摘要
为解决视频序列在复杂环境下,针对运动目标检测中存在噪声、光线变化及非目标物局部运动等影响目标检测精度的问题,利用Σ-Δ背景估计建立背景模型,用Kalman滤波背景估计对当前背景模型进行补偿校正,将两者进行结合建立稳定的背景模型,产生与实际背景模型相似的背景图像,使用背景减除法提取出运动目标。提出一种基于Σ-Δ背景估计的视频序列运动目标检测方法。实验结果表明,所提算法能抑制非目标物的影响,对缓慢变化的光线具有一定的适应能力,与其它算法相比,本文算法能较完整提取运动目标,具有较高的Precision、F1和Similarity指标,且具有较强的鲁棒性和自适应性。
To solve the problem of noise, light change and local motion of non-target which affect the accuracy of target detection in complex environment of video sequence, the Σ-Δ background estimation was used to establish the background model and Kalman filter background estimation was used to correct and compensate the current background model. The two methods were used to build the stable background and generate background image model which was similar to the real background. The moving object was extracted using the background subtract method. a moving object detection method based on Σ-Δ background estimation for video sequences was proposed. The experimental results show that the proposed algorithm can restrain the inf- luence of non-target objects and adapt to slowly changing light. Compared with other algorithms, the proposed algorithm can extract moving targets more completely. It has higher Precision, F1 and Similarity indices, and has stronger robustness and adap- tability .
作者
刘仲民
何胜皎
胡文瑾
LIU Zhong-min;HE Sheng-jiao;HU Wen-jin(College of Electrical and Information Engineering,Lanzhou University of Technology,Lanzhou 730050,China;Key Laboratory of Gansu Advanced Control for Industrial Processes,Lanzhou University of Technology,Lanzhou 730050,China;National Electrical and Control Engineering Experimental Teaching Center,Lanzhou University of Technology,Lanzhou 730050,China;College of Mathematics and Computer Science,Northwest Minzu University,Lanzhou 730000,China)
出处
《计算机工程与设计》
北大核心
2019年第3期788-794,共7页
Computer Engineering and Design
基金
国家自然科学基金项目(61561042)
关键词
运动检测
Σ-Δ背景估计
KALMAN滤波
背景差分法
目标提取
motion object detection
Σ-Δ background estimation
Kalman filter
background subtraction
object extraction