A FAST FUZZY-BASED APPROACH FOR RECOGNIZING OBJECTS
Atr State-key · 1997
A fast fuzzy-based algorithm for recognizing a number of objects is proposed. It uses some two-dimensional invariant features and forms the object models by learning and analyzing the sample sets. The member degree function of an object to be recognized is developed according to its feature parameters and the attributes of a model. At last that identification can be made by the max-min rules in the fuzzy mathematics. The system with the new method is used to one hundred object classes, and 99% correct rate is obtained. It shows the method developed is very efficient.