Applications of Video Structured Description Technology for Traffic Violation Monitoring

摘要

Action analysis and semantic interpretation in surveillance video have recently attracted increasing attention in the computer vision community. In this paper, video structural description model is proposed for practical applications for traffic violation monitoring. Conceptual space is defined to bridge the gap between low-level syntax which is quantitative and high-level semantic where information is handled by qualitative means. Based on the conceptual space, conceptual relating model is proposed to simulate and recognize the targets' behaviors in the scene. Applications for traffic violation monitoring experimental results demonstrate the performance of the proposed semantic interpretation model of video structural description.