边缘增强的MAP超分辨率遥感影像融合

Remote Sensing Image Fusion Using MAP Super-resolution with Enhanced Edge

遥感影像普遍存在光谱分辨率和空间分辨率不能兼顾的问题,影像融合技术可在保持高光谱分辨率的同时提高空间分辨率。基于MAP的超分辨率重建是一种新的融合方法,但是嵌入的Huber先验模型,虽然可以使重构的高分辨率图像具有较好的细节和边缘保持能力,却不能强化细节边缘。针对此问题,该文提出一种针对遥感影像边缘增强的MAP超分辨率融合算法,该算法首先对各低分辨率图像进行Laplace边缘增强预处理,然后针对边缘增强的低分辨率图像进行基于Huber-MRF MAP重构算法。实验结果表明所提的融合算法是有效的。

The spacial resolution of multispectral Remote sensing image is often low. There are many fusion method to be employed to resolve this problem. The MAP based superresolution reconstruction is a promising fusion algorithm. But the added Huber prior model which can preserve more edges and details of image can not enhance edges and details. Aiming at this problem, this paper proposed a MAP based image superresolution reconstruction using edge enhanced preprocessing algorithm, which preprocesses every low resolution （LR） image by Laplace enhancement kernel and then fuses high resolution （HL） image from enhanced LR images using Huber-MRF MAP method. Experiment shows the proposed algorithm is effective.