基于图像的TO型光器件快速定位算法  

Image-based fast localization algorithm for TO type optical devices

作　　者：巩玉国 谷玉海[1,2] 左云波 GONG Yuguo;GU Yuhai;ZUO Yunbo(Beijing Key Laboratory of Electromechanical System Measurement and Control,Beijing Information Science&Technology University,Beijing 100192,China;Beijing Kexin Mechatronic Research Institute Co.,Ltd.,Beijing 100016,China)

机构地区：[1]北京信息科技大学机电系统测控北京市重点实验室,北京100192 [2]北京科信机电技术研究所有限公司,北京100016

出　　处：《北京信息科技大学学报（自然科学版）》2023年第4期37-43,共7页Journal of Beijing Information Science and Technology University

摘　　要：提出了一种基于OpenCV图像处理的光器件快速定位算法。对获取的图像进行二值化处理,利用腐蚀和膨胀操作,剔除图像中由于光线或微小划痕等造成的微小区域;采用改进的双边滤波对图像进行去噪处理,保留图像的边缘信息,在传统双边滤波的基础上引入自适应的参数,提高了区域边界的轮廓清晰度;利用Laplacian算子提取图像边缘,得到完整的轮廓,根据边缘位置信息和器件特征信息,确定光器件内微小激光器芯片的位置。改进算法可以实现同轴型光器件定位误差控制在5μm情况下,平均定位时间430 ms。A fast positioning algorithm for optical devices based on OpenCV image processing was proposed.The acquired image was binarized,and the areas with fewer pixels caused by light or small scratches in the image were removed using corrosion and dilation operations.The improved bilateral filter was used to denoise the image and retain the edge information of the image.Compared with the traditional bilateral filter,the adaptive parameters were introduced to improve the contour definition of the region boundary.The Laplacian operator was then used to extract image edges and obtain complete contours,and the position of the micro laser chip inside the optical device was determined based on edge position information and device feature information.The improved algorithm can achieve an average positioning time of 430 ms to control the positioning error of TO type optical devices under 5μm.

关 键 词：双边滤波 边缘检测 图像定位 最小二乘法 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]