CAJ | 학술논문

为提高群体性事件智能视频监控的可靠性，提出在传统图像处理算法实现目标检测和特征提取的基础上，采用贝叶斯网络进一步对事件特征进行分析，重点介绍人群聚集事件判定贝叶斯网络建模、参数设置和推理等关键过程，并以具体的实例进行验证。结果表明：该方法具有建模简单、对智能视频分析算法依赖性低、鲁棒性和移植性强等特点，对建立可靠的智能视频监控系统具有参考价值。
위제고군체성사건지능시빈감공적가고성，제출재전통도상처리산법실현목표검측화특정제취적기출상，채용패협사망락진일보대사건특정진행분석，중점개소인군취집사건판정패협사망락건모、삼수설치화추리등관건과정，병이구체적실례진행험증。결과표명：해방법구유건모간단、대지능시빈분석산법의뢰성저、로봉성화이식성강등특점，대건립가고적지능시빈감공계통구유삼고개치。
In order to improve the reliability of the intelligent video surveillance system used for group events, introduces a method adopting Bayesian network to analyze the more event features, based on using the image processing algorithm to realize target detection and feature extract. It focuses on the key procedure of the group events judgment Bayesian network including modeling, parameter setting and reasoning, and validated it with specific examples. The results show the method has the advantages of modeling simply, relying on the intelligent video analysis algorithms lowly, and high reliability and portability. It has reference value for building reliable intelligent video surveillance system.