Video target tracking covers a variety of interdisciplinary subjects such as pattern recognition, image processing, computer graphics and artificial intelligence. In recent years, visual tracking research methods have made significant progress, and scholars have proposed many excellent algorithms. Based on this, this paper uses the basic tracking algorithm and block orthogonal matching pursuit (BOMP) algorithm of image reconstruction, respectively, from the run time, the quality of reconstruction, reconstruction error of the two algorithms to do simulation experiments, and compare their performance, the results show that the BOMP algorithm running time is short, has extensive application, therefore, to determine the BOMP algorithm as a sparse representation model is the core of the method. Then establish the target observation model, introduce the sparse display into the particle filter framework, and update the sparse representation coefficients so that the norm reaches the optimal solution and ensure the accuracy of the tracking target. Finally, through the simulation experiment, the success rate of the target coverage is calculated. The results show that the BOMP algorithm can maintain high tracking accuracy and strong stability in the case of appearance changes caused by illumination changes, partial occlusion and attitude changes.