基于改进的自组织映射网络的遥感影像融合研究

MODIFIED KOHONEN NETWORKS BASED REMOTE SENSING IMAGE FUSION

多源遥感影像融合是富集遥感海量数据的最有价值的技术手段。本文给出了一种新的基于改进的自组织映射网络的遥感影像融合模型。选择浙江绍兴为典型研究区 ,以LandsatTM( 1 0m)与SPOT 4Pan ( 1 0m)融合数据为例 ,进行了融合实验与分析。实验结果表明 ,应用基于改进的自组织映射网络模型进行融合 ,分类融合结果较好 ,较基于基本自组织映射网络的影像融合分类精度提高约 8%。

The paper puts forward a modified Kohonen networks based remote sensing image fusion model.Chosen the Shaoxing city,Zhejing Province,as the study site,a fusion experiment was done using Landsat TM(10m)data and SPOT-4 Pan(10m)data.The result shows that the fusion image using the modified Kohonen networks in better than that using the Kohonen networks only,and the classification accuracy is improved about 8 percent.