Shape-directed fundus image scaling method applied on an ophthalmology CAD system

Scaling fundus image to the required resolution is one of the important tools in the ophthalmology CAD system. In this paper, we develop a shape-directed fundus image scaling method for the ophthalmology CAD system in which a blood vessels extraction tool is also provided. The method first applies two kinds of functions to record the extracted blood vessels in order to maintain the shape of blood vessels including curvature and relative width at any resolution. The scaling procedure is then directed by these functions and the scaled image with this method has good shape maintenance of blood vessels without jag or blur border usually brought by other methods. The experiment results verify the effectiveness of our method and the measurement of the cross-correlation coefficient between the original image and the scaled image shows the correction of the interpolation kernel selection according to the Fluency information theory.