Abstract
A novel scheme is proposed to process the real-time images according to
requirements, captured in a computer web camera. By making use of open
source computer vision library (OpenCV) and applying its in-built
functions, we can easily detect human faces and then match them against
the training sets (face images stored in the database) to recognize a
human’s identity. Image handling and processing are done using basic
library methods. These library functions are able to capture images from
a webcam, create windows to hold the images at run time, save them in
the database and process them to check exact likeness. CascadeClassifier
is used for face detection, and HaarCascade is used for this purpose.
FaceRecognizer class is used for face recognition. EigenFaces,
FisherFaces and Local Binary Patterns are the basic methods that can be
used for this. Images are stored in form of a CSV database.