Figure 4

Retinal model structure. Structure of the normal retina model (simulating the foveal vision and the functions of the OPL) firstly by simulating the foveated vision using a multi-scale resolution sampling approach. Pixels in the fovea region are set to 1:1 from the input image while peripheral pixels are blurred by a Gaussian function with exponentially growing kernel size with radial distance from the fovea. The macular degeneration block is used here to simulate the degenerated photoreceptors. Then color is separated into four channels; Luminance, Red, Green, Blue and Yellow to be used in simulating the color opponent channels. The processed image is then reconstructed in the reconstruction module (three channels are reconstructed here; the Luminance, R/G and B/Y channels). Images are shown beside each stage for illustration.