Real_Time Multiple Human Compressive Tracking

Multiple object real-time tracking is a hot research field of computer vision.It has a wide application in the field of intelligent transportation and intelligent monitoring.However,a tracking system with high stability and robustness is very challenging to the scientific research and engineering applications.The compressed sensing algorithm can greatly reduce the complexity of the signal,and can improve the stability and real-time performance of the system.In this paper,based on the theory of compressed sensing,a real-time tracking system with high robustness is designed.The main research contents are as follows:(1)The AdaBoost target detector based on Haar features is studied and implemented.This paper introduces the principle of image Haar feature and AdaBoost cascade classifier,and trains a target detector based on a large number of positive and negative sample of multi angle heads.(2)The Naive Bayesian target tracker based on the compressed sensing is studied and implemented.This paper introduces three key parts of compressed sensing:signal sparse representation,measurement matrix generation and signal reconstruction.Using the compressed sensing algorithm we compressed the target feature,constructed the Naive Bayesian online learning classifier,and improved the learning method of the classifier.(3)A multi-target real-time tracking system is designed and developed,which can accomplish many functions,such as automatic detection,automatic tracking and follow-up data analysis in the general scene.In this paper,the compressed sensing algorithm is applied to the target tracking system,and the target tracking system has a very high Engineering research value and social application value.