Classification of Object using Fuzzy Reasoning and Maximum Likelihood in Visual Information Processing.
Toshiro Noritsugu, Hong Li Gao · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 1992
This paper proposes an algorithm for the classification of overlapped objects using fuzzy reasoning and maximum likelihood in visual information processing, which can decrease the amount of calculation, simplify the classification procedure and improve reliability. Fuzzy reasoning is used for estimating the contour category with incomplete contour feature information. Maximum likelihood is used for estimating the surface shape class of the surrounding contour based on the local range data acquired with the laser range finder. Each area surrounded by contours is classified under an object class by integrating the results of fuzzy reasoning and the estimation based on the maximum likelihood. Finally, verification at object level is performed by integrating the relationships between the contours of the same object and geometric relationships between the surfaces of the object. Experiments proved the effectiveness of the proposed algorithm.