摘要
图像融合是解决多源遥感图像信息综合的最有效技术手段,针对不同数据源选择最佳的融合方法是提高图像融合质量的关键。本文在介绍三种新的融合方法:Gram-Schmidt、Pansharp和Ehlers融合基本原理和融合过程的基础上,选择广东新会郊区的QuickBird卫星影像为数据源进行试验。对融合结果从定性和定量角度进行综合评价。定性评价包括色调、纹理和清晰度等;定量评价指标包括均值、标准差、信息熵、平均梯度、相关系数和相对偏差等。所选指标能够反映空间分辨率、信息量和保持融合图像的光谱性质等,以便客观准确地评价融合效果。试验结果表明:三种融合方法都有较高的空间结构信息与光谱信息的保真度。其中,Pansharp方法融合结果图像所含量信息最大,细节信息最丰富;而光谱保真方面:Gram-Schmidt方法最好,其它两种融合方法次之。
Image fusion is the most effective technological means to deal with multi-source remote sensing images. Choosing the best fusion method is a key to improve the picture quality according to different data sources. This paper presents basic principles of three kinds of fusion algorithms ： Gram- Schmidt,Pansharp and Ehlers fusion. Using the QuickBird satellite image data, a test has been made for the suburb of Xin hui, Guanngdong Province. Then the fussing results by the three methods are analyzed and compared in qualitative and quantitative ways, respectively. The qualitative analysis includes hue, texture and definition etc. ,while quantitative evaluation includes average values, standard errors, entropy, average gradient and spectrum authenticity. The test result shows that the three fusing methods have relatively high resolution and better fidelity of spectral information. The Pansharp method can contain the largest amount of information and details. In terms of spectral information fidelity, Gram-Schmidt is the best, while both the Pansharp and Ehlers are slightly inferior.
出处
《地质与勘探》
CAS
CSCD
北大核心
2010年第4期705-710,共6页
Geology and Exploration
基金
新疆与内蒙重点矿区遥感地质调查
