基于最小风险的Bayes决策方法在交通检测中的应用  

Application to Traffic Detection of Bayes' Strategy Based on Minimum Loss

作　　者：迟晓君[1] 孟庆春[1] 陈鹏[1] 

机构地区：[1]中国海洋大学计算机科学系,山东青岛266071

出　　处：《计算机应用研究》2005年第12期204-205,220,共3页Application Research of Computers

基　　金：山东省自然科学基金资助项目(Y2002G18)

摘　　要：提出了一种在智能交通系统中实时检测交通参数时更新路况背景的新方法,即基于最小风险的Bayes决策分类的背景提取方法。通过比较在不同类别路况模式下的期望风险值,找到最小风险值所对应的决策方案,相应地采取不同的策略替换路面背景。这种方法可达到精确地检测车辆的目的,并将其应用于智能交通参数的视觉检测过程中。其结果满足实时性检测的要求,可作为智能交通系统中高层控制管理的基础。Proposes a new method of background extraction on real-time traffic detection which is based on the minimum loss of Bayes＇ strategy. By contrasting the loss values in different circumstances the best strategy method can be found as well as corresponding background strategies can be adopted. The extracted background based on this method is more accurate that can be used to detect vehicles. The results meet the need of real-time detection and can be used as the basis of advanced traffic control and management in intelligent transportation systems.

关 键 词：Bayes决策 最小风险 视觉检测 智能交通系统 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]