many aspects, such as resolution or dynamic range. This allows us
generate and display images of unprecedented levels of realism but it
also poses new challenges - how to capture, process, store and display
all the new dimensions of visual information. Addressing these
problems requires a comprehensive approach where we need to integrate
the knowledge from imaging, computer graphics, image processing,
signal encoding and human vision research.
In this talk I will discuss four topics in the general area of
perception-motivated imaging, spanning computational photography, eye
tracking, image quality metrics and experimental protocols. In
particular, I will talk about the method of capturing high dynamic
range images in a single exposure using a cross-screen filter to
encode high intensity values in the spatial domain. Then I will
present a method of improving accuracy of eye tracking using
information from computer graphics rendering. The improved accuracy
let us simulate a gaze-contingent depth-of-field effect. Finally, I
will briefly talk about image quality measurements using both
computational metrics (HDR-VDP) and experimental protocols.