机构地区: 深圳信息职业技术学院
出 处: 《深圳信息职业技术学院学报》 2009年第2期1-4,共4页
摘 要: 在机器视觉领域,针对监控视频的行为分析和语义理解技术日益受到关注。本文提出了一种应用于实际监控视频的语义理解模型,通过概念空间连接低层定量的语法特征和高层定性的语义描述。基于概念空间使用概念关联模型建模和识别视频场景中的目标行为。实验结果证明了本文所提出模型的有效性。 Action analysis and semantic interpretation in surveillance video have recently attracted increasing attention in the computer vision community. In this paper, a semantic interpretation model is proposed for practical applications in visual surveillance. 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. Experimental results demonstrate the performance of the proposed semantic interpretation model.
领 域: [自动化与计算机技术] [自动化与计算机技术]