PERFORMANCE BOUNDS FOR TRACTABLE POISSON DENOISERS WITH PRINCIPLED PARAMETER TUNING
Chinmay Talegaonkar, Ajit V. Rajwade · 2018
We present an algorithm for image denoising under Poisson noise using the theory of variance stabilization transforms. We derive worst-case performance bounds for our algorithm. Our proposed estimator allows for easy and very principled parameter tuning unlike existing approaches which require specification of signal dependent parameters. Moreover our estimator is computationally tractable. We also demonstrate numerical results on image denoising under Poisson noise to support the theoretical results.