结合同名点及核线约束的近景影像直线匹配

Line matching for close-range images by combining corresponding points and epipolar constraint

针对影像间尺度变化、直线提取断裂、直线邻域纹理断裂引起的近景影像直线匹配难题,该文提出一种结合同名点及核线约束的近景影像直线匹配算法。该算法在已有同名点和直线提取结果的基础上,首先确定目标直线两侧离目标直线距离最近的同名点,利用其在右影像上对应的同名点连线得到的虚拟直线为基准构建矩形候选区;其次,分别计算左、右影像上同名点连线得到的虚拟直线与待匹配直线的夹角,根据两角度差值进一步对候选直线进行筛选;然后,利用核线约束计算目标直线和候选直线的重叠部分,并以此为中心构建直线支撑域,统一两支撑域窗口尺寸,根据窗口灰度相关确定同名直线;最后,采用距离约束对匹配结果进行检核完成直线匹配。选取具有不同几何变换的三组近景影像进行直线匹配实验,结果表明该文方法能获得可靠的直线匹配结果。

Aiming at problems caused by scale variation of images,breaks of line extraction,and discontinuity of texture around lines,a line matching algorithm for close-range images by combining corresponding points and epipolar constraint was proposed in this paper.Based on the corresponding points and line extraction results,firstly,the closest corresponding points of both sides of the target line would be found.Rectangular candidate region would be established based on the virtual line obtained by connecting the corresponding points on the right image.Then,the angle between the virtual line connected by corresponding points and the matching line in the left and right images was calculated respectively,furthermore,the candidate lines would be selected according to the two angles difference.Secondly,the overlapping parts of the target line and the candidate line were calculated,which would be the center line of line support region,using the epipolar constraints.Then the window sizes of the two supporting regions were unified,and the corresponding line would be obtained according to intensity correlation of the windows.Finally,the check of line matching result would be implemented by using distance constraint.Three pairs of close-range images with different geometric transformation were selected for the line matching experiments.The results showed that the algorithm could obtain reliable line matching results.