面向工业现场的矩形工件定位算法  

Rectangular workpiece positioning algorithm for industrial field

作　　者：朱曾生 佟强 杨大利[1] 侯凌燕[1] ZHU Cengsheng;TONG Qiang;YANG Dali;HOU Lingyan(Computer School,Beijing Information Science&Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学计算机学院,北京100192

出　　处：《北京信息科技大学学报（自然科学版）》2023年第3期66-71,共6页Journal of Beijing Information Science and Technology University

基　　金：北京市教委科研计划科技一般项目(KM202111232003)。

摘　　要：为了提高工业现场矩形工件的定位效率,提出了一种基于多边形近似的矩形检测算法,将矩形检测分为粗定位和精定位两步。在粗定位阶段,利用多边形近似算法剔除掉多余轮廓,筛选出矩形,获取矩形4个角的粗定位坐标;在精定位阶段,利用亚像素技术对矩形的4个角进行精确定位,计算出其亚像素坐标。实验结果表明:所提算法的平均检测误差为0.55像素,检测效率分别是基于Harris方法和Hough变换方法的10.7倍和7.9倍。该算法满足工业应用中的快速、高精度要求,具有一定的实用意义。In order to improve the efficiency of rectangular workpiece positioning in industrial field,a rectangle detection algorithm based on polygon approximation was proposed.The rectangle detection was divided into two steps:coarse positioning and fine positioning.In the coarse positioning stage,the polygon approximation algorithm was used to eliminate redundant contours,screen out the rectangle,and obtain the coarse positioning coordinates of the four corners of the rectangle.In the fine positioning stage,the sub-pixel technology was used to precisely locate the four corners of the rectangle and calculate the sub-pixel coordinates.The experimental results show that the average detection error of the proposed algorithm is 0.55 pixels,and the detection efficiency of the proposed algorithm is 10.7 times and 7.9 times that of the Harris method and the Hough transform method,respectively.The algorithm meets the requirements of high speed and high precision in industrial applications,and has some practical significance.

关 键 词：图像处理 矩形检测 多边形近似 亚像素 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]