| Compared with the traditional natural images,the remote sensing images integrate rich spatial,spectral,and radiation information.The images are analyzed to deliver useful products to urban monitoring,fire detection,and flood prediction.Remote sensing im-ages are often degraded by various types of noises,the most common of which is stripe noise.The stripe hides the ture information in the observed image,which affect the image visual quality and limit the subsequent processing tasks.Therefore,image destriping has become an important branch of remote sensing image processing.In this paper,we study the remote sensing image destriping problems.Based on a full consideration of the stripe properties,we propose the new mathematical model by combining with the regularization methods and apply the efficient numerical algorithms to solve the proposed model.The main contributions are listed as following:Most existing destriping methods concentrate on estimating the clean image directly from the observed image while paying little attention on the stripe component.Beside the directional and structural properties of the stripes,the stripe position is also an important feature.Accurate stripe detecting is conducive to destriping,which in turn is conducive to accurate stripe detecting.In this paper,we propose a novel destriping model that can detect the stripe position and remove the stripes simultaneously,by fully exploiting the properties of the stripes.Our model exploits both the structural and the directional properties of stripes.On the one hand,we use the?w,2,1-norm to characterize the joint sparsity of stripes.In addition,we use the iterative support detection（ISD）to calculate the weight vector in?w,2,1-norm,which shows the stripe position in the observed image.On the other hand,we use uni-directional total variation（UTV）to characterize the directional property of stripes.We apply the alternating direction method of multipliers（ADMM）.In the solving process,we achieve stripe detection by calculating weight vector.Meanwhile,we design new indices to analyse the accuracy of detection.Comparison results on simulated and real data sug-gest that the proposed method can remove and detect stripes effectively while preserving image edges and details. |
