Research On Multi-sources Data 3D Reconstruction Technology For Space Targets

The three-dimensional reconstruction technology of space targets is a key technology for realizing space security and service automation.This paper conducts research and discussion on the multi-source 3D reconstruction technology of gray image sequences and depth image sequences,and provides a set of full-link solution methods including target characteristics analysis,system parameters design,geometric calibration,reconstruction strategy,and reconstruction evaluation.Through the study of key technologies in the program,the problems of the three-dimensional reconstruction of space target are analyzed and relative solutions are proposed.In order to study the imaging properties of space targets,the space targets are modeled and analyzed.This paper studies the reflection characteristics of satellites based on the bidirectional reflection distribution function.Simulation shows that the intensity of the satellite reflection direction is distributed in a certain direction,and it changes significantly with the target material differences.The imaging performance is calculated based on the visible light imaging system and the target modeling.The imaging signal to noise ratio is about 50 dB,and good imaging quality can be obtained.A geometric calibration study of a multi-source data imaging system is carried out.First,the visible light imaging system is calibrated,and image segmentation is used to eliminate the effect of non-calibrating plate position noise.A depth camera model is established for depth imaging system.In order to reduce the influence of infrared projection on camera calibration in the calibration process based on depth model,a calibration method based on image enhancement is proposed,and the camera system parameters are obtained.In the space targets 3D reconstruction based on image sequence,the spatial target often appears oversaturated in imaging and the imaging dynamic range needs to be expanded.This paper proposes a multi-exposure fusion method based on bilateral filtering and information entropy combination.The fusion method improves the corresponding dynamic range of imaging and the feature information of the image are enhanced.The fusion method is introduced into the three-dimensional reconstruction,which significantly improves the number of point cloud.Aiming at the symmetry of the space target,an optimization strategy is proposed.First,the image sequence is sorted in order,the matching between the images is limited within the symmetry range,and the matching field range is reasonably adjusted according to experimental values.At the same time,during image matching,zoom in or zoom out the matching threshold according to the adjacent distance.The proposed optimization strategy effectively suppresses position and pose calculation errors caused by symmetry-induced,and can correctly reconstruct the target.The quality of the reconstructed model is evaluated.This paper proposes an evaluation method for 3D reconstruction,uses 3D design and printing to obtain the actual physical model and standard data model for comparison,and compares the surface sampling of the reconstructed model with the standard model.The evaluation parameters are obtained,and the proposed evaluation method effectively avoids the subjectivity of visual evaluation and the dependence on highprecision scanning instruments.In the three-dimensional reconstruction of spatial target based on depth image,the non-uniformity of the flat region of the depth image is easily ignored.In this paper,an average correction method is used to correct the image,and the image correction is introduced into the depth 3D reconstruction process.In this paper,two generations of Kinect V1 and V2 are used as the representative of the structured light and time of flight(DOF)depth imaging modes.The adaptive analysis of the three-dimensional reconstruction of the space target for both methods are carried out,the results show that,as for target edge noise,the structure-based light is better than TOF,and a good reconstruction effect is obtained.Based on the structured light,the engineering parameters are verified,and the boundary conditions for the three-dimensional reconstruction of the depth image under certain conditions are obtained.