计算机工程与应用 ›› 2016, Vol. 52 ›› Issue (21): 175-179.

• 模式识别与人工智能 • 上一篇    下一篇

Haar型LBP纹理特征的行人检测研究

周书仁1,2,王  刚1,2,徐岳峰1,佘凯晟1   

  1. 1.长沙理工大学 计算机与通信工程学院,长沙 410114
    2.长沙理工大学 综合交通运输大数据智能处理湖南省重点实验室,长沙 410114
  • 出版日期:2016-11-01 发布日期:2016-11-17

Haar characteristics LBP text feature for pedestrian detection

ZHOU Shuren1,2, WANG Gang1,2, XU Yuefeng1, SHE Kaisheng1   

  1. 1.School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha 410114, China
    2.Hunan Provincial Key Laboratory of Intelligent Processing of Big Data on Transportation, Changsha University of Science and Technology, Changsha 410114, China
  • Online:2016-11-01 Published:2016-11-17

摘要: 针对行人检测中直接在灰度图像下提取局部二元模式(Local Binary Pattern,LBP)特征受噪声影响大导致检测率低的问题,提出了基于HSV颜色空间提取改进型Haar型LBP(IHLBP)特征的方法。首先将图像由RGB颜色空间转化到HVS颜色空间,然后对HSV图像的H、S、V空间分别提取IHLBP特征,最后将3个IHLBP特征归一化后串接为一个特征向量,得到最终的IHLBP特征。在INRIA Person数据集上采用支持向量机(Support Vector Machine,SVM)作为分类器进行测试。实验结果表明,该方法能有效地提高识别率,可达98.5%。相比于HOG特征、HPG-LBP特征和WLD-LBP特征具有更好的实验效果。

关键词: 行人检测, HSV颜色空间, 特征提取, 改进型Haar型局部二元模式(IHLBP)特征, 支持向量机

Abstract: In order to solve the problem of low detection rate caused by noise when directly extracting Local Binary Pattern(LBP) feature from gray images in pedestrian detection, a method extracting Improved Haar-like LBP(IHLBP) feature based on HSV color space is proposed. Firstly, images are converted from RGB space to HSV space. Secondly, IHLBP features are extracted from H, S, V channel respectively. Thirdly, the final IHLBP feature is obtained by normalizing and seriating three IHLBP characteristics acquired from second stage. Tests are conducted in INRIA Person dataset using Support Vector Machine(SVM) classifier. Experimental results show that this approach achieves higher recognition rate reaching up to 98.5% and has a better performance when compared to HOG, HPG-LBP, WLD-LBP feature.

Key words: pedestrian detection, HSV color space, feature extraction, Improved Haar-like Local Binary Pattern(IHLBP) features, Support Vector Machine(SVM)