Recognizing Po1:yhedral 0 bjects using Neural Networks*
Humberto Sossal, Jesus Figueroa-Nazuno · 1995
Two different Neural Networks: the Backpropaga.tion (RPNN) and the Generalized Regression (GRNN) Neural Networks were used to solve the polylideral object recognition problem. Some comparisons between these two NNs with the absence and presence of noise -characterized by the absence of significant segments or the presence of spurious ones- were done. The proposed schemes (using either a BPNN or the GRNN) coiisist of t’wo phases: model building and object recognition. During the moc!el building process, each characteristic view (CV) of t,he object is described by a feature vector containing the normalized distances from each CV‘s vertex to the CV’s centroid. During the object recognition pha,se, a. set of vectors is used to train the NN. Finally a vector representing one of’ tdlie object,’s CVs is presented to the NN tto t,est its performance as a classifier.