CAJ | 학술논문

道路环境中的视觉障碍检测是智能车辆导航的一大难题，光流检测算法则为解决该难题提供了有效的途径。提出了一种复合特征光流检测算法，借助于不同类型局部特征的检测与匹配，通过图像分割发现感兴趣区域，通过光流检测为该区域的可通行性判定提供决策依据。与传统光流检测方法的对比实验证明了算法的有效性。
도로배경중적시각장애검측시지능차량도항적일대난제，광류검측산법칙위해결해난제제공료유효적도경。제출료일충복합특정광류검측산법，차조우불동류형국부특정적검측여필배，통과도상분할발현감흥취구역，통과광류검측위해구역적가통행성판정제공결책의거。여전통광류검측방법적대비실험증명료산법적유효성。
Feature optical flow algorithm supplies a valid resolution for the visual obstacles detection in road environment which is a big problem of autonomous navigation for intelligent vehicles.A compound method of feature optical flow,by virtue of detec-tion and match of local features,was proposed to detect the regions of interest (ROI)with image segmentation.Then,optical flow detection of ROI supplies a decision proof for passable judgments.The contrast experiments to traditional optical flow meth-od demonstrate the validity of the proposed method.