Objects Detection And Tracking In Video Survillance

The research of moving target detection and tracking is an important branch of thecomputer vision in video surveillance. The research involves deformations includingrotation, variation of appearance, etc. It has an important significance for the daily lifesecurity and behavior analysis in late advanced image processing. Taking moving targetsfor example, this text researches the situation of objects motion based on the colorcharacteristics, texture feature and SIFT local descriptor.First of all, this text analyses the tracking algorithm for the moving targets deeply inorder to improve existing tracking algorithm effectively.Then, the algorithm tracks targets of the moving. First, the algorithm uses the methodof the combination of color and texture feature which combine the two features’advantages. LBP is a texture descriptor which is not sensitive to color changes withdetailed description and better local description. And using the combined feature tracks thetargets which.include color feature and the block LBP texture descriptor.Finally, in order to increase the local position information, the algorithm extractslocal feature using the method of the combination of color and SIFT local descriptor basedLLC coding encoding and particle filter predicting. Extracting local descriptors andencoding it by LLC coding which can get a low-dimensional description. Using theparticle filter tracks targets of the moving by predicting and tracking the low-dimensionaldescription.