Open-ball-operators for 3-D object recognition

Recognition of three-dimensional objects is a crucial and challenging problem. This paper presents a method of three-dimensional object recognition using the Hopfield associative memory in neural networks. The input vectors to the Hopfield associative memory are obtained via the three-dimensional feature extraction open-ball operator (OBO). This approach is invariant to shift, translation, and rotation (R/sup 3/) of three-dimensional objects.