Facial Features Extraction

parathyroidsanchovyΤεχνίτη Νοημοσύνη και Ρομποτική

17 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

77 εμφανίσεις




Facial Features Extraction




Amit Pillay

Ravi Mattani


What Are We Doing !


Finding Features on a Face


Eyes


Mouth


Nose


Why Facial Feature Extraction?


Biometrics


Facial recognition system


Video Surveillance


Human Computer Interface

Difficulties !


Face Variation


Physical characteristic vary


Non
-
uniform lighting


Face position



Previous Work.


Many Face Extraction Methods



Main Trend : Combine image information and knowledge of face



Ian
-
Gang Wang, Eric sung in their article have proposed a morphological
procedure to analyze the shape of segmented face region. Several rules
have been formulated for the task of locating the contour of the face.




Terrillon et al., 1998 mentions the problem of how other body parts such
as neck may lead to face localization error



Haalick and Shapiro, 1993 demonstrate how morphological operations
can simplify the image data while preserving their essential shape
characteristics and can eliminate irrelevances.





Our Process


skin color
segmentation
Input
image
Morphological
image
-
processing
Skeletonization
Line
segmentation
and contour
detection
Facial feature
extraction using
facial geometry
.
Output
image
Skin Segmentation



Depends on color space


Used the finding by Yang & Waibel(1995,1996)


Normalized r
-
g color plane.


Took seed pixel


Classified the pixels based on whether the pixel
lies within the threshold


Same process carried out for the R and G plane



Skin color segmentation

Morphological Image Processing



Dilation


Fills the holes



Erosion


Restores the shape of the face


Morphological Image Processing

Skeletonization


Reduces binary image objects to a set of
thin strokes.


Retains important information about the
shape of the original object


Skeletonization

Contour Tracing


Certain vertices of these skeleton lines called
fitting points can fit the contour of the human
face.



Certain rules are then applied to deduce these
fitting points by analyzing the skeleton lines.

Contour Tracing


Rule 1
-

The contour fitting points should be the vertices of the
roughly horizontal skeleton line segments that are long enough.



Rule 2
-

The left vertex will be selected as candidate for contour
fitting if most of the horizontal line segments are positioned at
the left of the symmetry axis and vice versa



Rule 3
-

The contour points should be above a vertical position
that is set at 3/4 of the height from the top of the symmetry axis



Rule 4
-

The point set satisfying the above will be doubled using
symmetry axis


Contour Tracing

ROI


Feature extraction within the ROI


Edge Detection using Sobel Operator


Vertical position by horizontal integral
projection


Lip line maximizes the projection


Bounded by a rectangular box


Same process is repeated for nose and
eyes regions within the fixed vertical
positions



Results










Conclusion


No. of images experimented with = 30


No. of images in which features are
correctly identified = 27


Percentage correctly identified = 90


Average time taken to get the output in
MATLAB = 15
-
20 secs


Future Work


More robust and dynamic


Extended for profile views of image


More efficient code for faster execution
(applicable especially for MATLAB !!!)

References


Frontal
-
view face detection and facial feature
extraction using color and morphological
operations

by Jian
-
Gang Wang, Eric Sung


A Model
-
Based Gaze Tracking System

by Rainer
Stiefelhagen, Jie Yang, Alex Waibel


Digital Image Processing Using MATLAB

by
Gonzalez, Woods &Eddins,Prentice


Images taken from www.faceresearch.org


Prof. Gaborski’s lecture slides


www.wikipedia.com

Questions???