Funduscopy Image Stitching Based on SURF Algorithm
Liang Huang, Xianchang Chen, Hongli Wang · International Conference on Business Computing and Global Informatization · 2013
Aiming at the low contrast, uneven illumination, view limitations and geometric distortion between the different view of funduscopy images, This paper proposed a kind of image mosaicing algorithm based on improved SURF(speed up robust features) feature matching. SURF method is a fast and robust feature extraction algorithm, the algorithm is used to extract image features, The improved BBF (best bin first) fast matching algorithm is used to find image matching points, The RANSAC algorithm is used to eliminate false matching points, The linear optimization method is used to optimize homography matrix. Finally, the maximum fusion method is used in image fusion, effectively eliminate the stitching marks, and maintain high resolution. Experimental results verify the efficiency of the algorithm on rotation, scale and illumination changes of funduscopy images.