摘要
随着人工智能的发展,提出了使计算机系统具有模拟人类通过视觉接收外界信息、识别和理解周围环境、协助或代替人类感知的能力,基于视频序列的运动目标分析也就应运而生。本文针对目前常用的背景减法,帧间差分法和混合高斯背景建模的运动检测方法的优缺点,提出了一种3者相结合的运动目标检测算法。在讨论数学模型的基础上,通过OpenCV进行了实现,并对传统算法进行了简要介绍。实验结果表明该算法具有很好的检测效果和鲁棒性。
With the development of artificial intelligence, people begin to proposed to have computer system have capability of simulate human vision to accept the outside information, identify and understand the surrounding environment, to assist or replace human perception, analysis of video sequences of moving objects is emerged. In this paper, a new motion detection algorithm which combined commonly used background subtraction, frame difference and the Gaussian mixture model is proposed. Mathematic model is realized by OpenCV, traditional algorithm is briefly introduced. Experimental results show good detection performance and robustness.
出处
《电子测试》
2011年第7期27-29,33,共4页
Electronic Test
关键词
背景模型
背景减法
帧间差分
OPENCV
background modeling
background subtraction
frame difference
OpenCV
