一种基于小波系数特征的遥感图像融合算法

A Remotely Sensed Image Fusion Method Based on Wavelet Coefficient Features

多光谱图像和全色图像是目前卫星遥感领域最常见的传感器图像。为了更充分地发挥这两类遥感图像数据的价值,人们利用两类数据的互补性,将多传感器融合技术引进了遥感图像处理领域。在IHS彩色空间变换和小波多分辨率分析的基础上,利用图像高频小波系数的多个特征来定义特征量积,并利用特征量积作为依据提出了一种图像融合新算法。通过一组多光谱图像和全色图像数据进行融合仿真试验,并将该算法与IHS,HPF等算法和归一化矩算法作了比较。证明该方法能在保留多光谱图像光谱信息的基础上,有效地提高多光谱图像的空间分辨率。

Now multispectral images and panchromatic images are two of multisensor images which are widely used in remote sensing. To adequately take advantage of the high spectral resolution of multispectral images and the high spatial resolution of panchromatic images, multisensor data fusion technique has been proposed in remote sensing image processing. A new method based on transform of IHS and wavelet multiresolution analysis is introduced. Using characteristic moments of the wavelet coefficients of high frequency component under different resolutions,we can define characteristic product. According to the local characteristic product, the new method determines its measurement. Finally computer simulations of image fusion are carried out by a group of multispectal images and panchromatic images. Compared with the result of IHS,HPF fusion method and the normalized correlation moment method, it is proved that this method can get an excellent result for the aim of improving spatial resolution of the multispectal images.