基于改进Rank变换和分段MER的镜片疵病检测算法

Lens Defects Detection Algorithm Based on Improved Rank Transform and Segment Minimum Enclosing Rectangle

为了解决目前光学镜片表面疵病宽度测量主要依赖人工检测而导致弯曲划痕宽度检测精度较低的问题,提出一种基于Rank变换和分段最小外接矩形(MER)的镜片表面划痕宽度测量方法。该方法首先通过改进的Rank变换算法比较像素灰度相对关系来提取疵病目标,然后计算其最小外接矩形,根据矩形长宽比初步判别疵病类型(划痕或麻点),最后将疵病分割为多个小区段,分别计算每段的最小外接矩形并获得其宽度,比较取最大宽度作为疵病宽度值。试验结果表明,该方法具有较高的可行性与精确性,能够有效提取出疵病和计算其特征参数,可以满足工程实际的需要。

Today the measurement method of defects width on optical lens surface is mainly manual detecting, which leads to low bent scratch width detection accuracy. To solve this problem, a new method to measure surface defects scratches width based on rank transform and segment minimum enclosing rectangle（MER） is proposed. This method gets target con- tour of defects by comparing pixel gray level with improved rank transform algorithm firstly. Then the MER of the defects and classify the defects types（scratch or spot） are calculated according to the length-width ratio of the MER. At last, defects are divided into multiple sections and the minimum enclosing rectangle of each section is calculated to get each width, then the maximum width is the result. Experiments verify its accuracy and the feasibility, which shows that this method can meet the requirement of engineering practice.