遥感图像条带噪声的多尺度变分模型去除  

A destriping method with multi-scale variational model for remote sensing images

作　　者：霍丽君[1,2] 何斌[1] 周达标 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院大学,北京100039

出　　处：《光学精密工程》2017年第1期198-207,共10页Optics and Precision Engineering

基　　金：吉林省科技发展计划资助项目(No.20126016)

摘　　要：多片CCD拼接遥感成像系统由于存在非均匀性问题,导致遥感图像中常存在条带噪声,本文在分析条带噪声的主要来源和模型的基础上,提出了多尺度变分模型的条带噪声去除方法。首先,分析了条带噪声的特点并建立了图像退化模型。其次,结合条带噪声的单向性特点与多尺度分层分解方法构造能量泛函。然后,利用不动点Gauss-Seidel迭代法多尺度分级极小化能量泛函,将条带噪声和图像有用信息分离。最后,对各尺度结构分量和细节分量进行累加,得到去噪图像。实验结果表明:对于周期条带噪声,图像畸变量为2‰,图像辐射质量提升到11.715dB;对于随机条带噪声,图像畸变量为3.3‰,图像辐射质量提升到11.092 5dB。与典型条带噪声去除方法相比,不管是周期条带噪声还是随机条带噪声,本文方法均能够在保证畸变量很小的情况下,将其完全去除,满足遥感图像低畸变量的预处理要求。Non-uniformity often occurs in multi-detectors remote-sensing imaging system,resulting in the existence of strip noise in remote sensing images.A destriping method with multi-scale variational model has been proposed on the basis of the analysis on the main sources and model of stripe noise.First,the characteristics of strip noise have been analyzed and the degradation model of the image has been formulated.Secondly,the unidirectional characteristic of strip noise and multi-scale hierarchical image decomposition have been combined to structure J-functional.Then,the method uses fixed point Gauss-Seidel iterative method to minimize multi-scale J-functional and separate stripe noise and useful information.Last,structural and details component under different scales will be accumulated to obtain the destriped images.The experiment result on real remote sensing images indicates that the image distortion is 2‰ and IF increases to 11.715 0 dB for regular stripe noise;the image distortion is 3.3‰ and the IF increases to 11.092 5 dB for random stripe noise.Compared with typical destriping methods,the method in this paper can ensure that stripe noise will be removed completely and preprocessing requirements of small distortion for remote sensing images will be met,for both regular stripe noise and random stripe noise.

关 键 词：光学遥感 条带噪声 多尺度分层分解 非均匀性校正 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]