OBJECT RECOGNITION USING MULTILAYER HOPFIELD NEURAL-NETWORK

An object recognition approach based on concurrent coarse-and-fine mat ching using a multilayer Hopfield neural network is presented, The pro posed network consists of several ;cascaded single-layer Hopfield netw orks, each encoding object features at a distinct resolution, with bid irectional interconnections linking adjacent layers. The interconnecti on weights between nodes associating adjacent layers are structured to favor node pairs for which model translation and rotation, when viewe d at the two corresponding resolutions, are consistent. This interlaye r feedback feature of the algorithm reinforces the usual intralayer ma tching process in the conventional single-layer Hopfield network in or der to compute the most consistent-model-object match across several r esolution levels, The performance of the algorithm is demonstrated for test images containing single objects, and multiple occluded objects. These results are compared with recognition results obtained using a single-layer Hopfield network.