Two variables compactly supported fluency sampling functions-based scalable conversion for fundus photograph

To match the different resolution capabilities of devices used in the telemedicine network, we must perform resolution conversion for fundus photograph to the required resolution with the information important to ophthalmologists´ diagnosing well preserved. In this paper, in order to keep important information, such as color and shapes of blood vessels, optic disk and lesion areas on the fundus photograph after resolution conversion, we adopt three kinds of fluency functions to approximate shapes and propose a shape-directed resolution conversion method by using two variables compactly supported fluency sampling functions. The method leads the fundus photograph resolution conversion with features of scalability and good information preservation.