Recognition of 2-D Objects by Optimal Matching
Lu Liu, Fang Luo, Nanno J. Mulder · 1994
Abstract: This paper introduces an approach of recognizing 2-D objects by optimal matching. The method consists of two stages: object identification and object localization. Both of them are accomplished through optimal feature matching, in which the radiometric distribution of an object as a global feature extracted from an image is matched directly to the object model. A cost function is defined as a quantitative evaluation of the feature fitting and the recognition process is based on cost minimization. In this method, every subproblem in object recognition is formulated as an optimization problem and techniques of optimization are utilized to solve these problems. Key Words: object recognition, optimization, cost function, object modelling. 1.