A LABELED OBJECT IDENTIFICATION SYSTEM USING MULTILEVEL NEURAL NETWORKS

This paper describes the design of a neural network based labeled obje ct identification system, to be used for product classification at the final inspection stage of an IBM personal computer manufacturing line . The objective was to design an identification system using existing equipment that would provide robust and accurate classification, as we ll as a simple means for adding new product models to the system. In t he first stage of the identification system, an image of the product i s obtained, and the region containing the label is segmented from the rest of the image. Preprocessing operations are performed to extract t he region of interest from the segmented image. Normalized and preproc essed images of the labels are compressed using a fully-connected back -propagation autoencoder network. Features extracted in this manner ar e used as inputs to a Learning Vector Quantization (LVQ) network, trai ned to classify the labels. The system so designed is shown to satisfy the primary requirements of a typical industrial classification syste m.