改进的低秩稀疏分解及其在目标检测中的应用

针对传统低秩稀疏分解算法用于运动目标检测时,前景提取结果容易受噪声干扰以及检测结果不完整的问题,提出了一种新的低秩稀疏分解模型。考虑到视频前景目标呈结构化分布,以及动态背景对前景提取结果造成影响,该模型利用结构化稀疏范数对前景进行约束,且将稀疏部分所代表的运动区域进一步划分为动态背景部分与前景部分;然后采用广义交替方向乘子法对提出的模型进行求解,并分析了算法的复杂度;最后进行仿真实验将其应用到运动目标检测中。实验数据结果验证了提出的方法比其他基于低秩稀疏分解的运动目标检测方法更加稳定有效,更具有普适性,且对不同类型的噪声均具有一定的抗噪性。

Improved low rank and sparse decomposition with application to object detection

The traditional method of moving object detection is based on low rank and sparse decomposition algorithm, which has the problems of foreground extraction results affected by noise and incomplete detection of moving objects. To solve these problems, one kind of new low rank and sparse decomposition model is proposed. Not only the structural distribution characteristics of the video foreground object but also the influence of the dynamic background on the foreground extraction result are considered. The foreground is constrained by using the structured sparse norm. The motion area represented by the section is consisted of a dynamic background part and a foreground part. Then, the generalized alternating direction multiplier method is used to solve the proposed model, and the complexity of the algorithm is analyzed. Finally, simulation experiment is carried out and applied to moving object detection. Experimental results show that the proposed method is more stable and effective than other moving object detection methods based on low rank and sparse decomposition, which indicates its university. It also has certain anti noise ability for different types of noise.