摘要
人体运动目标检测一直是计算机视觉应用领域中一个重要研究课题,但检测过程中易受到背景抖动、环境光线变化等外界因素影响造成目标提取失败。为了消除噪声干扰,提高识别能力,在分析现有方法的基础上,提出一种基于帧差法和背景减除法相结合的人体目标检测方法。首先利用高斯模型构建自适应背景模型,并结合帧差信息对其进行选择性背景更新,将两种方法得到的检测结果进行逻辑运算,分割出完整可靠的前景目标。实验结果表明改进方法准确率高,适应能力较强,从而验证了目标检测的有效性。
Moving-objects detection has long been studied as an important research subject in computer-vision field.But the detecting result is always influenced by the changes of background scene,illumination and so on.In this paper,a new method of moving human detection based on fusion of background subtraction and temporal differencing is proposed through analyzing traditional methods.The method constructs the auto-adapted background model by Gauss,and adapts temporal difference to update the background,using background subtraction method to extract movement areas from the background model.Then the new method integrates the two foreground regions for object identification,obtaining the complete reliable moving target finally.The experimental results confirm that targets can be detected accurately and the approach adopted is very feasible.
出处
《计算机仿真》
CSCD
北大核心
2011年第2期308-311,共4页
Computer Simulation
基金
北京市教委科技创新平台项目(2008176)
北京市优秀人才培养资助项目(2009D005003000001)
北京市属高等学校人才强教深化计划学术创新团队项目(PHR201007123)
北京工商大学青年教师科研启动基金项目(2009-09)
关键词
运动目标检测
帧差法
背景减除法
背景更新
Moving-object detection
Temporal differencing
Background subtraction
Background updating