Cloud implementation of logistic regression for hyperspectral image classification

Classification of remotely sensed hyperspectral images is a challenging task due the enormous amount of information comprised in these images, that contain hundreds of continuous spectral bands. This creates a need to develop new techniques for hyperspectral classification using high performance computing architectures. Despite the availability of multiple algorithms adapted to parallel environments (such as multicore computers or accelerators like field programmable gate arrays or graphics processing units, the application of cloud computing techniques has not been as widespread, although there are many potential advantages in exploiting cloud computing architectures for distributed hyperspectral image analysis. In this paper, we present a cloud implementation (developed using Apache Spark) of a successful technique for hyperspectral image classification: the multinomial logistic regression probabilistic classifier. Our experimental results suggest that cloud computing architectures allow for the efficient classification of large hyperspectral image data sets.