An investigation of the selection of texture features for crop discrimination using SAR imagery.
Soares J. V., Rennó C. D., Formaggio A. R., Yanasse C. C. F., Frery A. C.
Author Affiliation: INPE-Instituto Nacional de Pesquisas Espacials, Caixa Postal 15, 12201-970 Sáo José dos Campos, SP, Brazil.
Remote Sensing of Environment 59 : 234-247
Abstract : A methodology is presented for the selection of texture measures to maximize the discrimination of agricultural land use classes in spaceborne synthetic aperture radar (SAR) images. The images were acquired during the first flight of the Shuttle Imaging Radar-C (SIR-C) experiment during April 1994, over the semiarid region of the Bebedouro, Brazil, where a government scheme aims to irrigate over 1 million hectares for agricultural use. Individual farms ranged from 5 to 12 ha and crops included mangoes, vines, tomatoes and watermelons, with land types such as grazing pastures, bare soil and shrubland. L (24 cm)- and C (5 cm)-band SAR data at HH (horizontal transmitting and receiving), HV (horizontal transmitting, vertical receiving), and VV (vertical transmitting and receiving) polarizations were analysed both in ground range and slant range and on 2 different passes. The kappa statistic was used to identify meaningful texture measures and to discriminate the 7 classes of crop and land type listed above. It was observed that the classifications of land use based only on tonal averages produced a kappa coefficient only slightly higher than 0.5. A kappa threshold of 0.9 was reached with the simultaneous inclusion of 15 texture measures for the 6 images (2 bands, 3 polarizations). It was also observed that the inclusion of texture features when only one band and one polarization was used could produce kappa values higher than 0.85.