Research And Application On The Algorithm Of Object Re-identification And Tracking Based On Video Image

Moving target tracking have played a pivotal role in field of social security, trafficmonitoring, medical examination, military defense. It is not only the core content of computervision, but also the driving force for the development of social science and technology. It owna wide range of prospects at practical and application. In the image tracking process,real-time tracking and the robustness of tracking is a basic requirement for accurate tracking.Accordingly, the specific contents as follows:The article introduces the particle filter which could effectively handle non-linear andnon-Gaussian system at first. The article made a detail introduction about the particle filtertheory. Then through the comparative experiments demonstrate the effectiveness of thealgorithm and change the variables found out the algorithm’s best track conditions.Secondly, this paper proposes a particle filter tracking algorithm based on active contourmodels. According to the active contour model get the shape characteristics of the target as atemplate matching in subsequent frames. At the same time, get the target position and obtainsome points as the source sample of particle filter then predict the target position of nextframe and match template at predicted position. When there are large deviation between theobtained target and template correct the template using active contour model.Thirdly, according to the complementary of SIFT features and clumps proposed thetarget recognition algorithm based on clumps with SIFT features. At recognition process, taketarget which extracted the features of SIFT features and clumps as template to recognise in theother frames. Meanwhile a algorithm of the template update and memory is proposed.Finally, combined the algorithm of particle filter tracking based on active contour modeland the target recognition algorithm based on clumps with SIFT features presented in thispaper. By the simulation results verify the effectiveness of the algorithm.