基于模糊推理背景分割的目标检测方法  

Object Detection Method Based on Fuzzy Inference Background Segmentation

作　　者：陈勇[1] 肖刚[1] 陈久军[1] 高飞[1] 金章赞[1] 

机构地区：[1]浙江工业大学信息工程学院,杭州310014

出　　处：《计算机工程》2010年第11期190-191,194,共3页Computer Engineering

基　　金：浙江省自然科学基金资助项目(Y1080636)

摘　　要：为解决传统背景差分法存在的背景更新缓慢问题,提出基于模糊推理背景分割的目标检测方法。该方法在传统的背景差分方法中引入帧间差分方法,结合IFTHEN推理规则进行模糊推理,实现了背景的快速更新及目标的正确检测。引入抗噪声推理机制,抑制跟踪目标抖动,增强方法鲁棒性。通过对鱼的运动检测实验表明,该方法能有效快速地提取干净的背景,对运动目标进行实时检测。In order to solve the problem of updating background slowly in the traditional background differencing, this paper proposes a new approach to object detection, which is based on fuzzy inference background segmentation. This method introduces frame difference and combines the IF THEN fuzzy inference rules to update the background rapidly and detect the object accurately. It introduces the anti-noise fuzzy inferenee, which can suppress the bounce object and improve robustness. This method can efficiently get the clean background and detect the real-time moving object through the experiments on detecting the fishes.

关 键 词：背景分割 目标检测 模糊推理 背景差分法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]