Research Of The Camera Calibration Algorithm Based On Subpixel Level

Vision is important for people to gain information and see the world.With the development of electronic science technology and vision theory,computer vision is used more widely in biomedical engineering,industrial detection,military target tracking,autonomous navigation and also other fields.Camera calibration has become a hot research field as a key part of getting three-dimensional space:information through computer vision.Camera calibration is to determine the correlation relationship between object coordinate in three-dimensio:nal space and its pixel coordinate,which also means to gain internal and external parameters according to the given camera model.The precision of camera calibration directly influences the precision of the whole vision system.So,the research of calibration algorithm has the vital significance.This thesis set improving the precision of camera calibration as a goal.The main research of this thesis is subpixel level corner extraction and the related theory and algorithm of camera calibration.Some improved algorithm is proposed,the main work can be concluded as below:The camera calibration is realized by using plane template in this thesis.Because the corner extraction is the basic of camera calibration,this thesis makes a research on the ways of corner extraction in the calibration template firstly.A subpixel level of the improved Harris corner detection based on the symmetry feature is proposed after comparing and analyzing the original corner detection algorithm.This proposed algorithm get candidate corner coordinate using Harris algorithm firstly.Then,detect the true corner from the candidate corner by using the symmetry feature of the checkerboard corner.Finally,gain the subpixel coordinate through the property of gray gradient around the corner point.Simulation results show the precision of corner detection in the improved algorithm proposed in this thesis is increased.In order to improve the precision of camera calibration,a new algorithm of camera calibration based on ICSA-PSO is proposed.PSO(Particle Swarm Optimization)has a simple working principle and a small number of adjusted parameter,so this algorithm gains a fast calculation speed and strong commonality,While,it easily falls into local optimum.The operator designed in ICSA(Immune Clonal Selection Algorithm)has a good diversity.Combine the feature of above two intelligent algorithms,a new algorithm of ICSA-PSO camera calibration based on particle swarm optimization and immune clonal selection algorithm is proposed in this thesis.This new algorithm gains the space coordinate of the calibrated object through establishing the world coordinate system firstly.Then,the corner coordinate of image is gotten by the improved corner detection based on symmetry feature.Finally,the internal and external parameters of the camera can be obtained by the calibration algorithm based on ICSA-PSO.Simulation results show the effective of camera calibration algorithm based on ICSA-PSO proposed in this thesis.