SEMINAIRE DU 5 Mars 2015 – 16H @ LTCI – Salle C49
Data Adaptive Dual Domain Denoising
We present a « last step denoising » method that takes as input a noisy image and as a guide the result of any state-of-the-art denoising algorithm. The method performs frequency domain shrinkage on shape and data-adaptive patches. Unlike other dual denoising methods, the proposed one doesn’t process all the image samples, which allows it to use large patches. The shape and data-adaptive patches are dynamically selected, effectively concentrating the computations on areas with more details, thus accelerating the process considerably. The proposed method also reduces the staircasing artifacts sometimes present in smooth parts of the guide images. The experiments show that this method improves the result of almost all state-of-the-art methods, and this improvement requires little additional computation time.
This is joint work with N. Pierazzo, M. Rais, and J.-M. Morel.
Gabriele Facciolo est Chercheur Postdoctoral à l’Ecole des Ponts et ENS Cachan