道路信息提取的GF-2影像融合方法对比分析

Comparison and analysis of GF-2 image fusion methods for road information extraction

本文针对适用于GF-2卫星影像道路信息提取的影像融合方法,进行了对比与分析。通过对GF-2影像的全色及多光谱数据进行融合实验,比较分析了6种常用的影像融合方法。从主观定性评价和客观定量评价两方面进行分析,筛选出融合后影像质量较优的方法;然后选取较优融合方法进行道路信息提取,并对比了道路提取的精度,得出最适合GF-2影像的道路信息提取融合方法。结果表明,6种融合方法中NNDiffuse变换和GS变换融合质量较好,基于NNDiffuse融合的影像的道路信息提取精度高于GS变换。因此,NNDiffuse变换相比于其它5种融合方法更适用于GF-2卫星影像在道路信息提取中的应用。

In this paper,the image fusion methods for road information extraction from GF-2 satellite images are compared and analyzed.Through the fusion experiment of panchromatic and multispectral data of GF-2 image,six commonly used image fusion methods are compared and analyzed.The methods with better image quality after fusion are screened based on two aspects of subjective qualitative evaluation and objective quantitative evaluation.Then the optimal fusion method is selected for road information extraction by comparing the accuracy of road extraction,and the most suitable road information extraction fusion method for GF-2 image is obtained.The results show that NNDiffuse and GS transform have better fusion quality among the six fusion methods,and the extraction accuracy of road information based on NNDiffuse is higher than that of GS transform.Therefore,NNDiffuse is more suitable for the application of GF-2 satellite image in road information extraction compared with other five fusion methods.