Infrared Image Stripe Noise Removal by Solving First-And Second-Order Total Variation Inverse Problem

Stripe noise is a prominent type of noise in infrared imaging systems. It has a unique directional property where it usually appears as vertical lines imposed on the image. This paper introduces an algorithm that turns an image denoising problem into solving inverse problems for each row of the noisy image, giving users a powerful tool for destriping infrared images. This was done by integrating first- and second-order total variation (TV) regularizations, which has led to excellent denoising capabilities around edges and smooth areas. The algorithm's performance is carefully examined, along with comparisons to other denoising techniques, using simulated and real-world data, emphasizing the algorithm's superiority in terms of noise reduction and reconstruction accuracy. Comparing the proposed method to state-of-the-art techniques, it shows greater performance in improving the visual quality of the produced image.