Research On Algorithm Of Detection And Tracking Recognition In Intelligent Video Surveillance

The intelligent video surveillance processes the video data of the monitoring area to obtain the motion information of the object in the area,wherein the moving object detection algorithm,the target tracking algorithm and the target recognition algorithm are the core of the intelligent video monitoring.Video data has characteristics of large amount of information,short timeliness,high complexity and strong correlation.It puts forward higher requirements for video data processing algorithms.The main research work of this paper is as follows:Aiming at the problem of "cavitation" in the inter-frame difference algorithm and the poor anti-interference performance of the background difference algorithm,this paper presents an improved algorithm based on single Gaussian background modeling and codebook background modeling.The experimental results are used to analyze the motion detection algorithm of the improved single Gaussian background model and the motion detection algorithm of the codebook background model.The feasibility of the algorithm is verified.The influence of noise such as illumination in the scene and the mutual occlusion of the target deformation and the target increase the difficulty of tracking the moving target.Aiming at the particle weight degradation problem and the lack of diversity in the particle filter target tracking algorithm,an improved method based on fireworks algorithm is presented.The fireworks algorithm is used to optimize the distribution of particles to ensure that most of the particles can be in the high likelihood region and have good diversity.The improved particle filter algorithm is verified by experiments.The results show that the accuracy of the target tracking algorithm is improved.Target recognition is the distinction between a particular target or a type of target from other targets or other types of targets.In order to solve the problem that the kernel function coefficients of multi-core support vector machines are difficult to determine,a multi-class single-core support vector machine fitting algorithm is proposed.By adjusting the training samples of each type of single-core support vector machine,the single-core support vector machine is improved.For the recognition of specific targets,multi-class single-core support vector machines are used to identify the target and make overall decisions for the recognition results.It can be seen from the simulation experiment that the support vector machine fitting algorithm improves the classification ability of the target sample.