基于SIFT的UAV载组合宽角相机影像匹配方法

针对无人驾驶飞机UAV(Unmanned Aerial Vehicle)航空组合相机获取的大像幅影像旋偏角较大、 大尺度变化和颜色差异明显的问题, 提出基于极几何和单应约束的SIFT(Scale Invariant Feature Transform)特征多尺度LSM(Least Squares Matching)算法。该算法顶层金字塔影像采用SIFT快速匹配, 对匹配结果利用改进的RANSAC(Random Sample Consensus)算法计算影像间单应矩阵和基本矩阵； 对影像进行Harris特征提取, 根据极几何和单应约束采用双向一致性相关系数算法进行密集匹配； 通过更新单应矩阵, 设定阈值删除误匹配点； 对匹配的同名点进行最小二乘匹配获取子像素级精度。通过对具有较大旋偏角、 大尺度变化和颜色差异的3组实际航摄影像的试验对比表明, 与传统方法相比, 该算法具有较高的匹配成功率和较好的有效性。

Image Matching Method Based on SIFT for UAV Images from Combined Large Frame Camera

In order to solve the problems in the characteristics of UAV(Unmanned Aerial Vehicle) image with large frame, i.e., large rotation angle, large difference in scales an
d color difference, a matching method named multi-scale LSM(Least Squares Matching) algorithm based on SIFT (Scale Invariant Feature Transform) features with epipolar and homography constraints, which can improve the matching success rate is designed. On the top pyramid images, SIFT image matching is done to obtain matching points. The homography matrix and basic matrix are calculated with the matching points by the improved RANSAC(RANdom SAmple Consensus) algorithm. And the harris feature extraction is used to obtain many feature points. According to epipolar and homography constraints two-dimensional concordance correlation coefficient algorithm is used to dense stereo matching. The homography matrix is updated for deleting false matchingpoints by setting threshold. Corresponding image points are used to obtain sub-pixel accuracy by LSM. Based on three groups of comparative tests with actual aerial photograph images, i.e., images with large rotation angle, lager different scales and color difference, it is proved that this method is effective.