多源遥感数据融合方法的新探索

目前常用的几种遥感图像融合方法在不同的应用层面上有各自的优势,但也存在比如光谱信息丢失严重等现象。为此,首先分析对比了传统的4种融合方法,用高分辨率SPOT图像和多光谱TM图像分别运用4种融合方法进行图像融合的研究。在此基础上提出改进的多进制小波包融和方法,并通过主观与客观相结合的质量评价方法对融合效果进行比较。分析结果表明,主成分分析融合要比B rovey融合能够更好的刻画细节;多进制小波包融合所丢失的图像光谱信息为最低,能更好的刻画细节问题;乘积融合法在增强建筑物信息方面有比较突出的优势。

New Exploration for Multi Source Remote Sensing Image Fusion Method

According deferent sources Remote Sensing image and deferent applied purpose,using three levels fusion method-pixel level,character level and decision-making level,it improved information analysing and picking-up ability from image and made more nicety of classification decision-making and positioning.In this thesis,using high spatial resolution SPOT imagery and multi-spectral TM imagery,the author carried out respectively four imagery fusion methods and analytical results indicated that Principal Component Analysis is more detailedly potried the imagery than Brover fusion method,wavelets analysis fusion losing less imagery information than others,product fusion has prominent advantages in enhance imagery information.It concluded that the reciprocal fusion method should be chosen based on deferent applied purpose.