Research On Target Tracking Algorithm Based On Compressed Sensing

The research in the field of moving target tracking is one of the hot research directions in the computer vision.It includes advanced technologies in many fields,such as artificial intelligence,pattern recognition and image processing,and it has important research significance not only in the military field but also in the civil field.In recent years,in the research field of target tracking,the compressed sensing algorithm has obtained a lot of research and application.For a large number of feature dimensionality reduction,the Bayes classifier model of binary classification is used,and the processing results show that the effect is very good.However,in the complex background environment,the robustness and real-time performance of the target tracking technology need to be improved.For the target tracking algorithm based on compressed sensing,the main research work of it includes the following parts:(1)For the core part of the compressed sensing theory,it conducts in-depth discussion and research.The application of compressed sensing technology provides the main theoretical basis for the real-time problem in the target tracking algorithm.The improved algorithm based on the real-time compression tracking algorithm also has good real-time performance in the application of compression sensing technology.(2)The algorithm mentioned in the thesis is mainly based on the dimension of haar-like feature to adaptively determine the sparsity of the matrix and the number of columns so as to select the window at the best scale as the final tracking target region.The improved algorithm has good real-time performance in target tracking.(3)When constructing Naive Bayes classifier,the classification algorithm mentioned in the thesis uses the method of setting threshold to update learning parameters,and selects the maximum probability value as the location of target features,that is,the final location of the target.The results show that the improved algorithm proposed in the thesis has a good tracking effect in target tracking and improves the problem of target drift.