IM (Image Mosaicing) is the main method for achieving images with high resolution and wide field of view. This technique has a great role in ophthalmic image processing used in imaging systems like fundus camera and OCT (Optical Coherence Tomography). Moreover, in IM method obtained images from different parts of retina (having limited field of view between 30 to 60 degrees with overlapping) have been registered with each other to create a picture with wide field of view. In present research, a new approach to achieve this goal has been simulated.

First, images from different parts of retina have been obtained using Topcon fundus camera. Then, preprocessing has been done to remove artifacts and enhance image contrast through curvelet transform. Afterwards, retinal blood vessels structures have been extracted using curvelet transform and edge detection algorithms have been applied on these structures. Subsequently, the most appropriate edges have been selected as the correspondent points utilizing feature selector algorithms. Then, by transfer function these similar points in two images have been registered with each other applying affine transform. Finally, differences in brightness at merged edges have been regularized using image blending techniques. These processing has been conducted using Matlab software.

Proposed algorithm has revealed promising results in image mosaicing of retinal pictures. This algorithm is invariant to rotation, scaling, translation and distortions.

It is suggested for future studies that region based feature extraction methods may be more effective in image registration than point based feature extraction algorithms.