基于LBP纹理和改进Camshift算子的车辆检测与跟踪  

Vehicle Detection and Tracking Based on the Local Binary Pattern Texture and Improved Camshift Operator

作　　者：宋晓琳[1] 王文涛[1] 张伟伟[1] 

机构地区：[1]湖南大学汽车车身先进设计制造国家重点实验室,湖南长沙410082

出　　处：《湖南大学学报（自然科学版）》2013年第8期52-57,共6页Journal of Hunan University:Natural Sciences

基　　金：国家自然科学基金资助项目(51175159);中国博士后基金资助项目(20110490263);汽车车身先进设计制造国家重点实验室自主课题团队重点项目(61075004)

摘　　要：提出了利用背景图像LBP(局部二值模式)纹理和当前帧图像LBP纹理的相似度分析提取前景的方法,克服了车辆检测中常用的帧差法、背景差分法对光照比较敏感的缺点.同时基于H,S,V分量及改进的LBP纹理的联合直方图与金字塔L-K光流法中心跟踪相结合的Camshift跟踪算法,有效地解决了背景目标颜色相近可能会导致跟踪的目标区域加入背景后变大、处理较大帧间位移的视频跟踪上搜索窗口的位置准确度较低的问题.实验证明,该方法具有良好的检测和追踪效果.A method of extraction prospect,which uses the background image LBP（local binary pattern）texture and current frame image LBP texture similarity analysis,was put forward.This method overcomes the sensitivity to illumination methods in vehicle detection,such as frame difference method and background difference method.The Camshift tracking algorithm combines the H,S and V components,the improved LBP texture of the joint histogram with the centroid tracking by pyramid L-K optical flow.This method can effectively solve two problems：one that the similar background color may lead to the tracking of the target area bigger,and the other that the search window position accuracy is low when dealing with large displacement between frames of video.The experimental results prove that the method has good detection and tracking effect.

关 键 词：车辆检测 车辆跟踪 LBP纹理 CAMSHIFT算法 L-K光流法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]