Research on Improved Joint Denoising and Demosaicing Algorithms

In the process of image processing, image denoising and image demosaicing have been the research hotspots and problems. Denoising and demosaicing are two adjacent important modules in Image Signal Processing (ISP). The existing traditional joint denoising and demosaicing algorithms have problems such as poor image quality and high complexity. This paper proposes a joint denoising and demosaicing algorithm that combines the wavelet threshold denoising algorithm and the nonlocal means (NLM) algorithm. The experiments show that this method has some improvement in time efficiency and objective numerical peak signal-to-noise ratio (PSNR)/Structure Similarity Index Measure (SSIM) metrics under different levels of noise compared to two traditional algorithms.