Evaluation of four digital classifiers for automated cartography of local soil classes based on reflectance and elevation in Mexico

被引：7

作者：

Cruz-Cardenas, G. ^{[1
]}

Ortiz-Solorio, C. A. ^{[1
]}

Ojeda-Trejo, E. ^{[1
]}

Martinez-Montoya, J. F. ^{[2
]}

Sotelo-Ruiz, E. D. ^{[3
]}

Licona-Vargas, A. L. ^{[4
]}

机构：

[1] Colegio Postgrad, Texcoco 56230, Mexico State, Mexico

[2] Colegio Postgrad, Salinas 78620, San Luis Potosi, Mexico

[3] Inst Nacl Invest Forestales Agr & Pecuarias, Texcoco 56230, Mexico State, Mexico

[4] Univ Autonoma Chapingo, CRUO, Huatusco 94100, Veracruz, Mexico

来源：

INTERNATIONAL JOURNAL OF REMOTE SENSING | 2010年 / 31卷 / 03期

关键词：

PHOTO-INTERPRETATION; NEURAL-NETWORKS; CLASSIFICATION; KNOWLEDGE; MAPS;

D O I：

10.1080/01431160902894491

中图分类号：

TP7 [遥感技术];

学科分类号：

081102 ; 0816 ; 081602 ; 083002 ; 1404 ;

摘要：

The objective of this research is to evaluate the precision and accuracy of local soil class maps generated with four computer algorithms: minimum distance, parallelepiped, maximum likelihood, and artificial neural networks, using digital elevation models and spectral signatures of local soil classes as input data. The study was done in the states of Mexico, San Luis Potosi, and Veracruz. Statistical binomial proportion tests were done to compare the difference between maps' precision and accuracy. The conclusion was that the combination of reflectance and elevation improved the quality of soil class maps produced by CAC, due to the reflectance variation of local soil classes according to altitude, which helped to better identify them. The best precision was 84% and the best accuracy was 30%.

引用

页码：665 / 679

页数：15