Research On The Detection Technology Of The Pose For Nuclear Fuel Ball Based On Machine Vision

With the rapid development of China's nuclear industry,the demand for the nuclear fuel ball elements for high temperature gas-cooled reactor(HTGR)is gradually increasing.In order to accelerate the construction of HTGR nuclear power plants in China and promote the formation of the technical system of nuclear power design,manufacture,construction and operation in China,it is urgent to realize the intelligent production of nuclear fuel ball components.Pose detection of nuclear fuel balls is a necessary step in the production of nuclear fuel balls.However,most of the current pose detection is done by manual,and the accuracy and efficiency are difficult to guarantee,which can not meet the needs of automatic production of nuclear fuel balls.Based on this,this paper studies the technology of nuclear fuel ball pose detection based on machine vision,which has certain theoretical significance and application value for improving the efficiency and accuracy of ball pose detection and realizing the automatic production of nuclear fuel ball.With the support of the Ministry of Science and Technology's Innovative Method Project(2016IM0300)and the Ministry of Information Technology's Intelligent Manufacturing Standardization and New Mode Application(2016ZXFM02016),this paper studies the technology of nuclear fuel ball pose detection based on machine vision,proposes a method of feature recognition and ball pose detection.A nuclear fuel ball pose detection system based on machine vision is expected to improve the production efficiency of nuclear fuel ball.The main research work of this paper is as follows:(1)Based on the analysis of the basic principle of camera imaging,the advantages and disadvantages of the camera imaging model are summarized.The common coordinate system,the conversion relationship between coordinate systems and the camera calibration parameters in the camera calibration process are elaborated in detail.Then the causes and expression methods of the non-linear distortion in the imaging process are analyzed,and Zhang Zhengyou's calibration method is explained in detail.Finally based on the principle of selecting industrial camera,lens and light source,the hardware scheme of ball pose detection system is established.(2)To solve the problem of incomplete arc segmentation and inaccurate ellipse recognition,an image segmentation algorithm based on morphology and Otsu threshold method and an ellipse fitting algorithm based on prior information are proposed.Firstly,morphological structural elements are designed to perform open-close operation on the sphere image,and then the object clustering method based on prior knowledge is introduced to segment the sphere arc region.After segmentation and arc feature extraction,the ellipse is identified by ellipse fitting algorithm based on prior information,and the experiment is carried out by using the collected ball image.The experimental results show that the proposed image segmentation algorithm can effectively segment the arc region on the ball,and its segmentation effect is better than the traditional Otsu threshold algorithm.The proposed ellipse fitting algorithm based on prior information has higher accuracy and better robustness.(3)A ball pose detection method based on perspective distortion correction is proposed.Firstly,an image perspective distortion correction model based on inverse perspective transform is proposed for sphere features.Then,a ball pose detection method based on circular feature parameters is proposed,and an experimental verification scheme for ball pose detection is designed.The experimental results show that the proposed ball pose detection method based on perspective distortion correction can effectively eliminate the influence of perspective distortion on ball image,and make the ball pose detection more accurate.(4)A ball pose detection software based on machine vision is designed and developed.The system uses LabVIEW software to collect the image of calibration plate or sphere,and then calls the camera calibration toolbox in MATLAB to calibrate the camera.At last,it uses the function of MATLAB to extract image features and detect the sphere pose.The test results are displayed on the system interface,which realizes the automatic detection of the nuclear fuel ball pose.