A low cost methodology for multispectral image classification

Multispectral and hyperspectral remote sensing have significantly improved territorial surveys and mapping. However aerial images are often expensive being acquired through aircraft and satellite sensors. Furthermore, the processing and classification of these images need commercial software that increases the entire cost of the analysis. For these reasons, we propose an approach of data acquisition and analysis based on supervised classification to obtain accurately maps of the area of interest in reduced time. The images have been acquired through 3-channels Tetracam ADC-Lite camera, and processed with free and open source software, PixelWrench2 and QGIS. The results obtained demonstrate that the approach can compete with traditional acquisition and classification methods, due to simple operational procedures, low operational costs, and high accuracy of supervised classification. This approach provides promising results that encourage its development and optimization of these technologies for other purposes, such as the mapping of asbestos-cement (AC) roof coverings.