一种基于HT系数预测的遥感图像多分辨融合方法

A HT-based Algorithm of Multi-Resolution Image Fusion

信息融合的概念自 70年代形成后 ,由于其强调应用对象结果的信息优化 ,在许多方面比传统的信息处理方法有更大的优势 ,在遥感领域得到了广泛的应用。本文在分析前人算法的基础上 ,揭示了高斯拉普拉斯金字塔 ( GLP)结构融合方法与传统的哈达玛变换 ( HT)之间的内在联系 ,提出了一种基于 HT系数预测的多分辨融合方法。通过实验表明 ,这一算法对于光谱差异较大的遥感图像的融合 ,在几何分辨率改善的同时可获得较好的光谱保真度。

The notion of image fusion was conceived about 1970s and fusion algorithms pay much attention to information optimization. It has been widely used in image processing of remote sensing because of its superiority over the traditional methods. After having analyzed algorithms of remote sensing image fusion using multi\|resolution analysis(MRA), this paper expatiates on relations between Gaussian\|Laplacian pyramids and Hadamard transform. A new algorithm based on predicting HT coefficients is developed and the experiments are performed multi\|resolution images. The studies show that the algorithm hight MRA/HT has good performance in preservation of the spectral fidelity as well as in resolution enhancement, especially acting on the bands with different spectrum.\; Most of the fusion algorithms of remote sensing image are based on MRA method for the purpose of improving resolution of multispectral images. The fashionable algorithms are divided into two typical classes: Gaussian\|Laplacian pyramids and wavelet\|based algorithms. In course of implementation, we found their common limitations that the algorithms could not work well for two images without spectral superposition. MRA/HT put forward in the paper solved the problem partially. We use panchromatic band image of SPOT to sharpen three multispectral bands in the experiments. As the result, MRA/HT promises to accurately maintain the spectral characteristics of the original low\|resolution image XS3 which has little spectral overlap with panchromatic band image.