Enhancement of retinal fundus image using multi-scale tophat transformation

In biomedical image processing, enhancement of fundus image is the challenging problem to reveal the hidden features of a retinal image. To enhance the features of retinal images, different image enhancement techniques are used. This paper presents a multi-scale tophat transformation technique for retinal image enhancement. Simulation is performed on the DRIVE dataset of fundus image using MATLAB. Various performance metrics have also been measured for checking the efficacy of the proposed method like CII, PSNR, clarity index (PL), mean square error (MSE), NAE. Comparative analysis of proposed transformation technique is made with existing techniques like HE, CLAHE, and BPDFHE in terms of different performance metrics. The proposed method achieved improvement in CII, PSNR, PL and reduction in MSE and NAE when compared with the previous existing techniques.