Research On Object Tracking Method And Application For Ios Platform

The problem of moving objecttracking is a very important research topic in the field of computer vision.It attracts many scholars,researchers and business people at home and abroad to carry out research on relevant issues,and has achieved very remarkable results,successfully applied to intelligent video Monitoring,intelligent human-computer interaction,intelligent traffic monitoring,image retrieval and military applications.Although the problem of target tracking has been developed very rapidly in a large number of related research work,the existence of factors such as change of illumination,posture transformation,occlusion and background disturbance still shows that there is still a lot of research space for object tracking related problems.In recent years,the development of mobile devices such as mobile phones is quite rapid,people use the phone even more time than the use of computers,mobile devices to achieve the target tracking is very meaningful.But because of mobile devices and PC performance is still a certain gap,the use of mobile phone camera real-time target tracking will encounter some new difficulties and challenges.The main work of this paper is as follows:(1)By analyzing the shortcomings of the current keypoints-based moving object tracking algorithm,a new adaptive feature point clustering tracking method is proposed.Firstly,the algorithm models the appearance of the target by detecting the FAST keypoints of the target and calculating the color histogram of the target area.Secondly,the active feature points of the current frame are calculated by the optical flow method,and the feature points of the outliers are excluded by the adaptive clustering method proposed in this paper.Finally,the center point position of the current frame tracking target and the bounding box of the target are estimated by the correct keypoints?(2)A new object tracking template update strategy is proposed.According to the current frame tracking results to calculate the target area of the color histogram,by comparing with the color histogram template to obtain the current frame tracking confidence.If the confidence level is high,the color histogram model is updated and local keypoints matching is performed.If the confidence is low,the color appearance model is not updated and the global keypoints matching is performed.In addition,according to the number of current active keypoints,different strategies are selected to update the keypoints set,and the computational cost is reduced and the computational efficiency is improved while ensuring the robustness of tracking.Finally,the validity and robustness of the proposed algorithm are verified by qualitative analysis and quantitative analysis by comparing with a variety of classical tracking algorithms in different challenging video sequences.(3)Designed a moving target tracking system based on iOS platform.The system implements the object tracking algorithm proposed in this paper on the iOS platform,and can track the target in real time through the mobile phone camera.This paper proves the robustness and effectiveness of the proposed algorithm in response to various tracking challenges when real-time tracking of mobile phone cameras.In addition,the impact of parameter setting in the real-time tracking process of the camera is analyzed,and the corresponding parameter adjustment when dealing with different challenges can effectively improve the tracking accuracy and robustness.