Vehicle Dynamic Information Extraction Method Of Remote Sensing Satellite Video

The appearance of video satellite brings a new favorable opportunity for real-time observation of remote sensing,and provides a kind of new data for dynamic monitoring and target tracking.Based on the difference between target detection of remote sensing satellite video and traditional surveillance video,this paper studies the moving vehicle target detection and tracking algorithms based on satellite video sequences respectively from aspects of the detection and tracking of single target,the tracking of group targets.The specific research works in this paper are as follows:1.This paper begins with the study on the moving target detection technology based on video satellite image sequences.The paper indicated the problems about applying existing target detection algorithm directly to satellite video.A new approach which combining the background subtraction and frame subtraction technologies is proposed for better detecting the moving target on remote sensing satellite video.2.This paper turns to the study on the target tracking technology based on Mean Shift.A Mean Shift target tracking algorithm based on SSFDLC is proposed to improve the Mean Shift tracking algorithm by solving the problem of missed targets caused by error convergence.This tracking algorithm makes use of the interaction between detection and tracking to correct the tracking errors in real time and to realize robust target tracking by reducing the tracking error in complex scenes.3.This paper focuses on the study of target extraction and monitoring technology of satellite video image in urban commercial district.On the basis of vehicle target detection,the distribution of vehicle parking is analyzed,the point cluster of the parking is extracted through the DBSCAN clustering algorithm,and the central location of the parking is extracted,and the range of the parking is obtained according to the point cluster.The kernel density analysis theory is introduced into the analysis of time series remote sensing image,and the vehicle density of different locations can be extracted,and the vehicle parking density in the parking lot is analyzed.4.The validity and stability of the proposed algorithms in this paper are all verified by the contrast experiments by using UrtheCast and Skybox as experimental data,the effectiveness of the moving object detection and tracking algorithm proposed in this paper can effectively suppress the moving background edge and residual noise interference,improve the detection rate and quality,the motion object tracking algorithm to update the target information,effectively improve the traditional tracking technology,and tracking positioning more accurate.This paper has good application potential in the detection and tracking of remote sensing satellite video moving targets,the paper finally discusses and prospects.