CAJ | 학술논문

针对监控视频中静止物体的检测，提出了一种基于拉普拉斯分布模型的检测方法。该方法首先改进?-D背景建模方法，快速提取视频背景，构成初级背景，然后在初级背景中引入拉普拉斯分布模型，从而构成精确的自适应动态背景，最后比较初级背景与动态背景之间的差异达到检测静止物体的目的。实验结果表明，该方法能在标准视频数据库中有效地检测到静止行李，并对人群拥挤和光照变化等复杂场景有良好的检测效果。
침대감공시빈중정지물체적검측，제출료일충기우랍보랍사분포모형적검측방법。해방법수선개진?-D배경건모방법，쾌속제취시빈배경，구성초급배경，연후재초급배경중인입랍보랍사분포모형，종이구성정학적자괄응동태배경，최후비교초급배경여동태배경지간적차이체도검측정지물체적목적。실험결과표명，해방법능재표준시빈수거고중유효지검측도정지행리，병대인군옹제화광조변화등복잡장경유량호적검측효과。
In view of static objects detection in surveillance video, this paper proposes a method to solve above problem based on Laplacian distribution model. To begin with, the background modeling procedure of ?-D method is improved to abstract fast the background of video and form the primary background. And then, Laplacian distribution model is intro-duced to the primary background so that to abstract video self-adaptive dynamic background accurately which can reflect the background of video better. At last, the absolute difference between the primary background and dynamic background is used to detect the static objects. Experimental results prove the validity of this method to detect static luggage based on video standard databases and this method is robust in some complex scenes including crowded and illumination change.