Feature Matching Approach for 3D Reconstruction from Multiple Images

Abstract

This paper handles a robust method to handle matching problem of the 3D building from remote sensed images, where two or more images are captured from the different views without any camera information. The system is based on the SIFT interest point detector on foreground object. The system provides a method for extracting distinctive invariant features from images that can be used to perform reliable matching between different view of an object or scene. The features are invariant to image scaling, translation, and rotation, and partially invariant to illumination changes and affine or 3D projection. The system also describes the outlier removal and the homography matrix computation approach between two planes using corresponding points based on the RANSAC method. The proposed method gives a sufficient number of matches distributed on the image and is particularly suitable for 3D reconstruction from feature correspondence.