Automatic Retinal Image Registration Using fully Connected Vascular Tree

Retinal image registration plays very important role in medical imaging, as it combines the information of different field of view images and temporal images. This paper presents the novel method for retinal image registration based on bifurcation structure matching. In proposed method, segmentation is based on the multi-scale ridge detection, and the preprocessing steps used to avoid false detection of vessels. Also, Dijkstra's algorithm is used to get a fully connected vascular tree. A novel technique is proposed for removing the false correspondence. This method is invariant to translation, rotation and scaling since feature vector contains normalized lengths and angles. The registered image is visually good as well as experimental performance shows very good results.