80 likes | 392 Views
Facial Feature Extraction. Mert ZEYBEKLER. Introduction. Detection of the features Extraction of feature points Sending the coordinates of these points(output) as an input to the expression recognition system. Methods Used. OpenCV computer vision library
E N D
Facial Feature Extraction Mert ZEYBEKLER
Introduction • Detection of the features • Extraction of feature points • Sending the coordinates of these points(output) as an input to the expression recognition system
Methods Used • OpenCV computer vision library • Robust face detection framework proposed by Viola & Jones. • Haar caracteristiques, Integral image, Cascaded classifiers. • Simple image processing methods.(Thresholding,
System Flow • Grab the frames from the source (webcam/video file) • Detect the face for each frame • Using the face anatomy minimize the search areas for eyes, nose, mouth. • Detect eyes, nose, mouth, eyebrows and chin. • Detect the feature points by fitting lines, ellipses, circles to the edges.
Face Anatomy Used • Serves for minimization of the search window for each feature. • Helps gaining time in computation. • Eliminates most of the uninteresting parts.
Current Output (Screenshot) • Still working on the prototype.. • Trying to fit ellipses to the eyes.(a little buggy)
Future Work • Ellipse fitting to the mouth. • Line fitting for the eyebrows and chin.