Abstract: A new approach for automatic stitching of aerial image sequence was proposed, based on the SIFT (Scale Invariant Feature Transform) features. It consisted of two core steps: image registration and mosaic. The variations between two consecutive frames in aerial sequence are always very notable, thus the present commonlyused featurebased methods are not very suitable for their registration. Concerning this, an iterative registration method using the SIFT features was put forward. Moreover, for image mosaic, a new color fusion method was proposed to achieve smooth stitching results based on human visual features. Finally, experiments on real aerial sequences prove the effectiveness of the proposed stitching approach.

Key words: automatic stitching of aerial image sequence, automatic stitching, Scale Invariant Featuer Transform (SIFT) feature, image geometric transform, image mosaic

摘要： 针对航拍图像序列拼接问题，提出一种基于尺度不变特征变换（SIFT）特征的自动拼接方法，该方法主要包括图像配准和镶嵌两个步骤。由于航拍序列各帧图像之间存在较大差异，因而常用的基于特征的配准方法适用性较差，对此，提出利用SIFT特征来实现准确、稳健的航拍图像配准。进一步的，提出一种基于视觉特征的色彩融合方法以取得平滑的镶嵌效果。最后，通过对真实航拍序列进行拼接实验验证了所提方法的有效性。

关键词: 航拍图像序列, 自动拼接, SIFT特征, 图像几何校正, 图像镶嵌