| Information extraction is a very important branch of image processing.Moving object information extraction from remote sensing satellite video data is one of the research hotspots in the field of image processing.This paper mainly studies the application of super-resolution reconstruction technology to the motion information of vehicle targets on the road.Due to the large number of vehicles on the road and the high complexity of the video satellite image background,it is difficult to extract vehicle target information from satellite video data with weak visibility and the accuracy of vehicle moving target detection needs to be improved.Aiming at the problem of poor visibility of vehicle targets,the single frame image of satellite video data is preprocessed by using the iterative back projection algorithm AF-IBP(Iterative Back Projection algorithm combined with Anisotropy Filtering)to increase the visibility of the vehicle.In order to improve the detection accuracy of moving objects in traditional satellite video moving object detection algorithms,a DF-GMM(Double three-Frame combined with Gaussian Mixture Model)detection algorithm based on frame difference method and mixed Gaussian background modeling is proposed.Finally,through experiments,the effectiveness of the algorithm is verified,and the research results in this paper are applied to practical projects.The main work of this paper is as follows:(1)Taking remote sensing satellite video data as the breakthrough point,the super-resolution image reconstruction technology is studied to detect vehicle moving target information.The two problems of weak visibility of vehicle target and low accuracy of vehicle moving target detection in video data are studied.This paper summarizes the current situation of video satellites,the current research situation of target detection and the applicable fields of satellite video data,and summarizes and combs the basic theoretical knowledge such as relevant technologies and algorithms involved in the two problems to be solved in this paper.(2)An improved algorithm AF-IBP based on reverse iterative projection algorithm is proposed to solve the problem of weak visibility of moving targets in satellite video data.In order to enhance the visibility of the vehicle target,an improvement is made based on the back iterative projection algorithm and anisotropic filtering idea.Aiming at the sawtooth effect in the iterative process,an anisotropic filtering operator is proposed to be added,a smoothing factor adjustment is added to alleviate the sawtooth effect,an adaptive coefficient adjustment iteration error is added,and the improved algorithm is used to carry out super-resolution image enhancement processing on a single frame image of satellite video data,and the effectiveness of AF-IBP algorithm is verified through experiments.(3)Aiming at the problem of low detection accuracy of typical algorithms for moving object detection in satellite video data,a DF-GMM detection algorithm based on frame difference method and Gaussian mixture background modeling is proposed to improve the accuracy of vehicle moving object detection on satellite video data.Based on the three-frame difference method and the mixed Gaussian background modeling algorithm,the DF-GMM detection algorithm proposed in this paper updates and improves the parameters of the mixed Gaussian background modeling algorithm,and gets the vehicle detection target by algorithm fusion with the double three-frame difference method.The DF-GMM algorithm is verified and analyzed on satellite video data,which proves that the proposed algorithm can improve the detection accuracy of vehicle targets.(4)The research results of this paper and the proposed algorithm are applied to the relevant modules of the research and development project of super-resolution reconstruction system that I participated in,and the effectiveness of the algorithm is verified by the actual data provided by the project team. |
PDF Full Text Request