Abstract
Abstract: :
Purpose: Magnification is an effective aid for people with conditions causing resolution loss. However, it inherently restricts the field of view. We have developed a system of magnifying television images while addressing the field restriction in two ways. Methods: The system requires a point in each video frame that is to be centered on the screen when the image is magnified. This selection should maintain the most relevant details in the scene. The coordinates of this «center of interest» point can potentially be included with each frame. The eye movement patterns of normal-vision observers were used to determine the centers of interest within scenes. Six ten-minute video segments were selected from a broad range of scenes. To the extent possible, centers of interest were defined in terms of objects. For example, if a view fixated on a person’s nose and then moved fixation to an ear, the object of fixation in both cases was defined as the person’s head. In typical use, the visually impaired person uses a remote control to adjust the magnification («zoom») and can also override the pre-set center of interest («roam»). To maintain wide field context, an edge-enhanced outline of the full scene can be superimposed on the magnified image. Results: A system implementing the magnification and shifting was built and integrated to use the preset center coordinates. A real-time outline derivation system that superimposes an outline of the full frame on the magnified image was developed. Eye movement patterns were analyzed off line. A comparison across groups allowed us to compare differences in areas of interest or eye movement patterns based on age or gender. The location of the center of each object was translated to the center of interest attached to the frame. Modified videos were shown to low vision patients with and without the full field outline. Magnification preferences, override frequency, and satisfaction were recorded. Patients were also interviewed regarding the impact of information in the scene that might have been missed due to the magnification. Conclusion: A demonstration system will be available for use during the poster session.
Keywords: 459 low vision • 429 image processing • 583 space and scene perception