Ab
Abstract
-
- II -
Abstract
With the worldwide population explosion and city development, there are more and more densely populated public places. While these places provide people convenience, they also provide places for crime. Therefore, public security is an urgent issue faced by the entire world. As a main method to control the crime in public paces, video surveillance is applied more and more widely. Yet, because of human cannot focus on every single details in the whole scene and they may be tired during the work, using technological means to automatically analyze the video and discover anomalies is a popular research topic in computer vision community.
For abnormal behavior detection in crowded scenes, many models and algorithms have been published at top conference and journals in recent years. Moreover, several publicly available dataset referred to crowds behavior are published, too. Although there have been a plenty of works, many flaws still exist in them. Especially when in extremely crowded areas or occlusion frequently happen, the performance of detection has a lot of space to improve.
In this paper, we present a novel method based on sparse coded motion saliency for detecting abnormal events in crowded scenes. Unlike existing sparse coding based approaches, our model dose not need to learn a dictionary. Instead, it directly sparsely represents the motion feathers of the object region using its surrounding areas. The sparse representation error is used to measure the motion saliency intensity.
Additionally, to reflect the situations in the real world more precisely, we assume that: the area with more intense or more disorder motion has more probability to exist abnormal behavior. Based on the assumption above, we devise two attributes: motion intension attribute and motion consistency attribute and design two algorithms to measure these two attributes.
To evaluate our approach, two publicly available datasets – UMN dataset and UCSD ped1 dataset are