Fuzzy and CNN Based Novel approach for Real Time Image De-noising

Abstract

Abstract The discriminative Model for image de-noising has been recently attracting considerable attention due to its favorable de-noising performances. In this paper, a model has been constructed for image de-noising for the standard dataset and real-time dataset. In this paper a novel approach has been implemented known as adaptive fast and flexible de-noising network (AFFDNet). This model is totally based upon the fuzzy and CNN network, with the combination of both image de-noising has been done. Our AFFDNet model is able to capable for handling the different noise level as well as different image sizes. The noise level has been taken 15, 25 & 50 with image size 256*256, 512*512 & 1024*1024. Our model give better result at all noise level also on different size of images. The residual learning is used in this novel approach to remove the latent clean image from hidden layer. In the future scope it can be used in for reduce the complexity because real-time makes algorithm very complex.