能量法的自适应背景更新算法  

A Self-Adaptive Background Updating Algorithm of Energy Method

作　　者：田军[1] 魏振华[1] 武思远 

机构地区：[1]华北电力大学计算机科学与技术系,北京102206

出　　处：《计算机科学与探索》2009年第2期218-224,共7页Journal of Frontiers of Computer Science and Technology

基　　金：国家自然科学基金No.60775058;国家教育部科学技术研究重点项目No.107028;华北电力大学博士科研基金No.200722019~~

摘　　要：光流法不适用于复杂的环境,计算量大、用时长,且实时性不高,而背景图像差分法只适用于静止不动的背景,且不能使背景图像随监视场景背景变化而及时地更新。因此,提出了能量法的自适应背景更新算法SBUE。该算法主要用于运动目标检测,即在复杂背景变化的情况下,能够实时地修正或更新粗背景模型CBM,再结合背景图像差分法与运动物体能量相似性分割出完整的运动目标,具有普遍的适用性。通过实验证明了算法可以快速、准确地更新变化复杂的背景,提高运动目标的识别精度。With the large computation quantity, long computing time and low real-time performance, the optical flow is not fit for the complex surroundings. The background image difference is only fit for the still background and no reasonable approach has been designed and implemented for automatic background updating along with the background variance. To solve the above problems, a self-adaptive background updating algorithm of energy method （SBUE） is proposed. In the complicated background, this algorithm is mainly applied to the moving object detection, which can modify or update the coarse background model （CBM） in real time. Moreover, combining the background image difference and energy similarity of moving objects, this algorithm can extract the moving objects completely and its application is universal. Experimental results demonstrate that the proposed self-adaptive back- ground updating algorithm of energy method （SBUE） can update the variation background exactly and quickly and improve the recognition accuracy of the moving objects.

关 键 词：运动目标检测 能量法 粗背景模型 背景图像差分法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]