Design And Implementation Of Automatic Calibration Algorithm For Internal And External Parameters Of Multi Cameras Based On Natural Scene

With the development of science and technology,to assist human beings in completing specific tasks,more and more unmanned autonomous control systems have been developed such as self driving cars,mobile robots and aircraft.In these systems,multi-camera system is widely used in navigation,positioning and obstacle avoidance.In such system,the accuracy of internal parameters and external parameters often determine the accuracy of the system.In order to improve the accuracy and convenience of calibration of internal and external parameters of multi-camera system,an automatic calibration method of internal and external parameters of multi-camera based on natural scenes is presented in this paper.It can effectively improve the convenience and accuracy.This paper mainly includes the following aspects.Firstly,the special requirements of automatic calibration of internal and external parameters in natural scenes are analyzed to define the input conditions of calibration methods and the data acquisition process.Then,according to the requirements of data processing and parameter optimizing,the overall framework of automatic calibration method is given.The method is divided to point cloud reconstruction,simultaneous calibration of multi-camera internal and external parametersSecondly,the point cloud reconstruction algorithm is designed.It is based in incremental structure from motion to calculate the pose of each frame and point cloud map which are both used to calibrate the internal and external parameters.Then,the initial point cloud map construction algorithm is designed.And the optimization of point cloud and pose is built to update and the point cloud and poses iteratively.Finally,the correctness of the point cloud reconstruction algorithm is verified by experiments.Thirdly,the calibration algorithm of multi-camera internal parameters based on point cloud map is designed.Considering the special problems of internal parameter calibration in natural scenes,in order to ensure the accuracy of internal parameter calibration,the criteria for determining whether point clouds satisfy the accuracy requirements is analyzed and given.On this basis,the camera internal parameter calibration problem is transformed into a multi-parameter optimization problem,and an optimization algorithm based on Eigen is designed.Finally,the validity and accuracy of the proposed camera internal parameter optimization algorithm are verified by experimental data from natural scenes.Then,the calibration algorithm of multi-camera external parameters based on point cloud map is designed.In order to eliminate the effect of cumulative error caused by point cloud reconstruction,a preprocessing algorithm for point cloud and poses is designed.Then,the parameter optimization problem is built and calculated using the time relationship between frames.Finally,an evaluation algorithm of the calibration results is proposed to evaluate the quality of the calibration results of external parameters.Finally,the proposed method is applied to a vehicle which is mounted four fisheye cameras.Compared with the traditional method,the mileage error calculated by the camera internal and external parameters calibrated by this method is smaller,which fully illustrates the accuracy and superiority of the proposed calibration method.