The Method of Pavement Image Splicing Based on SIFT Algorithm

According to the characteristic of pavement image, a new splicing method for pavement image based on scale invariant feature transform was proposed. The SIFT algorithm is used to detect the feature points, FCM clustering is used to find the most similitude cluster relative to the number of feature point, and match the feature descriptors in the cluster ,then the false feature points pair is rejected by RANSAC algorithm. Finally the correct match feature point´s pair is used to realize image splicing in order to display the panoramic images of pavement images. There is a great reference value for improving the system of pavement image Automatic detection.