基于亚像素边缘检测的LED芯片定位算法研究  

Research on Positioning Algorithm of LED Chips Based on Sub-pixel Edge Detection

作　　者：李德龙[1] 龚时华[1] 王子悦[1] 

机构地区：[1]华中科技大学机械科学与工程学院,武汉430073

出　　处：《半导体光电》2017年第3期369-374,共6页Semiconductor Optoelectronics

基　　金：国家自然科学基金项目(51375192);湖北省科技支撑计划项目(2014BAA009)

摘　　要：LED芯片的定位精度直接决定了LED制造装备的生产质量和效率,为了提高LED芯片的定位精度,提出了一种基于亚像素边缘检测的芯片定位算法。该算法首先采用Gamma变换方法增强图像的对比度,并利用Blob算法获取芯片有效区域;接着采用Canny算法进行芯片亚像素边缘轮廓的提取;最后,通过拟合芯片边缘轮廓,获取芯片位置中心,完成芯片位置的精确识别。该算法不需要人工训练模板进行匹配,提高了边缘提取的定位精度,实验表明,该算法能在平均4ms内完成一颗芯片的识别,且重复精度达到0.1pixel,满足LED芯片高速高精度定位需求。The position of LED chips directly determines the accuracy of LED detection equipment.To improve the accuracy of the chip positioning,and to minimize the effect of template matching,the sub-pixel positioning algorithm based on the edge extraction was put forward in this paper.Firstly,the algorithm uses the gamma transform to enhance the contrast between the chip and the background area.Then,it uses Blob algorithm to get the chip area of interest.Finally,the Canny edge algorithm is used to extract the sub-pixel edge profile and the chip edge profile was fitted to locate the chip position.The proposed algorithm does not need artificial training to match template,so the position accuracy of the edge extraction can be improved.Through actual measurement and test,the results show that the image repeatability positioning precision is up to 0.1pixel,and the high-speed and high-precision positioning demands for LED chip has been realized.

关 键 词：LED芯片 图像分割 CANNY算法 亚像素 边缘提取 

分 类 号：TN312.8[电子电信—物理电子学]