Abstract: Memory-based object recognition methods work by comparing an object against many representations stored in a memory, and finding the closest match. However matches are generally made to representations of complete objects, hence such methods tend to be sensitive to clutter and occlusion and require good global segmentation for success. We describe a method that combines an associative memory with a Hough-like evidence combination technique, allowing local segmentation to be used. This resolves the clutter and occlusion sensitivity of traditional memory-based methods, without encountering the space problems that plague voting methods for high DOF problems. The method is based on the two stage use of a general purpose associative memory and semi-invariant local objects called keys. Experiments using keys based on a curve segmentation process are reported, using both polyhedral and curved objects.
Full Publication List Recent Publications