Real-time bio-inspired contrast enhancement on GPU

This paper describes a new contrast enhancement technique inspired in the way the human retina works to increase the degree of detail, mitigate glares and enhance the visibility of the scene in both bright and dim lighting conditions. The proposed operator and its real-time implementation are specially aimed as an aid system for visually impaired people who struggle to manage themselves in environments where illumination is not uniform or changes rapidly. This aid device processes in real-time, with minimum latency, the input of a camera and shows the enhanced image on a Head Mounted Display (HMD). To meet all these requirements we have implemented the whole model in a light-weight netbook which integrates a GPU NVIDIA ION2 and performs at a maximum rate of 27 frames per second when working with VGA resolution images (640×480 RGB images).