作　　者： ; ; ; ; ;

机构地区： 河海大学计算机与信息学院

出　　处： 《测试技术学报》 2011年第6期544-547,共4页

摘　　要： 在图像测量、目标检测过程中,传统的背景更新算法对于背景的扰动变化过于敏感.为此,本文提出了一种基于图像分块和Hausdorff距离判断的背景更新方法.首先采用高斯混合模型初步得到背景,然后对图像进行分块,借助图像匹配的思想计算当前图像块和背景图像块的Hausdorff距离,判断是否满足背景更新条件.若满足,则更新背景;若不满足,则再通过计算其色调值的Hausdorff距离来决定是否更新背景.实验表明,新算法通过图像灰度、色调等多种参数判别,能较好地克服背景扰动变化的干扰,背景更新的效果较好,有利于后期的目标检测与识别. To solve the problem that traditional background update algorithms are always sensitive to background disturbances when used for image measurement and target detection, a novel background update algorithm based on image blocks and Hausdorff distance is proposed in this paper. First, the initial background is obtained according to the Gaussian mixture model. Then all the images are divided into blocks with the same size, and the luminance Hausdorff distance are calculated for each pair of the current image block and the background image block. If the luminance Hausdorff distance is smaller than the threshold, then the background block is updated with the current block, and else the hue Hausdorff distance is calculated and compared with the threshold to determine whether the background block be updated. Experimental results show that this algorithm can overcome background disturbances and achieve good background update effect, which will be helpful for object detection and recognition.

关 键 词： 图像检测 高斯混合模型 距离 图像分块 背景更新

领　　域： [自动化与计算机技术] [自动化与计算机技术]