Research On Binocular Vision Detection Of Probe Position In Tool Detection System

The main purpose of the tool inspection system is to measure geometric parameters of the tool in the progress of design,manufacture and re-grinding of the tool to ensure their quality.At the same time,the geometry and accuracy of the tool cutting part during the to tooling Measurements are used to ensure tool cutting performance and product quality.In this paper,the binocular stereoscopic visual position detection is used to detect the position of the scanning probe in the tool detection system,so as to realize the non-contact and on-line measurement.First of all,based on the basic theory of computer vision,this paper analyzes the binocular vision model and the measurement system,and elaborates the principle of binocular vision measurement in detail.Secondly,the camera calibration method and calibration process are analyzed in detail,and the relationship among image coordinates,camera coordinates,the conversion between world coordinates and causes of camera distortion are analyzed.The calibration principle of the plane calibration method is mainly studied.Based on the initial value of this calibration,the improved genetic algorithm is used to calibrate the camera,and experimental results show that the method has high calibration accuracy.Thirdly,the feature extraction process is studied to have a research on the feature extraction principle,and the classical algorithm is analyzed.On the basis of studying the classical Harris corner extraction,slow extraction speed and the determination of pixel level extraction are improved.In this paper we improve the computational speed and subpixel extraction by combining the adjacent pixels and the difference method,and the effectiveness of the proposed algorithm is proved by experiments.Finally,the principle of pose solution is analyzed in detail,and the position of the probe is solved by the parallax principle.The spatial posture of the probe is solved by analyzing the spatial geometry.Experimental results show that the proposed method can achieve the expected effect and complete the pose detection.