Research On Calibration Method Of Line-Scan System In Three-dimensional Reconstruction Use Hyperspectral Images

The detection of three-dimensional objects can only distinguish the targets from their structural features because the traditional three-dimensional information acquisition methods focuses on the acquisition of spatial information of threedimensional objects.For the targets with the same or similar spatial structure but with completely different physical and chemical properties,the traditional three-dimensional object detection methods cannot resolve the question very well.Hyperspectral imagery can provide this kind of information,so this report combines photogrammetry with hyperspectral remote sensing technologies,uses hyperspectral cameras with ultra-wide spectral band to cross-angle imaging.And obtains hundreds of spectral bands of stereo images,extracts parallax information from them for three-dimensional reconstruction.In the meantime,the high resolution spectral information can be obtained by radiation processing in the same reference system.The geometric and radiation integrated information can be extracted with high accuracy because of geometric and radiation information extraction can be carried out under the same geometric reference system.Camera calibration is an indispensable step for three-dimensional reconstruction from hyperspectral images.At present,the mainstream hyperspectral imaging system is composed of hyperspectral linear array camera and scanning platform,so the traditional frame camera imaging model and calibration method are not applicable.The problems brought by the unique imaging characteristics of linear array scanning system for camera calibration make it difficult for line-scan system to be used in threedimensional reconstruction.Based on a comprehensive summary and analysis of the calibration methods of linear array cameras at home and abroad,this paper focuses on the difficulties of applying the images acquired by line-scan system to threedimensional reconstruction.The main work of this paper and the innovations are as follows:1.According to the imaging model of linear line-scan system,the calculation process of model parameters is deduced.The accuracy of the imaging model of linear line-scan system and the validity of the optimization method for parameter solution are verified by computer simulation data.The influence of noise in calibration points,the number of abnormal calibration points and the number of calibration points used in calibration on the performance of the calibration method is also verified.2.To solve the problem of complex and unstable solution of rotating line-scan system model,a calibration method of rotating line-scan system based on virtual frame imaging is proposed.Firstly,a virtual frame imaging model is constructed.The model can be projected into a frame image based on the parameters of the rotating platform of the camera and the position relationship between the image plane and the tangent plane.Then,according to the imaging model constructed,the direct linear transformation method and nonlinear optimization method were used for camera calibration and parameter optimization.The experimental results show that the method is simple and feasible with high accuracy.3.The adjustment model of model parameters of rotating line-scan imaging system considering system is deduced,and the camera parameters calibrated based on virtual frame imaging model are used initial to calculate values.Using the real calibration data from the camera calibration site of the three-dimensional,the model is solved and the accuracy is evaluated.