摘要
对周边环境类型的自动辨识对无人驾驶汽车十分关键。近红外成像能够捕捉到可见光成像所无法发现的重要环境特征,因此有助于增强无人车的环境辨识能力。本文提出一种基于近红外成像的环境类型识别方法,通过对不同图像区域分别提取边缘与纹理特征并进行特征层融合,生成近红外环境场景图像的关键特征,在此基础上构建出两步式环境类型分类器。实验结果表明,本文方法可有效鉴别出不同类型近红外场景图像的特征差异,实现在近红外波段中对多种常见无人车周边环境类型的辨识。
Automatic surrounding environment type identification is critical for unmanned vehicles. Near infrared imaging can capture important environmental characteristics that are invisible for visible light imaging,thus it could enhance the capabilities of unmanned vehicles in environmental identification. In the paper,an environment type identification method based on near infrared imaging is proposed. The key features of near infrared scene images are generated by the feature- level fusion of edge and texture features extracted from different image blocks. On the basis,a two- step environment type classifier is built. The experimental results demonstrate that the proposed method can effectively distinguish the feature differences between various near infrared scene images,and identify a variety of common surrounding environments of unmanned vehicles in the near infrared band.
出处
《激光杂志》
北大核心
2015年第6期82-85,共4页
Laser Journal
基金
河南省教育厅高等学校重点科研项目(14A320029)
河南省科技攻关项目(112102210338)
关键词
机器视觉
环境类型识别
特征层融合
无人车
Machine vision
Environment type identification
Feature-level fusion
Unmanned vehicle