摘要
主要研究了智能视频监控系统中的运动目标检测算法,试图将非负矩阵分解算法引入运动目标检测算法中,通过非负矩阵分解算法对视频序列的背景进行建模,使用背景差分法将当前视频帧图像与建立的背景模型比较获得运动目标。针对运动目标检测中基本非负矩阵分解批处理算法的不足,提出一种基于滑动窗非负矩阵分解的运动目标检测算法。通过滑动窗处理控制非负矩阵分解模型中被分解矩阵的规模,降低了算法的计算复杂度和空间复杂度,并在一定程度上增加了模型的非记忆性。实验结果表明,该算法能够更好地自适应背景模型的动态改变,并且在视频场景中存在光照突变和较小运动目标时具有较好的检测效果。
The algorithms of detecting the moving objects in the intelligent video surveillance system are mainly studied. The Non-nega- tive Matrix Factorization （NMF） algorithm is introduced into the moving target detection algorithm for modeling the background of a video sequences, and the background subtraction method is used to obtain the moving target by comparing the differences of the current video frame and the background model. In order to solve the problem of the NMF algorithm with a batch process,a moving target detec- tion algorithm of NMF based on sliding window is put forward, which controls the size of the decomposed matrix in NMF matrix decom- position model by adjusting the length of the sliding window. The proposed algorithm can reduce the computation and space complexity, and to some extent, it can increase non-memory characteristic of the model. The experiments show that the proposed method can adap- tively change the background model and has better detection effect when there is light change and small moving target in video scene.
出处
《计算机技术与发展》
2017年第1期20-24,共5页
Computer Technology and Development
基金
安徽省科技攻关强警专项(1101b0403030)
中国科学院上海微系统与信息技术横向研发基金课题
关键词
非负矩阵分解
运动目标检测
背景差分
滑动窗
non-negative matrix factorization
moving object detection
background difference
sliding window
