Research And Application Of Multi - Target Tracking Algorithm Based On PRSF + Mean - Shift + Area Method

The non-rigid Multi-target tracking of the meteorological video images can appear the problem of tracking target deformation, merger, division, disappear and the number of target instability which bring great difficulty of the target tracking and the judgment. At present, the algorithms which can achieve the Multi-target tracking of the meteorological video images are scarce and most of them stay in the single target tracking. According to the characteristics of the meteorological image, we put the edge profile information which is extracted by PRSF integrated into the Mean-Shift tracking algorithm, it makes the judgment of target more accurate, although the tracking effect of the improved algorithm than Mean-Shift target tracking has enhanced greatly, it still can’t achieve the target tracking of great change and can’t achieve more new target identification, judgment, tracking in the next frame image. Therefore, it is difficult to realize the multi target tracking of non-rigid the meteorological video images accurately in the PRSF+Mean-Shift target tracking algorithm.On the basis of the PRSF+Mean-Shift target tracking algorithm, the thesis fused the regional area calculation of target area with PRSF+Mean-Shift target tracking algorithm to improve it, and used the meteorological video images as the research object to carry on the multiple target tracking research. Firstly, the thesis used PRSF model to extract the edge profile information from the image. Then it processed target screening and determination with area method. Finally, it used Mean-Shift target tracking algorithm which fused edge contour information to perform target tracking. Through Multi-target tracking experiment in the precipitation echo domain of the reflectivity map,it found that the improved algorithm in the thesis has strong feasibility. The target regions which occur easily deformation, merger, division in the domain of no-rigid precipitation echo can effectively track. A new target can be determined and tracked in the next frame, which solved well the problem of the target number of instability. The tracking effect is greatly progress, but also meet the real-time requirements.