基于分形盒子维数的车辆定位和识别方法  

Vehicle location and identification method based on fractal box dimensionality

作　　者：王荣本[1] 顾柏园[1] 郭烈[1] 余天洪[1] 

机构地区：[1]吉林大学交通学院,长春130022

出　　处：《吉林大学学报（工学版）》2006年第3期331-335,共5页Journal of Jilin University:Engineering and Technology Edition

基　　金：吉林省科技发展计划项目(20050316)

摘　　要：提出了一种基于计盒子维数的方法识别和定位前方的车辆。首先,利用差分和线性滤波获得车辆的水平边缘信息。然后,通过统计法缩小盒子维数的运算区域。最后,通过盒子维数的计算确定车辆的候选区域,并对该区域内的车辆进行定位。同时,利用对称性测度法对定位的车辆进行确认。在不同环境和光照条件下对车辆图片进行测试,表明了该算法具有较好的鲁棒性和可靠性。A frontal vehicle location and identification method based on the fractal box dimensionality （FBD） was proposed to help the auto - guided vehicle development. The horizontal edge informations of the located vehicle were obtained by the image difference and linear filtering. The search area of the FBD was reduced by the statistical method. The potential vehicle region was determined by calculation of the FBD and the location of the vehicle was achieved in this region. At the same time, a validation algorithm based on the symmetry measures was proposed to check the correctness of the vehicle location. The tests of the vehicle images under different environments and illuminations show that the proposed method is characterized by good robustness and reliability.

关 键 词：交通运输工程 视觉 分形盒子维 车辆定位与识别 

分 类 号：U495[交通运输工程—交通运输规划与管理]