摘要
提出了一种新的基于柯西分布的光照模型(shading model,SM)变化检测方法。在一种快速动态背景图像初始化的基础上,建立了Gaussian统计背景模型;基于使用统计假设检验方法检测变化区域的结果,利用YCbCr颜色空间的亮度、颜色信息,识别和消除视频序列图像中的阴影和反光等。试验表明,该文所提出的方法可以承受整体或局部的、缓慢或突然的背景光线变化,以及由场景背景中小的背景扰动、阴影或反光导致的噪声,可以较精确地检测背景目标,改善了SM方法在较暗区域的目标检测效果。
A novel illumination-invariant Cauchy distribution based change detection using shading model （SM） for a fixed visual surveillance system is proposed. Both the initialization method of Gaussian background models and the estimation of parameters for the Cauehy dis- tribution model under the maximum likelihood estimation are presented. Based on results of change detection using statistical hypothesis test, the intensity, hue and saturation in the YCbCr color space are employed to recognize and eliminate shadows and reflections in video sequences. Finally, experimental results demonstrate that the proposed method of back-ground modeling can tolerate changes in illumination, and noises caused by some small motions, shadows or reflections in a background scene. The proposed approach also improves the performance of objects detection in darker regions.
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2008年第12期1216-1220,共5页
Geomatics and Information Science of Wuhan University
基金
国家863计划资助项目(SQ2006AA12Z108506)
关键词
运动目标检测
背景建模
阴影
柯西分布
moving object detection
background modeling
shadows
Cauchy distribution
