Minimax Parameter Selection Work for NLM To Be Published in IEEE Transactions on Image Processing

Novel parameter selection methods for self-weights in the NLM filter have been investigated using Baranchik’s minimax estimation theorem and the results will be published in IEEE Transactions on Image Processing (IF: 3.735). This work was done by former Master student Minh Phuong Nguyen (now with TU Delft) and Prof. Se Young Chun. We extended the work of Wu et al. and proposed methods to overcome limitations of the previous work such as visual artifacts. We also found that in our methods, even global smoothing parameters in NLM may be able to be estimated without the ground truth images. For more information, please visit



Comments are closed.

Create a free website or blog at

Up ↑