Font Size: a A A

Research On The Background Change Of Moving Target Tracking System

Posted on:2017-05-21Degree:MasterType:Thesis
Country:ChinaCandidate:S T JiangFull Text:PDF
GTID:2348330485951459Subject:Control engineering
Abstract/Summary:PDF Full Text Request
Moving targets tracking technology is one of the most important research topics of research in computer vision.The applications are most value,not only in science field but also in engineering field.It combines sensor technology,image processing,pattern recognition,artificial intelligence and others advanced technologies together.These applications are mainly apply in transportion,military,enviornment,medical imaging,and many other areas.It is still the hottest research to design an robust and real-time algorithm in computer vision.This paper firstly introduce the targets tracking theories based on Kalman filters and particle filter,and research the targets tracking algorithm.which has been proved the robustness and the validity of the algorithm by the experiments of moving target tracking.Moreover,in order to improve the stability of moving target tracking algorithms in different complex environments,in particular,on the background of the real-time changed.this article conducts systematic and in-depth research on target tracking algorithm based on compressed sensing.Aiming at the problems of tracking failure and high error rates in the case of target occlusion and textures change in moving targets tracking,an improved compressive sensing algorithm is proposed.The threshold of a sigmoid function was calculated to judge whether occlusion exist.This threshold determined the classifier update occasion in order to avoid object model mistake update when heavy occlusion.To solve the tracking instability cause by single feature,a feature fusion method under rules that combined grey features and texture features was presented.Tracking was guided by the fused feature.Experiments show that the improved algorithm has17.84% higher success tracking rates,and11.59%lower average error rate compares with the traditional algorithm.Finally,the article putting forward the system of target tracking based on ARM9,and detailed the design ideal and method.It has completed thedesign of system,and the configuration of hardware and the operation system.At last,we have completed transfer of partial software.
Keywords/Search Tags:target tracking, Kalman filter, particle filter, compressive sensing, ARM9
PDF Full Text Request
Related items