A new method for impulse noise reduction from digital images Based on Adaptive Neuro-Fuzzy System and Fuzzy Wavelet Shrinkage
Sude Tavassoli, Alireza Rezvanian, Mohammad Mehdi Ebadzadeh · 2010
Impulse noise reduction is one of the important processes in the pre-processing of digital images. Most primitive approaches used neighbour pixel values to replacement of noisy pixels. But these methods affected on the all pixels including corrupted noisy pixels and uncorrupted noisy pixels. So the images loosed vital texture such as edges. Recently researchers have been proposed classification based methods, in this case at the first detect noisy pixels then replace only noisy pixels with new value, un-noisy pixels and detected normal texture remain unchanged. This paper proposed novel method, containing two stages, which in the first stage, Noisy pixels detect based on an Adaptive Neuro-Fuzzy Inference System (ANFIS), then in the final stage, the new changes apply on these pixels using the Fuzzy Wavelet Shrinkage (FWS). To illustrate the proposed method, some experiments have been performed on several standard gray level test images, also based quantitative and qualitative criteria were compared with popular methods. The results show that the proposed method relatively has the desirable performance.