Improved Target Tracking Method And System Design

This thesis expounds the research background and research significance of target tracking technology and describes the principle of Compressive Tracking algorithm and Camshift tracking algorithm respectively,and analyzes their tracking performance.The traditional Compressive Tracking algorithm only deals with the gray feature of the target region,so the robustness and accuracy of the algorithm still have a lifting space.Camshift tracking algorithm has high tracking accuracy under small illumination changes,since this algorithm mainly tracks the color characteristics of the target area,the change of the hue drift or illumination variation will have a serious impact on the tracking performance of the algorithm.Firstly,aiming at the defect of Compressive tracking algorithm,an improved Compressive Tracking method is proposed,which weighted the gray and differential features of the target area and used the multi-feature to describe the target area,to a certain extent,this method makes up for the defect of tracking performance instability caused by the single characteristic of the target area.Secondly,aiming at the influence of illumination variation and the hue drift on Camshift tracking algorithm,a Camshift Tracking method based on fuzzy histogram model is proposed.The probability projection image of the tracking scene image can be obtained by the back projection of the fuzzy histogram model,tracking with this probability projection image,can reduce the sensibility of the target model to the different dividing values of the hue rank.Through a series of simulations,it is proved that the two improved tracking methods proposed in this thesis can improve the tracking accuracy and robustness,and meets the real-time tracking system requirements.Finally,the hardware design of the intelligent tracking vehicle based on ROS system is introduced,and the improved tracking method is applied to tracking experiments of the intelligent tracking vehicle,the experimental results show that the improved tracking method has better target tracking effect.