Object recognition based on graph matching implemented by a Hopfield-style neural network

Abstract
A model-based object recognition technique is presented. For each model, distinct features such as curvature points are extracted, and a graph consisting of a number of nodes connected by arcs is constructed. Therefore, each node in the graph represents a feature that has an assigned feature type and a numerical feature value, and an arc between two nodes shows the relationship or compatibility between the features such as distances between feature points. Objects recognition is formulated as matching a model graph with an input image graph. A Hopfield binary network is implemented to perform a subgraph isomorphism to obtain optimum compatible matching features between graphs. The compatibility is defined so that it will tolerate the ambiguity of preprocessed features. The algorithm is also extended to detect one object among several objects which could be touching or overlapping. Some simulation results are shown to evaluate the performance of the system.

This publication has 17 references indexed in Scilit: