为解决变形较大的无人机影像配准问题,提出了点特征和边缘特征相结合的配准方法.用尺度不变特征变换(SIFT)算法提取点特征,完成影像的初步配准,并通过多项式函数对影像进行粗校正.在此基础上提取影像的边缘特征信息,根据距离相似性对边缘特征信息进行配准；依据色彩能量差筛选点特征信息配准结果和边缘特征信息配准结果,采用小面元微分校正的方法对变形影像进行校正.实验结果表明:提出的配准方法能够弥补点特征配准方法和边缘特征配准方法的不足,其配准的鲁棒性提高10％左右,可以较好地完成变形较大的无人机影像配准.%To solve the problem of registration of unmanned aerial vehicle (UAV) images with a large distortion, an improved method based on the combination of point and line features was proposed. With this method, preparatory registration is completed by extracting point features with the scale invariant feature transform (SIFT) algorithm, and a polynomial function is used to correct an image coarsely. On the basis of the above, the edge features of the image are extracted, and the registration of the edge features is carried out based on the distance similarity- The registration results of point and edge features are filtered by utilizing the color energy difference. A small-bin based differential correction method is used for finally accurate correction. The experimental results show that the proposed method can overcome the drawbacks of the point feature registration and the edge feature registration, the registration robustness is improved by 10% , and it can effectively complete the registration of UAV images with a large distortion.
展开▼
