基于协方差流形和LogitBoost的异常驾驶行为识别方法
李此君; 刘云鹏; Li S(李帅)
刊名激光与光电子学进展
2018
页码1-13
关键词异常驾驶行为识别 协方差描述子 黎曼流形 多类LogitBoost分类器
ISSN号1006-4125
其他题名Abnormal Driving Behavior Detection based on Covariance Manifold and LogitBoost
通讯作者李此君
产权排序1
中文摘要由于因驾驶员的因素引发的交通事故比例居高不下,因此研究一种通过分析驾驶员活动状态对异常驾驶行为正确识别分类的识别方法具有重要的意义。本文提出了一种基于协方差流形和基于二分类思想的多类LogitBoost分类器的异常驾驶行为识别方法。首先提取图像的纹理、颜色和梯度方向等基础特征,以克服基于单一特征识别驾驶行为的不足;并利用协方差流形进行多特征融合,以消除特征冗余,同时降低由于不同特征数值差异过大可能对图像处理及识别带来的影响;最后使用基于二分类的多类LogitBoost分类器进行分类识别。针对不同目标的正确识别率可达81.08%,实验结果表明,相对传统的直接使用LogitBoost多分类方法,基于LogitBoost二分类器的多类分类方法较大幅地提高了多分类的正确率。
英文摘要Because of the high proportion of traffic accidents caused by the driver's factors, it is of great significance to study a recognition method for the correct identification of abnormal driving behavior by analyzing the driver's active state. In this paper, a recognition method of abnormal driving behavior based on covariance manifold and two class classification based on multi class LogitBoost classifier is proposed. First, we extract the basic features such as texture, color and gradient direction, so as to overcome the shortage of driving behavior based on single feature recognition, and use covariance manifold for multi feature fusion to eliminate feature redundancy and reduce the influence of different characteristics of different features on image processing and recognition. Then we use the multi class LogitBoost classifier based on binary classifiers to classify and recognize. The correct recognition rate for different targets is up to 81.08%. The experimental results show that the multi classification method based on the LogitBoost two classifier is better than the traditional LogitBoost multi classification method, and the accuracy of multi classification is greatly improved.
语种中文
内容类型期刊论文
源URL[http://ir.sia.cn/handle/173321/21861]  
专题沈阳自动化研究所_数字工厂研究室
通讯作者Li S(李帅)
作者单位1.中国科学院沈阳自动化研究所
2.中国科学院大学
3.中国科学院光电信息处理重点实验室
4.辽宁省图像理解与视觉计算重点实验室
推荐引用方式
GB/T 7714
李此君,刘云鹏,Li S(李帅). 基于协方差流形和LogitBoost的异常驾驶行为识别方法[J]. 激光与光电子学进展,2018:1-13.
APA 李此君,刘云鹏,&Li S.(2018).基于协方差流形和LogitBoost的异常驾驶行为识别方法.激光与光电子学进展,1-13.
MLA 李此君,et al."基于协方差流形和LogitBoost的异常驾驶行为识别方法".激光与光电子学进展 (2018):1-13.
个性服务
查看访问统计
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。


©版权所有 ©2017 CSpace - Powered by CSpace