Rethinking sampling topologies for image quality estimation in computational imaging system design

Optical design includes an optimization process that evaluates image degradations caused by the lens design in each iteration. In joint digital-optical imaging system design the optimization cost function needs to include measured image quality after optical and digital processing. A commonly used merit function is the mean-squared-error (MSE) of the digital image after restoration. Low resolution rectangular sampling grids increase the speed of the optimization process, but cause approximation errors of the MSE merit function. Through adaptation of the MSE calculations to flexible non-rectangular sampling topologies, we achieve increased MSE accuracy and significant speed-up of the processing time.