Research On Image Feature Point Detection Algorithm For Visual Calibration

It is a huge leap in the intelligence of the robot that the vision is introduced to the robot,and the visual calibration technology of the robot is regarded as the base of the visual servo control and guidance of the robot movement and measurement work,therefore,the robot visual calibration has a crucial role in the field of the robot visual technology application.This paper mainly studies the application of camera calibration technology in visual guidance of robot vision,and the accuracy of visual guidance is directly affected by the accuracy of robot vision calibration.But the accuracy of the image feature point inspection directly affect the accuracy of the camera calibration,this paper focuses on the image feature point inspection technology for the camera calibration,the work of this paper is shown as follows:1.For the application,an improved Harris checkerboard corner detection algorithm is proposed This algorithm integrates the corner extraction technique of Harris corner detection algorithm,and makes it more accurate and reproducible in preserving the advantages of good reproducibility and relatively high detection efficiency of Harris corner detection algorithm.At the same time,it solves the difficult problem of matching the corner coordinates and the spatial points of the grid plate image in the calculation process of the camera,and effectively solves the over-reliance on the threshold of the original Harris corner detection algorithm.2.This dissertation presents a centroid detection algorithm based on geometric symmetry and applied to elliptic array images.The algorithm limits the search range by setting the constraint condition,avoids the case of omission or mutual confusion,and affects the accuracy of elliptical circle extraction.And partly solves the difficult problem that the calibration plate mark points and image feature points are strictly matched.3.In order to reduce the requirements of robot calibration to reduce the cost of the enterprise,this dissertation wants to find a completely automatic center feature extraction algorithm.A new automatic center feature extraction algorithm is proposed to solve the problem that the accuracy of the ellipse is not high at the same time.It is difficult to accurately sort the problem.An elliptic fitting algorithm based on the least squares method is applied to the elliptic array image.The algorithm does not need the man-machine interaction to complete the detection work,and all the automatic extraction and extraction of the center feature points and the matching degree relatively high,repeatability is good.4.The experiments are done by using the above three kinds of feature point extraction algorithm and plane calibration method,the camera's internal and external parameters are calculated,the comparison analysis and calibration accuracy calculation are also done;the calibration data is written to the robot controller,and the visual capture experiment verifies the correctness of the calibration algorithm.