Abstract:

Currently, in surveillance video groups, traditional methods for searching camera videos involve traversing and searching through all cameras or performing repetitive searches in a network topology. These approaches result in low efficiency and poor accuracy in tracking individuals. To address this issue, we propose an efficient method for selecting surveillance camera videos based on the principles of the queuing and vertex-weighted directed graph theories. In this method, we treat cameras as vertices and construct a weighted directed graph. By calculating weights, we can determine the optimal monitoring paths considering the connections and weights between cameras. The key advantage of this method is its efficient selection of surveillance camera videos. Additionally, by combining the optimal movement paths of target passengers in urban rail transit nodes with individual tracking, we use the concept of vertex-weighted directed graphs to enhance the accuracy and efficiency of person recognition. The research results show the great significance of this method in improving the performance of surveillance systems and individual tracking capabilities. By applying the queuing and vertex-weighted directed graph theories for individual tracking, we offer an innovative approach to address practical problems and enhance system performance. This method holds great importance in enhancing surveillance system performance and individual tracking capabilities.

Key words: queuing theory, vertex weighting, trajectory recognition method, cross-camera tracking