Abstract:

In order to improve the accuracy of automatic monitoring and fault diagnosis in the railway vehicles
and other engineering applications, the five methods of feature point detection are compared and researched in the paper. SIFT (ScaleInvariant Feature Transform) method is selected to detect feature points from the overlapping images, which can mosaic precisely the two images under most relaxed conditions. The experimental results demonstrate that the proposed method can mosaic effectively the overlapping images captured by ordinary camera and obtain high resolution image, eliminating the influences caused by image distortion, overlapping and rotation.

Key words: overlapping image, feature point detection, scale-invariant feature transform（SIFT） method, image mosaic