Colloquium Computing Science

Title: Encoding of high dynamic range video with a model of human cones

Abstract:
Natural stimuli typically encompass a large range of intensities, even within a single scene and for short stretches of time. This range is larger than can be accommodated by biological systems, such as the neurons in the human eye, or by technical systems, such as common display equipment. The human eye solves this problem by dynamic nonlinearities in the cones. In the talk I will present a computational model of this process that describes measured responses from the eye and that is also applicable to encoding and decoding of high dynamic range video. The model can be efficiently implemented on current Graphical Processing Units using fast recursive filters for the model components.