一种用于智能车环境探测的立体视觉传感器  

A Stereo Vision Sensor for Environmental Detection of Intelligent Vehicle

作　　者：李琳辉[1] 赵一兵[1] 张明恒[1] 郭烈[1] 

机构地区：[1]大连理工大学工业装备结构分析国家重点实验室；运载工程与力学学部；汽车工程学院,辽宁大连116024

出　　处：《传感技术学报》2010年第9期1247-1251,共5页Chinese Journal of Sensors and Actuators

基　　金：中央高校基本科研业务基金资助(20103013852007)

摘　　要：设计了应用于户外智能车环境感知的立体视觉系统,并提出了两种立体匹配算法,分别实现不同的功能。基于多特征提取的稀疏匹配算法首先通过图像局部增强处理减弱光照和场景变化的影响,然后在对极线引导和连续性约束下提取立体图像的角点特征和边缘特征进行匹配,能够实时探测环境的三维概貌,并突出环境中的障碍物信息,实现辅助车辆自主导航的功能;基于特征引导的稠密匹配算法采用了最小割/最大流算法,并利用多特征匹配的结果来剔除稠密匹配中的成块误匹配,能够重建出未知环境详细的三维信息,实现三维可视化功能。最后通过试验验证了两种算法的有效性。A stereo vision system is designed for environmental perception of outdoor intelligent vehicle, and a couple of stereo matching algorithms（MESM and FGDM） are proposed for different usage. For the multi-feature extraction based sparse matching algorithm（MESM）, local image enhancement technology is proposed firstly to weaken the influence of varying illumination and scenes. Then corner features and edge features are extracted for matching under epipolar guidance and continuity constraint. MESM method can provide a roughly description of 3-D environment information on real time and highlight the obstacles in the environment, which contributes to the achievement of autonomous vehicle navigation. For the feature-guided dense matching algorithm（FGDM）, a min-cut/max-flow algorithm is considered, and the result of MESM is utilized to eliminate pseudo matching of dense matching. FGDM method can provide a detailed description of 3-D environment information, which contributes to the visualization of unknown places. Experiments confirm the effectiveness of the proposed algorithms.

关 键 词：计算机视觉 智能车 环境探测 稀疏匹配 稠密匹配 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]