基于矩匹配和变分法的MODIS条带去除模型  

Destriping Model of MODIS Images Based on Moment Matching and Variational Approach

作　　者：胡宝鹏 周则明[1] 孟勇[1] 蒋林[1] 

机构地区：[1]解放军理工大学气象海洋学院,江苏南京211101

出　　处：《红外》2014年第11期28-36,共9页Infrared

基　　金：国家自然科学基金资助项目(41174164);公益性行业(气象)专项项目(GYHY201306068)

摘　　要：针对中分辨率成像光谱仪(Moderate Resolution Imaging Spectroradiometer,MODIS)影像条带噪声的特点,提出了一种结合矩匹配和变分法的条带去除模型。首先采用矩匹配方法去除多行和宽行条带噪声;然后,基于改进的条带检测方法定位矩匹配后的残余条带;最后,基于变分条带去除模型,使用Split Bregman迭代求解以去除条带。实验结果表明,该方法能够有效地去除MODIS影像中的条带噪声,并能有效地保持图像中的细节信息。与直方图匹配、矩匹配、低通滤波及单向变分去条带算法相比,本文模型的去噪效果更理想。According to the features of stripe noises in images from the Moderate Resolution Imaging Spectroradiometer（MODIS）,a destriping model which combines moment matching with the variational method is proposed.Firstly,the multi-line and wide-line stripe noises are removed by using the moment matching method.Secondly,the residual stripes after moment matching are located by using the improved stripe detection method.Finally,the stripe noises are removed by Split Bregman iteration on the basis of the variational destriping model.The experimental result shows that the method can remove the stripe noises in MODIS images effectively while preserving the details of the image.Compared with the destriping algorithms such as histogram matching,moment matching,low-pass filter and unidirectional variational destriping algorithms,the proposed model has more ideal denoising effectiveness.

关 键 词：MODIS影像 条带噪声 矩匹配方法 变分条带去除 图像处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]