Component Based Representation Using Probabilistic Neural Network Classifier
G Anjali · IOSR Journal of Computer Engineering · 2014
The law enforcement agency make use of face recognition system for the identification of criminals.In some cases, the photograph of criminals is not available.In this situation,sketch is drawn by the information provided by the eyewitness.In the early stages sketches are drawn by the artist by the information provided by the eyewitness.Sometimes the witness does not recall the exact apperance of the suspect.In that case,sketch drawn is not helpfull for investigation.In this networked world, the sketch construction with the help of computer generated software arises.The computer generated composite sketch is constructed by selecting the facial components provided in the system database.The composite sketch construction gives more faster results for the law enforcement agency.The component based approach for face recognition using probabilistic neural network classifier gives more accurate result in photo to sketch matching.The proposed approach performs facial decomposition, some transformations and feature extraction for photo to composite matching.The use of classifier is helpfull for comparing the features of face images and gives the degree of match to make a decision.The advantage of using Probabilisitic Neural Network( PNN) classifier is that it requires only short training time..