Amigo: A Tool For The Generation Of Synthetic Hyperspectral Images

摘要

The lack of publicly accessible hyperspectral image repositories with properly labeled and characterized images for the design and validation of processing algorithms, in particular, classification algorithms, is a longstanding problem in the field. In order to mitigate the need for benchmark libraries of real hyper-photographic material with reliable labeling, this work presents an tool, called Automatic Multi-dimensional Image GeneratOr (AMIGO) designed and implemented to automatically create synthetic multidimensional images based on spatio-spectral parametric descriptors that can be used to simplify and facilitate the adjustment and testing of classification algorithms. This tool has been used to properly train and validate classification and denoising algorithms in many research works found in literature showing satisfactory results.