Human Face Recognition Using Neural Networks Project is about the development of human face recognition system (HFRS) using multilayer perceptron artificial neural networks (MLP). The system takes the face image as input from video camera and also detects the presence of an object in front of the camera and detected facial area will now be used as input to neural network to perform recognition.
The operation of HFRS has three modules: Human head tracer (HHT), Eye locator (EL), and Recognition of face (RF).HHT module includes sensing the presence of an object in front of video camera and to locate the human head in image frame. EL module includes searching for location of human eyes in the image frame and responsible to scan for left and right eyes of face image. RF does the main task of human face recognition. The input of RF is the image of human face and is compared with several output nodes with set of face images fed in database.If nooutput nodes responds to input image, the input image is considered as not recognized else it is recognized.
HFRS process includes capturing head and shoulder of human object with equipment, detection of image and approximate location of human head and scans for eye features and once eye features are detected and face input area is known is compared with several output nodes with set of face images fed in database.
It concludes that when the optical bruin damage (OBD) technique is applied to MLP there is no significant increase in its ability to recognize face images.
And MLP without OBD has ability to reject non face images. But When is applied to MLP deletes the ‘unnecessary’ weights from network. This reduces MLP in making wrong classifications on non-face images.