Study on machine vision fuzzy recognition based on matching degree of multi-characteristics

Abstract

This paper presents a new method used for fruit category recognition based on machine vision and total matching degree of fruit's multi-characteristics. The ladder membership function was used to express each characteristic. The matching degree of each characteristic was calculated by its membership function, and then the total matching degree was calculated, fruit category recognition can be determined by the total matching degree. In this paper, a 5-input 1-output zero-order Takagi-Sugeno fuzzy neural network was constructed to achieve non-linear mapping between fruit characteristics and fruit type, then the parameters of membership function for each characteristic was designed as learning parameters of the network. Training the fuzzy neural network through a large amount of sample data, the corresponding parameters of the membership functions of recognized fruit can be determined. Taking apple recognition as an example, the experimental results show that the method is simple, effective, highly precise, easy to implement. © 2010 Springer-Verlag.