We present a novel visual representation for prosthetic vision that augments intensity in order to emphasise regions of structural change. This is achieved via the adaptation of a recently proposed method for measuring the extent of local variation of surface orientation in corresponding disparity images. The proposed visual representation demonstrates how intensity and depth data may be combined to provide a scene representation that shows visual appearance as brightness in the familiar way (i.e., intensity-based), but ensures structurally important features such as steps, doorways and drop-offs, as well as general items of interest remain perceivable, regardless of contrast. Qualitative comparisons of the proposed visual representation in simulated prosthetic vision (98 phosphenes) suggest potential advantages over non-augmented intensity for distinguishing between free and obstructed space in the scene, and for perceiving features of interest on smooth surfaces.
展开▼
