Integrating SIFT and Harris corner detector for definite keypoint extraction for image matching

Abstract

Feature point extraction automatically instead of manually is important and can improve the efficiency for photogrammetric tasks. Scale invariant feature transform (SIFT) is an important algorithm developed by Lowe (2004) to extract keypoints with invariance on scale and rotation for image matching and registration. Not all these kinds of keypoints do correspond to the definite position of object points in real world, therefore not all of them are suitable for precise localization in photogrammetric task. The definite keypoints can be used for image matching and precise location of object points. For photogrammetric applications, Harris corner detector is often employed to detect definite and precise keypoints, but those kinds of keypoints do not have the characteristics of invariance on scale and rotation. Therefore, this study will integrate SIFT and Harris corner detector to extract the precise and definite keypoints for image matching in order to determine the precise location of object points. Meanwhile, the efficiency of image matching from the extracted keypoints will be investigated in this study.