Image Mosaic Based on SURF Feature Matching

Abstract

In the paper, an image mosaic algorithm based on SURF feature matching is proposed. The algorithm uses SURF operator which has strong robustness and superior performance to extract features instead of conventional SIFT operator. The extracted features are matched by a novel matching scheme - fast bidirectional matching. Then a RANSAC algorithm is applied to eliminate outliers and obtain the transformation matrix between images. Finally images are stitched by a multi-band blending algorithm. The presented algorithm overcomes the disadvantages of the traditional image mosaic methods which are susceptible to different scale and moving objects, and has a great improvement in calculation speed. Experimental results show that the method with strong robustness performs fast and effectively.