Gradient and Multi Scale Feature Inspired Deep Blind Gaussian Denoiser

In this paper, a novel deep blind Gaussian denoising network is proposed utilizing the concepts of gradient information, multi-scale feature information and for removing additive white noise (AWGN) from images. The consists two modules where in first module generates an intermediate image whose concatenated with features second to generate final residual image. Subtracting noisy gives desired denoised block used middle enhances usage image, together multi scale block, further contributes quality Experimental results show superior performance method comparison several state art classical learning based methods like EPLL, BM3D, WNNM, DnCNN, MemNet, BUIFD, Self2Self ComplexNet by decent margin (an improvement up 2.4dB PSNR, 0.07 SSIM 0.03 FOM index best performing model) when experimented over BSD68, Set5, Set14, SunHays80 Manga109 databases.