Accelerating feature extraction and matching algorithms using GPUs for low vision images is the aim of this work. Feature relationships are built from pixel level features to facilitate subsequent learning and generalization of the object information in an image.
