Image Reconstruction

Here we show the quality of our approximation of convolution with the Green's function, which is used to solve the Poisson equation with a compactly supported function in right-hand side. Divergence is the input to the problem and we use the original image as the ground truth. Reconstruction A shows the result of the reconstruction using separable filters of size 7x7 for h1, h2 and 5x5 for g .

Reconstruction B shows the result of the reconstruction using separable filters of size 5x5 for h1, h2 and 3x3 for g. Error A and B show absolute differences (magnified x50) between the reconstructions and original image.

Instructions: Roll your mouse over each thumbnail to see a larger version of the image appear below. This makes it possible to easily flip back and forth between the images. Click on a thumbnail for a full resolution version of the image.

 


Original Image Reconstruction A Reconstruction B
Divergence Error A Error B
1 1 1

1