摘要
视频序列中的运动目标检测是计算机视觉、视频监控等领域的关键问题。背景差分法是目前运动目标检测中最常用的一种方法．而构造一个自适应更新的背景模型是背景差分法的核心。利用运动目标图像变化比背景图像变化要快的特点，提出了一种改进的构建并实时更新背景图像的方法。实验表明，该方法计算量小、实时性好、并且能够确保较好的检测精度。
Moving target detection in Video sequence is one of the critical issues in the fields of computer vision and video surveillance. Background subtraction is a typical approach to detect moving targets. The core of background subtraction is how to construct an adaptive updating background model. This paper provides an improved method for constructing the background image and updating it at time, based on the feature that the motion image changes faster than the background image. Experiments proved that this method has lower computational complexity, better real-time performance and higher accuracy of detection.
作者
潘亚男
白帆
PAN Ya-nan, BAI Fan （Lanzhou Jiaotong University, School of Electronic and Information Engineering, Lanzhou 730070, China）
出处
《电脑知识与技术》
2009年第10期8026-8027,8030,共3页
Computer Knowledge and Technology
关键词
运动目标检测
背景差分法
背景模型
motion detection
background subtraction
background model
