一种基于PCP的块稀疏RPCA运动目标检测算法

A Block-sparse RPCA Algorithm for Moving Object Detection Based on PCP

针对已有的块稀疏RPCA运动目标检测方法难以适用于动态变化背景的问题,提出一种基于PCP的块稀疏RPCA运动目标检测算法。该算法首先通过基于PCP的RPCA方法对视频序列降维,将观测图像序列分解成低秩背景矩阵和稀疏前景矩阵;然后根据运动特性的光流一致性特点,结合前景区域的空间相关性,进一步得到大致的前景稀疏块;再利用基于PCP的块稀疏RPCA方法,动态地估计前景运动区域,重构出前景目标。实验结果表明,该算法能有效地排除运动和变化背景的干扰,提高对小目标的检测率。

Aiming at the shortcoming of being unsuitable for dynamic background for the existing RPCA based block-sparse moving object detection method,this paper proposes a PCP based block-sparse RPCA object detec-tion algorithm. First,the observed image sequence was regarded as the sum of a low-rank background matrix and a sparse outlier matrix,and then the decomposition was solved by the RPCA method via PCP. According to the consistent optical flow of motion saliency,by imposing spatial coherence on these regions,the rough fore-ground regions were obtained. Finally block-sparse RPCA algorithm through PCP was used to estimate fore-ground areas dynamically and to reconstruct the foreground objects. Extensive experiments demonstrate that our method can exclude the interference of background motion and change,simultaneously improving the detec-tion rate of small targets.