An approach to three-dimensional object recognition using a hybrid Hopfield network

A neural network approach is presented as a solution to the three-dimensional object recognition problem. Specifically, a hybrid Hopfield network previously used to solve two-dimensional occluded object recognition problems is adapted to the three-dimensional object recognition problem. The authors proceed under the assumption that feature extraction has yielded a set of vertices for the model and a set of vertices for the input object. From these vertices local and relational features are proposed for use in a hybrid Hopfield network graph matching algorithm. Finally, three-dimensional single input object recognition is realized