融合提高视频车牌识别系统性能的研究  

Research on Video Vehicle License Plate Recognition System Performance Improvement by Amalgamation

作　　者：潘树龙[1] 王国新[1] 

机构地区：[1]烟台职业学院,山东烟台264670

出　　处：《烟台职业学院学报》2010年第1期71-76,共6页Journal of Yantai Vocational College

基　　金：山东省烟台市科技发展项目基金(2006120)

摘　　要：针对视频车牌识别系统中,同一车辆被多次采样,多次识别的情况,提出了一种结果融合的策略。该策略能从同一车辆产生的多次结果中选取一个最可靠的结果输出。该策略主要有两个部分组成:一是找出由同一车辆产生的所有结果。二是从这些结果中选取最可靠的一个作为输出。通过实验表明,采用此策略的系统识别率比相同质量的图像识别率提高6个百分点以上。In this paper,a result amalgamation method in video vehicle license plate recognition(LPR) system is proposed.This method can choose the best result from several results of the same car,since a car can be sampled several times in video LPR system.The method is divided into two steps: 1) find the results of the same car,and 2) judging which result is the best.An experiment has been performed to prove the effect.The correct rate of our video LPR system is 6 percent more than the same quality images correct rate.

关 键 词：模版匹配 识别分数 匹配分数 识别奖励 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]