Classification And Suppression Of Blending Noise Using CNN
Rolf H. Baardman · Proceedings · 2018
In this abstract a novel machine learning deblending algorithm is introduced. The method uses a convolutional neural netork (CNN) to classify data patches in a "blended" and a "non-blended" class. A second, regression based, CNN deblends the "blended" patches. Results are shown for a synthetic data example.