A Novel Approach for Image Denoising Based on Artificial Neural Networks
Şeref Sağıroğlu, Erkan Beşdok · DergiPark (Istanbul University) · 2012
This study presents a novel approach based on artificial neural networks (ANNs) to remove noises from defected images.ANNs were trained with two different learning algorithms, Levenberg-Marquardt and Extended-Delta-Bar-Delta, for speeding upthe training and feedforward calculation processes. The restored results were also compared to the classical techniques, FFT,Wiener+Median filtering and wavelet denoising. The results were shown that the proposed novel neural model providessimplicity and accuracy to remove noises from defected images without estimating any mathematical model than the others