Novel approach based on topological simplification algorithm optimized with Particle Swarm Optimization

Abstract

The movement of people can be considered as the flow of liquid, so we can use the methods employed for the flow of liquid to understand the motion of a crowd. Based on this, we present a novel framework for abnormal behavior detection in crowded scenes. We extract a topological structure from the crowd with the topology simplification algorithm. However, a conventional topology simplification algorithm can not work well if we apply it to the crowd directly because there is too much noises produced by the random motion of the people in the original image. To overcome this, we make a step forward by optimizing this model using Particle Swarm Optimization (PSO) to perform the advection of particle population spread randomly over the image frames. Then we propose two new methods for analyzing the boundary point structure and extraction of a critical point from the particle motion field; both methods can be used to describe the global topological structure of the crowd motion. The advantage of our approach is that each kind of abnormal event can be described as a specific change in the topological structure, so we do not need construct a complex classifier, but can classify the crowd anomalies dynamically and directly. Moreover, the approach monitors the crowd motion macroscopically, making it insensitive to the motion of an individual, disregarding the global movement. The result of an experiment conducted on a common data set shows that our method is both precise and stable.