Improved Adaptive Type-2 Fuzzy Detection and Simple Linear Regression-Based Filter for Removing Salt & Pepper Noise

Image denoising has gained in relevance as a component of image preprocessing due to the increased use of digital images in a range of applications, as well as the degradation of image quality caused by noise introduced by unavoidable occurrences. This work suggests a novel two-stage filter to remove salt and pepper noise from the images. It operates in two stages, the first stage uses an enhanced adaptive type-2 fuzzy noise identifier to identify the corrupted pixel, and the second stage uses a simple linear regression-based approach filter to denoise the corrupted pixel. We first identify a pixel as corrupted or uncorrupted using an improved adaptive type-2 fuzzy-based Gaussian membership function with variables both mean and variance for a specific corrupted image frame. The second step is denoising the damaged pixel using a linear regression-based technique. Herein, we propose a novel co-design method that uses the Gaussian membership function for detection and a linear regression-based denoising technique without any parameter tuning, resulting in better time efficiency. We validate the proposed improved adaptive type-2 fuzzy detection and linear regression-based filter (IAFDLRBF) on a variety of standard images and real-time images with varying noise density. We compare the simulation results with various state-of-the-art methods in terms of various assessment metrics. The results demonstrate the effectiveness of the proposed filter even at high noise densities by providing better detail and edge preservation of an image.