A novel random-valued impulse noise detector based on MLP neural network classifier
Saeed Soleimany, Mohammad Hamghalam · 2017
In this paper, a novel random-valued impulse noise (RVIN) detection based-on neural network is proposed. In order to precise noise detection, the feed-forward neural network with back-propagation training algorithm is applied. Five features of noisy images are extracted and used as input of the proposed network. Thus, the uncorrupted and corrupted pixels can be precisely classified. Experimental results confirm the superiority of the method in contrast to similar techniques in the noise detection. Moreover, we find the best compromise between RVIN undetection and misdetection.