Multi-source image classification using color attribute profiles

This work introduces a method to extract attribute profiles from RGB color images with high spatial resolution, for instance images acquired from Unmanned Aerial Vehicles (UAV). The resulting Color Attribute Profiles (CAP) are intended to improve the classification of low spatial resolution hyperspectral images by merging the attribute features with the spectral features of the hyperspectral image. Instead of treating the R, G and B bands separately, the color image is transformed into CIE-Lab space. In this color space, attribute profiles are extracted from the `L´ band, while the `a´ and `b´ bands are kept intact, and the resulting images are transformed back into RGB space. In our experiments, classification results using this methodology are compared to classification results using other strategies for extracting attribute profiles in CIE-Lab space, as well as regular grayscale attribute profiles.