首页 | 官方网站   微博 | 高级检索  
     

一种支持向量机和数字图像相结合的能见度检测算法
引用本文:刘南辉,周海芳,章 杰.一种支持向量机和数字图像相结合的能见度检测算法[J].福州大学学报(自然科学版),2018,46(1):59-64.
作者姓名:刘南辉  周海芳  章 杰
作者单位:1.福州大学物理与信息工程学院微纳器件与太阳能电池研究所,1.福州大学物理与信息工程学院微纳器件与太阳能电池研究所,1.福州大学物理与信息工程学院微纳器件与太阳能电池研究所
基金项目:福建省自然科学基金资助项目(2016J01298), 福建省科技厅引导性资助项目(2016H0016)
摘    要:提出一种数字图像处理和支持向量机相结合的道路能见度检测算法,针对白天和夜晚建立不同能见度检测模型.对于白天的能见度检测,通过暗通道先验原理提取场景的全局透射率,进一步计算韦伯对比度特征值和图像能量梯度值等作为支持向量机的输入,训练得到白天能见度模型;对于夜晚的能见度检测,提取明度与对比度关系值(POLC),图像的功率谱和全局梯度值作为支持向量机的输入,训练夜晚能见度检测模型.检测结果表明,该算法较好地满足了人眼视觉特性,准确度高,可应用于智能交通和辅助驾驶等领域.

关 键 词:能见度检测  暗通道先验  支持向量机
收稿时间:2016/10/13 0:00:00
修稿时间:2016/11/27 0:00:00

A visibility detection algorithm based on digital image processing and support vector machines
LIU Nanhui,ZHOU Haifang and ZHANG Jie.A visibility detection algorithm based on digital image processing and support vector machines[J].Journal of Fuzhou University(Natural Science Edition),2018,46(1):59-64.
Authors:LIU Nanhui  ZHOU Haifang and ZHANG Jie
Affiliation:Institute of Micro-Nano Devices and Solar Cells,College of Physics and Information Engineering,Fuzhou University,Institute of Micro-Nano Devices and Solar Cells,College of Physics and Information Engineering,Fuzhou University,Institute of Micro-Nano Devices and Solar Cells,College of Physics and Information Engineering,Fuzhou University
Abstract:Visibility has an important effect on the traffic safety. In this paper, a algorithm of visibility detection by combining digital image processing with support vector machines is presented, which builds two different visibility detection models for daytime and night. For daytime visibility detection, firstly, the global transmittance of scene is obtained according to dark channel prior. Then other input features for support vector machines such as the Webber contrast feature and image gradient are calculated in turn, and the daytime visibility detection model is obtained by training them. For night detection, the brightness and contrast (POLC) , the image spectrum and global gradient are calculated and trained as input of support vector machine, and the night visibility detection model is obtained too. Experimental results show the proposed visibility detection algorithm with high accuracy matches human visual characteristics, and it is suitable to be applied in intelligence transportation and assistance driving.
Keywords:Visibility detection  Dark Channel Prior  Support Vector Machine
本文献已被 CNKI 等数据库收录!
点击此处可从《福州大学学报(自然科学版)》浏览原始摘要信息
点击此处可从《福州大学学报(自然科学版)》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号