Research On Remote Sensing Image Oblique Stripe Noise Removal Based On Anisotropic Total Variation

Remote sensing technology is a technology to obtain target information from a distant place,and the image obtained by this technology is called remote sensing image.Remote sensing images have a wide range of applications,such as defense military,geological survey,climate and weather forecasting,etc.Remote sensing images are affected by many factors in the process of acquisition,transmission,and preservation,which cause the remote sensing images to be polluted by various noises.The most common type of noise pollution is stripe noise.Stripe noise will seriously reduce the accuracy of remote sensing image data and image visual quality,and cause a lot of trouble for the subsequent processing of remote sensing images.Therefore,before parsing and processing remote sensing images,stripe noise should be removed first.Existing methods for removing stripe noise in remote sensing images can be basically divided into the following types: statistical-based methods,filtering-based methods and model-based optimization methods.In recent years,some people have also proposed deep learning methods to remove stripe noise in remote sensing images.However,these common methods are mostly used to remove horizontal and vertical stripe noise,and rarely target oblique stripe noise existing in remote sensing images.The stripe noise acquired by the sensor is horizontal or vertical in the original remote sensing image,but in the process of producing high-level remote sensing products,those vertical or horizontal stripe noises will be converted into oblique stripe noises due to geometric registration and other reasons.And advanced remote sensing products usually have greater value and practicality in scientific applications,so it is also important to remove the oblique stripe noise problems in remote sensing images.This paper proposes an Anisotropic Total Variation method for removing oblique stripe noise in remote sensing images.The main work of this paper is as follows:1)Explain several commonly used stripe noise removal methods and summarize the current status of stripping remote sensing images.2)To improve the oblique stripe noise removal method based on the Total Variation(TV)optimization model,and propose an Anisotropic Total Variation oblique stripe noise removal algorithm for remote sensing images.This method uses oblique stripe noise to damage the continuity of the original clear image and the degree of horizontal and vertical destruction is different.The smooth nature and global sparse nature of the oblique stripe noise implements reasonable regularization constraints on the oblique stripe noise,to construct an oblique stripe noise removal model.3)A large number of experiments have been carried out on real and simulated data,and compared with several existing better methods to verify the versatility and effectiveness of the proposed method to remove stripe noise in remote sensing images.