- Wed Aug 29, 2007 2:26 am
Most range transducers such as ultrasonic devices would be limited by range, and would require some mechanical method for scanning, which could get quite complex. They also draw a lot of power, loose accuracy relatively quickly, and take some amount of time to capture distance information.
Perhaps an idea that would work is to take two consecutive images, changing the lens zoom a small amount on the second capture. Then software could create a sort of topographic grayscale image that distinguished layers based on the interposition of objects in the two images. For example, focus on an object in front of you, then move your head towards the object a little bit. There will be continuity between 'layers' in the image you see, which could be exploited using existing 'wand' tools that use frequency filtering to select similar regions of an image. I don't know if the idea is good at all, since I am unfamiliar with how image processing really works. Maybe ill play with the idea at school next year.