基于YC_bC_r颜色空间的背景建模及运动目标检测

高斯混合模型广泛应用于基于背景建模的运动目标检测中。首先在YCbCr颜色空间采用自适应高斯混合模型对背景的每个像素建模;然后,对输入的当前帧图像的每一像素值与该像素点对应的高斯混合背景模型的各个高斯模型进行比较,将前景运动区域（包括运动目标、投射阴影）从场景中提取出来;最后,采用局部二元图（Lo-cal Binary Pattern,LBP）来提取纹理特征,利用背景在阴影覆盖前后的纹理相似性去除投射阴影,同时结合阴影的空间几何特性优化运动目标检测结果。实验结果表明,该算法能有效地检测出投射阴影和运动目标,具有较高的实际应用价值。

Background Modeling and Moving Object Detection Based on YC_bC_r Color Space

Gaussians mixture model （GMM） has been widely used for moving object detection based on background modeling.In this paper,the background is firstly modeled using adaptive Gaussian mixture models in YCbCr color space,and the foreground regions including moving objects and cast shadow are extracted from current frame by comparing the each pixel with Gaussian model.Then,the texture of little patches is represented by local binary patterns and the cast shadow is detected and eliminated based on the texture similarity between shadow region and corresponding region in the background.Finally,the geometric features of cast shadow are imposed to further improve the performance of moving object detection.Experimental results demonstrate the proposed algorithm can effectively detect cast shadow and moving object,and has higher practicability.