基于边缘检测与码本模型的运动目标检测算法  

Moving Object Detection Algorithm Based on Edge Detection and Codebook Model

作　　者：雷飞[1] 黄文路[1] 张泽[1] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100124

出　　处：《科学技术与工程》2013年第24期7234-7239,共6页Science Technology and Engineering

基　　金：2012科研计划项目(002000546312507)资助

摘　　要：传统的码本模型能够快速完整地检测出运动目标,但检测出的运动目标边缘不准确;同时易受到阴影的干扰。针对这一问题,提出了基于边缘检测与码本模型的运动目标检测算法。该算法在码本模型提取运动目标的基础上,利用基于YUV色彩空间的阴影去除算法消除阴影获得一个运动目标图像;同时建立均值背景,然后对背景和当前图像进行Canny边缘检测和差分获得运动目标的边缘轮廓图像,然后利用形态学闭运算和连通性操作进行填充得到另一个运动目标图像,最后将两个运动目标检测图像进行"与运算"得到最终的运动目标。经过实验仿真,能够提取边缘轮廓准确和完整的运动目标,同时能有效地抑制噪声。The codebook model can quickly and completely detect moving objects, but the edge of the moving objects is inaccurate and subjects to the interference of the shadow. To solve these problems, an moving objects detection algorithm based on edge detection and codebook model. In this algorithm are proposed, the preliminary moving objects image are extracted by codebook model and shadow removal algorithm in YUV color space. And the other preliminary moving objects image are obtained by morphological and communicating operation to fill the edge of the moving objects that is the result of subtracting the Canny edge of current image and the respective Canny edge of mean background image. The ultimate moving objects is extracted by ＂and＂ operation of two preliminary moving objects images. Experiments show that this algorithm can completely extract moving objects with accurate edge feature, and inhibit interference of noise.

关 键 词：码本模型 CANNY边缘检测 阴影去除 YUV 运动目标检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]