Super-resolution on synthetically downsampled images

Reset to Original vs Set as Left Image

Ground truth IID Gaussian AR Gaussian IID Sparse AR Sparse λ=0.01 λ=0.001 True kernel Estimated kernel

These results illustrate super-resolution on synthetically downsampled images. As the eight test images are individually small, results for all images are shown together. There are two versions of the synthetic images for each combination, differing in how much Poisson noise is added: S = 100 is less noisy and S = 10 is more noisy. Super-resolution is performed both with the true kernel and kernels estimated assuming correlated image gradients (assuming at most 5 x 5 kernels, with elements less than 1% of the maximum subsequently set to 0).

λ is the regularization parameter; higher values give more importance to the prior. The sparse prior refers to the hyper-Laplacian distribution for image gradients with parameter α = 0.8. The AR priors assume that the image gradients are correlated (according to a simple independent 2-D auto-regressive model with correlation parameters 0.3 and 0.6). The IID prior assumes they are independent.