作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

一种基于HOG-LBP 的高效车辆检测方法

杨先凤,杨 燕   

  1. (西南石油大学计算机科学学院,成都610500)
  • 收稿日期:2013-09-27 出版日期:2014-09-15 发布日期:2014-09-12
  • 作者简介:杨先凤(1974 - ),女,教授,主研方向:智能识别,数字图像处理,数据库技术;杨 燕,硕士研究生。
  • 基金资助:
    四川省应用基础研究计划基金资助项目(2011JY0060)。

A Method of Efficient Vehicle Detection Based on HOG-LBP

YANG Xian-feng,YANG Yan   

  1. (Institute of Computer Science,Southwest Petroleum University,Chengdu 610500,China)
  • Received:2013-09-27 Online:2014-09-15 Published:2014-09-12

摘要: 针对形状特征在车辆检测中存在的误检现象,在分析误检原因的基础上,提出一种融合形状和纹理特征的车辆检测方法。对检测窗口中划分的胞元进行方向梯度直方图特征和统一化局部二进制模式算子的求解,统计检测窗口中各胞元的特征情况,在形成浏览窗口的形状和纹理特征过程中,采用主成分分析解决特征的高维度和冗余问题,结合支持向量机进行特征训练和检测实验。实验结果证明,该方法有效兼顾车辆图像的形状和纹理两方面的特征,在不影响检测速度的同时,明显降低了车辆检测的误检率,在时效和精度两方面都取得较好的效果。

关键词: 车辆检测, 误检, 方向梯度直方图, 局部二进制模式, 主成分分析, 支持向量机

Abstract: According to the feature erroneous inspection that consists in vehicle detection,this paper proposes a vehicle detection method based on the fusion shape and texture characteristics in analysis of the error reason. It calculates the Histogram of Oriented Gradient (HOG) feature and the unified Local Binary Pattern (LBP) operator for all cell in detection window,solves the problem of high dimension characteristic and redundancy by Principal Component Analysis (PCA) in the browser window and texture characteristics-forming process. Combined with the Support Vector Machine (SVM),it does the feature training and test experiment. Experimental results show that this method juggles both sides of the shape and texture characteristics in vehicle image effectively,significantly reduces the error probability of the vehicle detection when meets the detection speed,gets good effect both in efficiency and accuracy.

Key words: vehicle detection, erroneous inspection, Histogram of Oriented Gradient ( HOG), Local Binary Pattern (LBP), Principal Component Analysis(PCA), Support Vector Machine(SVM)

中图分类号: