Improved Pulse Coupled Neural Network based on Maximize SNR for Vessel Extraction

Aiming at the problems of branch loss and fracture in the fundus retinal vessel segmentation algorithm, with improved pulsed coupled neural network (PCNN) is proposed. As the matched filter can maximize the signal to noise ratio even though using the gaussian filter, therefore, it is used for reducing the noise. First, matching filters are used to enhance the retinal image, and then a segmentation algorithm is suggested using a combination of matching filter and PCNN. Finally, the evaluation on DRIVE database demonstrates that the sensitivity, specificity, and accuracy of the algorithm are 0.7310, 0.9640, and 0.9711 accordingly compared with other algorithms. Simulation experiments illustrate that the Matched-PCNN algorithm resolves the effect of low light and optic disc highlight on vessel segmentation, and demonstrates good connectivity between blood vessels.