Preliminary analysis of tm data for coffee, citrus and sugarcane discrimination in Furnas region, MG, Brazil

Abstract

Coffee, citrus and sugarcane play an important role on the national economics in terms of export products. Furnas region localized in Southeastern Brazil has been characterized by an increased agricultural area, mainly for coffee plantation. Several works have been carried out in Brazil with the objective of discrimination and mapping sugarcane using satellite images, but almost nothing has been done to study coffee and citrus through orbital remote sensing data. In this paper, digital image analysis was applied to TM data, bands 3, 4, 5 and 7 which are sensitive to red (B3), near infrared (B4) and middle infrared (B5 and B7) regions. Spectral discrimination and crop classification was obtained on an interactive image processing system using a maximum likelihood algorithm. Classification matrices, average performance, and percentage of reflectance are presented for all crops and channels. Results achieved for the test site show high level of discrimination (greater than 95%) among the studied crops.