一种提高工件上角点检测准确率的方法

Method for Improving Accuracy of Corner Detection on Workpiece

摘要: 角点是对目标物体进行三维重建的基础，但传统的Harris角点检测技术在识别工件的角点过程中识别准确率低。针对这一问题，本文以液压气缸法兰板为研究对象，提出一种更高准确率的角点检测方法。首先，对图像进行边缘检测，用Canny算子结合OTSU自适应阈值来提高对工件图像边缘识别的准确度；其次，利用八邻域轮廓跟踪法提取出工件边缘轮廓；最后，检查每个工件边缘像素点的周围八个像素点，根据其分布特点，判断出真实角点位置。研究结果表明，该角点检测方法与传统的Harris角点检测方法相比，角点检测的准确率得到了明显提高。

Abstract: Corners are the basis of 3D reconstruction of target objects, but the traditional Harris corner detection technology has low recognition accuracy in the process of recognizing corners of workpieces. To solve this problem, this paper takes the flange plate of hydraulic cylinder as the research object, and proposes a corner detection method with higher accuracy. Firstly, edge detection is carried out on the image, and Canny operator combined with OTSU adaptive threshold is used to improve the accuracy of edge recognition of workpiece image. Secondly, the contour of workpiece edge is extracted by eight-neighborhood contour tracking method. Finally, eight pixels around the edge pixels of each workpiece are inspected, and the real corner positions are judged according to their distribution characteristics. The results show that compared with the traditional Harris corner detection method, the corner detection accuracy has been significantly improved.