131 A Efficient Image Matching Based on SIFT

Image matching is a fundamental task in image matching and other computer vision applications. The Scale Invariant Feature Transform (SIFT) algorithm is generally considered as the most appealing application for practical uses, the SIFT descriptors remain invariant under rotation, scaling, changing viewpoint, affine distortion and variations in lighting conditions. A efficient Image matching based on SIFT is proposed by using the rotation and scale invariant property of SIFT. We generate the keypoint descriptor with the steps of scale-space extreme detection, accurate keypoint localization and orientation assignment, and then match the feature point by comparing the feature vector. The experimental results demonstrate that the SIFT algorithm is robust to scaling, rotation and noise.