| Image information is an important means for human to obtain information.Due to the limitation of sensor technology,multi-sensor imaging systems often suffer from the problem of stripe noise and other issues during the imaging process.Stripe noise degrades the imaging quality and brings great difficulty to the sequent image processing tasks.In recent years,domestic and foreign scholars have put forward many kinds of image stripe noise removal methods,but many of these methods have some deficiencies in the effectiveness of eliminating stripes,universal,simplicity and so on.The aim of this paper is to propose methods which have characteristics of universality,effectively removing stripe noise and preserving the information of original image as much as possible.In this paper,destriping methods are researched and the main contents are summarized as follows:This paper firstly introduces the research background and summarizes the domestic and foreign relevant literature on image stripe noise removal.Then the mechanism of production and characteristics of stripe noise are analyzed.The stripe degradation model and stripe-degenerated analog image are researched.Moreover,because unidirectional variational algorithm destroys the image information consistent with the direction of stripe noise,we propose a destriping method by combining the global sparsity and unidirectional variation.This method can be used to deal with the images produced by different imaging systems,which has common adaptability.At last,considering the high spectral correlation between the remote sensing images in distinct bands,we put forward a new destriping method based on the low rank matrix decomposition and unidirectional variational algorithm.And two proposed methods are compared with several state-of-the-art destriping methods.The experimental results show that two new methods could both effectively eliminate stripe noise and preserve the detail information of original image.Both methods perform well in image destriping. |
PDF Full Text Request