The Particle Filtering Target Detection And Tracking Algorithm Based On Neural Network And Multi-Feature Fusion

As a hot spot in digital image processing and computer vision field,detection and tracking of moving objects have very important application value in the field of intelligent navigation,digital traffic,national defense and military industry.In the past several decades,many researchers have done deep research in the field of target detection and tracking,but the target detection and tracking technology can not be widely used because of the complex and varied application scenes and the complicated rules of target motion.Therefore,the design of an efficient target detection and tracking algorithm is of great significance.In order to improve the efficiency and accuracy of the target tracking detection,this paper studies the target tracking algorithm based on multi-feature particle filter on the basis of improving the target region extraction of mixed Gaussian model and improves the particle filter tracking algorithm by using BP neural network.The main algorithm improvements and achievements are as follows:A mixed Gaussian model with improved frame difference method is proposed to extract the moving target region.The mixed Gaussian model can't detect the moving target completely,and it is easy to detect the background error as the foreground.In this paper,by combining the frame difference method and the hybrid Gaussian model,the moving object region is extracted completely by distinguishing the background highlight region from the target region and using different learning rates.A particle filter tracking method based on multi-feature fusion is proposed.The tracking algorithm for single feature has low accuracy and poor robustness.In this paper,the color feature and HOG feature of the target are extracted and the multi-feature observation model is constructed.The particle filter target detection and tracking are realized by the feature model.Experiments show that the algorithm is more accurate.A new particle filter tracking algorithm based on BP neural network is proposed to improve multi-feature fusion.The traditional particle filter algorithm has the problem of particle degeneration,and the number of particles is getting scarce.In this paper,the back propagation of BP neural network is used to adjust the updated particle weights and increase the diversity of particles.By combining the multi-feature model,the filtering performance of the algorithm is improved and the precision of target tracking is improved.In this paper,the target detection and tracking algorithm of particle filter is optimized by the above three aspects,and the experimental results show that the target tracking error is reduced and the precision is improved correspondingly in complex scenes such as occlusion,background similarity,complex motion and so on.