QAAD: Quality Aware Adaptive Denoising
Tejas Ajay Parse, Tanishq Awasthi, Dushyant Yadav, Piyush Joshi · 2024
Image denoising is a fundamental task in computer vision and image processing, crucial for improving the visual quality and interpretability of images captured in noisy environments. In this research, we propose a quality-aware adaptive denoising (QAAD) approach that leverages the Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE) as a key factor in determining the optimal denoising algorithm for an image. Our methodology involves an initial assessment of image quality using BRISQUE scores for a diverse set of images. Subsequently, we employ various image quality enhancement algorithms to enhance the image quality. Through extensive experimentation, we assess the performance of these enhancement algorithms with respect to specific ranges of BRISQUE scores. This evaluation enables us to determine which denoising algorithm performs optimally for images falling within particular BRISQUE score ranges. The innovation in our approach lies in its adaptability. When an image proceeds for denoising, its BRISQUE score is calculated and, based on this score, the most suitable denoising algorithm for that specific image is selected. This adaptive strategy ensures that the chosen denoising algorithm is tailored to the inherent noise characteristics of the input image, leading to superior denoising results. Our experimental results demonstrate the effectiveness of our approach in enhancing image quality and achieving state-of-the-art denoising results. This research contributes to the field of image processing by providing a data-driven and adaptive solution for image denoising, ultimately enhancing the quality of images in diverse real-world applications.