结合区域仿射变换的MSLD描述子与直线段匹配

MSLD Descriptor Combined Regional Affine Transformation and Straight Line Matching

针对直线匹配过程中直线描述子的稳定性问题,提出一种结合区域仿射变换的MSLD直线描述子用于直线段匹配。匹配在同名三角网和核线约束基础上,首先对左影像上目标直线建立直线平行支撑域;再根据目标直线段两端点及其平行支撑域四角点的核线确定候选直线对应区域及其平行支撑域;然后根据目标直线支撑域长度和宽度对目标直线和候选直线平行支撑域分别进行仿射变换,建立目标直线和候选直线的MSLD特征描述子;最后结合欧式距离和最邻近比值(NNDR)确定最终同名直线。选取具有不同类型纹理特征的近景影像进行实验,结果表明,本文方法能获得可靠的直线匹配结果。

Aiming at the stability of the line descriptor in straight line matching, a MSLD descriptor combined regional Af- fine Transformation is presented for the straight line matching in this paper. Based on the corresponding triangulations and epipolar constraint, a support region parallel to straight line is established for target line in the left image, and the corre- sponding endpoints of candidate line and its parallel support region are determined according to epipolar lines of the end- points of the target line and the four corners of its parallel support region; Secondly, Affine Transformation is performed on the parallel support region of target line and candidate line respectively according to the length and width of target line sup- port region; Thirdly, descriptors are constructed from the support regions of target line and candidate line after the Affine Transformation; Finally, the corresponding line in the right image is determined by Euclidean Distance in descriptors and NNDR. Experimental studies are carried out with different kinds of texture feature close-range images, and the results show that the proposed method is reliable for line matching.