张铮, 王孙强, 胡新宇,等.基于多特征融合和信息熵优化的前车检测[J].河南理工大学学报（自然科学版）,2022,41(3):138-145.

ZHANG Z, WANG S Q, HU X Y, et al.Preceding vehicle detection based on multi-feature fusion and information entropy optimization[J].Journal of Henan Polytechnic University(Natural Science) ,2022,41(3):138-145.

基于多特征融合和信息熵优化的前车检测

张铮, 王孙强, 胡新宇, 熊盛辉, 胡凌辉

湖北工业大学 机械工程学院，湖北 武汉430068

摘要:针对高速公路前车检测中单一特征易受光照、天气等环境因素影响的问题，提出一种基于多特征融合和信息熵优化的检测算法。首先，利用自适应大津算法对路面进行阈值分割，生成若干车辆假设区域；其次在方向梯度直方图（histogram of gradient， HOG）特征的基础上，引入几何特征、纹理特征和幅值特征构造特征向量，并根据信息熵对特征向量进行优化；最后，训练支持向量机（support vector machine， SVM）验证假设区域。实验结果表明，该算法提高了前车检测的准确率，扩大了前车检测的适用范围。

关键词:前车检测;支持向量机;多特征融合;特征优化;信息熵优化

doi:10.16186/j.cnki.1673-9787.2020080056

基金项目:国家自然科学基金资助项目（61976083）

收稿日期:2020/08/22

修回日期:2020/11/09

出版日期:2022/05/15

Preceding vehicle detection based on multi-feature fusion and information entropy optimization

ZHANG Zheng, WANG Sunqiang, HU Xinyu, XIONG Shenghui, HU Linghui

School of Mechanical Engineering， Hubei University of Technology， Wuhan 430068 ， Hubei， China

Abstract:Aiming at the problem that using a single feature was easily affected by environmental factors such as illumination and weather in the preceding vehicle detection on the expressways, a vehicle detection algorithm based on multi-feature fusion and optimization was proposed. Firstly, several vehicle hypothesis regions were generated by using an adaptive Otsu algorithm to segment the road surface. Then ,on the basis of histogram of gradient( HOG ) features, geometric features, texture features and amplitude features were introduced to build the features vectors which were optimized according to the information entropy. Finally, the support vector machine(SVM) classifier was trained to verify the hypothetical region The experimental results showed that the proposed algorithm improved the accuracy of the preceding vehicle detection and expanded the applicable scope of the preceding vehicle detection.

Key words:preceding vehicle detection;SVM;multi-feature fusion;feature optimization;information entropy optimization

基于多特征融合和信息熵优化的前车检测_张铮.pdf