Nowadays, aerial images present a “bird’s-eye” view of geographical environment, and historical one provides the spatial information in the past. Through multi-temporal aerial images, we can analyze dynamic environmental changes. In Taiwan, Research Center for Humanities and Social Sciences (RCHSS) of Academia Sinica, has collected and scanned abundant historical maps and aerial images. By being processed through methods of computer vision, those materials can achieve greater value. Most of the historical aerial images haven’t been registered since there were no precise POS system for orientation assisting in the past. To handle the great quantity of images, we develop an automatic process to match historical aerial images by Scale Invariant Feature Transform (Lowe, 2004). This matching algorithm extracts extreme values in scale space, and becomes invariant image features, which are robust in rotation, scale, noise, and illumination. If two images have the same image feature point, we can use these points to do affine transformation or projective transformation for image alignment. Research that using feature points of SIFT for automatic registration of historical aerial images has proven feasible (Rau, 2014).

After image matching and alignment automatically, we only have the relative orientation of images. We still have to add control points manually for registration through least square adjustment based on collinear equation. Finally, we can use those feature points extracted by SIFT to build control image database in future work. Every new image will be query image and be extracted. If features of new points match with the point data in database, it means that the query image probably is overlapped with control images and then become new control data. After feature extracting, all computation is based on point data instead of image data, so the requirement of computation is low. With the growth of the database, more and more query image can be matched and aligned automatically. Also, further study such as multi-temporal environmental changes can be investigated by using this temporal spatial data system.