occurrence matrix of a dark-field image is obtained.According the matrix,a criterion function for the calculation of the threshold of a dark-field image is generated.Applied the maximum entropy of information theory and combined the criterion function with an improved 2-D maximum entropy algorithm,the threshold,which will be used to extract the damage information from its dark-field image,can be calculated accurately and automatically.This achieved segmentation for dark-field image based on the algorithm will make the damage recognition,the precise location and the size calculation of the damage more easily.In order to validate the algorithm, many dark-field images of several optic components have been analyzed.The experimental result shows that the damage targets extracted with this algorithm is accord with the actual damage distribution,and the error is in the scope of permission.The algorithm provides a useful approach for online detection and automatization detection of damage in optics.