Research Of Tracking Multiple Objects Methods Based On Feature Fusion And Mean Shift

The technology of tracking multiple objects in video is a core area of research in image understanding,computer vision and so on.Now it is widely used in military visual guidance,robot vision navigation,traffic control and other fields.Therefore the research of tracking multiple objects has important significance.Because the tracking methods of features rely on segmentation and the tracking methods of Mean Shift are difficult to adapt to light change,the paper has researched the methods of tracking multiple objects based on feature fusion and the Mean Shift.The paper has completed a large number of experiments and yielded the following results:1.Rapid targets' area extraction algorithm.Based on the background difference method and the frame difference method,an improved extraction method of the background difference is put forward.It can quickly and efficiently extract thebinarization foreground of the moving objects.Afterwards apply the mathematical morphology and connectivity signs to extract the targets' area accurately.2.Tracking multiple objects based on feature fusion.Traditional tracking methods based on global features can not accurately track the objects which are occluded and don't split up.A method based on feature fusion is put forward.When multiple objects occur occlusion each other,the method adopts K nearest neighborhood classifier to classify the corners in the occluding region.And this method can effectively distinct multiple targets through occlusion.In the process of extracting corners,the dual-threshold Harris corner detection algorithm is proposed.It can adaptively and accurately extract corners. Compared with the traditional algorithm,the dual-threshold Harris corner detection algorithm can avoid manually setting the detection threshold.3.Mean Shift algorithm based on moving pixels purification.Because the traditional Mean Shift tracking algorithm considers the background components,it leads to the problem that tracking window center drift off the real central.So the Mean Shift algorithm based on moving pixels purification is put forward.Compared with the traditional algorithm,the improved algorithm can effectively remove the background components of the tracking window and the iterative target tracking window center is closer to the true center.4.Based on the study of feature fusion and Mean Shift,the method of tracking multiple objects based on feature fusion and Mean Shift is put forward.Because the method combines the advantages of feature fusion tracking methods and Mean Shift algorithm,it can track multiple moving objects more robustly.