Adaptive neuro fuzzy estimation of the most influential speckle noise distributions in color images for denoising performance prediction

This research paper analyzes the speckle noise distributions in images for denoising performance prediction through the prism of spatial domain. The values at the maximum, minimum and middle of spectrum in spatial domain are taken as reference values. All obtained results give a better overview of the “nature” of the digital images in comparison to the theoretical definitions of noises and images as digital signals. Therefore, analyses of the noises in the 2D spectrum give good recommendations for improvement of the filters. The main aim in this study is to investigate which speckle noise distributions in images has the strongest influence for denoising performance prediction. The clean images are available and we adopt it for evaluating our network. In our experiments, Peak Signal to Noise Ratio (PSNR), normalized color difference (NCD), and feature similarity index for color image quality assessment (FSIMc), are used to measure denoising performance. is selected as the evaluation index of the image. Studies on speckle noise distributions in images show that such distribution do have certain disciplines. ALOHA filter is the most influential for the denoising performance prediction.