Research On Target Tracking Algorithm Based On Compressive Sensing

Target tracking is a rigid demand in modern defense and civilian fields.Nowadays in the target tracking will be faced with huge data,Nyquist speed detection and processing has become impractical,the use of low sampling rate processing of the theory of compressed sensing(CS)provides signal processing New ideas.When the signal is sparse in a base or dictionary,sub-Nyquist rate acquisition is possible.By studying compressed sensing theory,target tracking combined with compressed sensing can effectively improve the tracking real-time performance and improve the robustness of the algorithm.Study the basic theory of compressed sensing.Aiming at the long time of image restoration in high-precision reconstruction and tracking of radar images,a computationally efficient radar image reconstruction algorithm based on compressed sensing is proposed.The algorithm is applied to high-resolution radar imaging and tracking,and theoretically deriving execution matrix vectors.Complexity will be significantly reduced when multiplied.Experiments show that the proposed algorithm has a fast image recovery speed and is used to improve the real-time performance during tracking.In order to solve the poor performance of the existing advanced compression tracking(CT),fast compression tracking(FCT),fuzzy tracking and other issues,research has improved the real-time compression tracker and the fast compression tracker.The proposed method introduces template matching and uses weighted multiframes to collect more available data to make more informed tracking decisions,instead of purely using the highest classifier output,and uses similarity measures to normalize cross-correlations during the tracking process.(NCC)and cosine similarity(CS).The experiment compares the performance on 18 commonly used publicly available image sequences.The proposed algorithm deals with issues such as fuzzy tracking and shows excellent overall performance,while maintaining a high frame rate.The compressive sensing and processing(CSP)method greatly compresses the acquired signal and brings a significant improvement in real-time performance when used in tracking.However,the output signal-to-noise ratio(SNR)is greatly reduced,and the tracking accuracy cannot be guaranteed.In addition,compression sensing andprocessing techniques are currently applied to single-target tracking.A radar multi-target tracking algorithm based on adaptive compressed sensing and processing is proposed.The proposed Adaptive Compressed Sensing and Processing(ACSP)method not only improves the signal-to-noise ratio and reduces the sidelobes of the ambiguity function,but also improves the tracking performance.Particle filter(PF)and joint data association(JPDA)are studied,and particle filters combined with joint data association are used for sequence estimation of target states,thus achieving multi-target tracking.Experiments show that compared with the original algorithm,the proposed ACSP method improves the output SNR of the signal and achieves multi-target tracking.