自然资源调查监测中国产卫星遥感技术运用分析

Application of Satellite Remote Sensing Technology in Natural Resources Survey and Monitoring

研究一种利用国产高分卫星遥感数据进行自然资源调查的大数据应用算法,重点研究神经网络深度迭代回归算法在卫星遥感图像像素级分析过程中的地籍边界确认算法,将新算法与高分卫星大数据工具包自带地籍边界划分算法进行对比,发现:自带软件较革新软件,在林木种类误判数量上高出6.8倍,在农作物类型误判数量上高出19.2倍,在水产类型误判数量上高出4.1倍。革新软件对比自带软件,其大资源区边界精度提升65.9%,小资源区边界精度提升67.2%,综合分析其边界划分精度提升62.5%。该结果t<10.000,P<0.01,具有显著的统计学差异,该革新算法可以大幅度提升资源调查效率和资源区划分精度。

Studies at a big data application algorithm for natural resources investigation using high resolution satellite remote sensing data made in China.This paper focuses on the identification of cadastral boundary in the pixel level analysis of satellite remote sensing images using neural network depth iterative regression algorithm.By comparing the new algorithm with the cadastral boundary division algorithm of Gaofen satellite big data toolkit,it is found that compared with the innovative software,the number of misjudged forest species is 6.8 times higher,that of crop type is 19.2 times,and that of aquatic product type is 4.1 times higher.Compared with the self-contained software,the boundary accuracy of large resource area is improved by 65.9%,that of small resource area is increased by 67.2%,and the boundary division accuracy of comprehensive analysis is improved by 62.5%.The results were T<10.000,P<0.01,with significant statistical difference.The innovative algorithm can greatly improve the efficiency of resource investigation and the accuracy of resource division.