多通道SAR图像滤波的向量总变分模型

Vectorial total variation model for multi-channel SAR image denoising

研究多通道SAR图像滤波的向量总变分模型与算法.通过引入向量总变分模型,设计了模型快速求解的不动点迭代算法,证明了迭代算法的收敛性.通过改进不动点算法迭代过程中滤波系数的构造方法,建立了保分辨率的多通道SAR图像的自适应向量总变分模型,给出了相应的迭代求解算法和收敛定理.通过处理RADARSAT-2多极化、多时相图像分析了本文模型的滤波性能和分辨率保持性能,并验证了本文迭代算法的收敛性和快速收敛速度.

The vectorial total variation model and algorithm are studied for multi-channel SAR image denoising.After introducing the vectorial total variation model,an accelerative fix-point iterative algorithm was proposed and its convergence was proved.By improving the filter coefficient of the fix-point iterative process,an adaptive vectorial total variation model was developed for multi-channel SAR image denoising,whose iterative algorithm and convergence theorem were present.The performance of denoising and resolution preservation of our models was tested by multi-polarimetric,multi-temporal RADARSAT-2 images,in addition to the validation of the convergence and the convergent speed of the proposed algorithms.