彩色自然场景下的多特征融合车辆检测算法  

Vehicle Detection Algorithm in Natural Environments Using Multi-features fusion

作　　者：陈明亮[1] 岑峰[1] 

机构地区：[1]同济大学电子与信息工程学院,上海201804

出　　处：《系统仿真技术》2015年第3期191-195,共5页System Simulation Technology

基　　金：国家自然科学基金(61273284)

摘　　要：在自然场景下,由于车型种类繁多,而且车辆所处环境受光照、背景、天气等因素的影响,车辆检测一直是目标检测中的难点问题。针对这种自然场景下的车辆检测,提出一种融合稀疏自编码特征和方向梯度直方图(Histogram of Oriented Gradients,HOG)特征的多特征描述子的车辆检测算法。实验结果表明,提出的方法对复杂自然场景下的车辆具有较好的分类效果。In the natural scene,because of a wide range of vehicle models and the environment which is impacted by lighting, background, weather and other factors. Vehicle detection has been one of difficult problems in target detection. To solve these problems, we proposed a vehicle detection algorithm based on multi-feature descriptor integrated sparse auto-encoder andHistogram of Oriented Gradients （HOG） features. The experiment results show that the proposed method provides higher classification accuracy when vehicles in complex natural scene

关 键 词：自然场景 车辆检测 稀疏自编码 多特征融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]