期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2000
卷号：XXXIII Part B2
页码：582-589
出版社：Copernicus Publications
摘要：A method for classifying agricultural crops using multi-temporal, multi-spectral and multi-source remotely-sensed data is described. The procedure characterises all the pixels in a scene by considering their intensity values as a function of time of imaging and spectral waveband. An analytical surface is interpolated through these data points, which may be irregularly spaced. Two fitted function interpolation methods were used to generate and parameterise the analytical surfaces. Then, the surface coefficients were input to three different supervised classifiers (Maximum Likelihood, Artificial Neural Network and Minimum Distance Rule algorithms). Results show that classification accuracy is significantly improved in comparison with the use of any single-date image. The advantages of the methodology described in this paper are that it takes account of the reflectance spectra at different points in the growing season, and that the time periods between images, as well as the wavebands, need not be the same at each date. Thus, the procedure can handle data from sensors such as SPOT HRV and Landsat TM. In addition, the use of coefficients to represent the analytical surfaces significantly reduces the amount of data processing, whilst maintaining information reliability
关键词：Multitemporal; Classification; Multisensor/multispectral fusion; Land use; Integration
