|
Title:
|
AN AUTOMATIC SYSTEM OF HUMAN FACE DETECTION AND RECOGNITION |
|
Author(s):
|
Ammar.i.shihab |
|
ISBN:
|
978-972-8924-63-8 |
|
Editors:
|
Yingcai Xiao and Eleonore ten Thij |
|
Year:
|
2008 |
|
Edition:
|
Single |
|
Keywords:
|
Wavelet transform; Edge detection; Symmetry detection; Face detection; Face recognition; Template matching
Correlation; Principal component analysis; K-L e |
|
Type:
|
Short Paper |
|
First Page:
|
211 |
|
Last Page:
|
215 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
An Automatic System of Human Face Detection and Recognition is proposed in this work, it consists of five steps, First
input image is decomposed using Haar wavelet transformation and reducing redundancy. Secondly facial edge detection,
in this step all elements of the sub-bands HL and LH are converted into binary, and all pixels of image mixing by NOR
operation. The result image matrix 128x128 pixels is partitioned into blocks size 8x8 pixels by selecting ≥ 8/64
thrshulding as an edge block. Finally, Facial Edge Block is yielded by matching between edge block and special type of
mask called T-Ship mask. In this stage we use a T-Shape mask in to matching at the output of second stage for some
condition, the target areas (matrix x,y) were successfully reduced to less than 20% of the entire area of edge image on
average.
Third step called symmetry axis detection, in this step is narrows down face areas further using gradient orientation, and
identifying the face location from image, and make a horizontal vector divide the human face, Fourth face detection, in
this step is restrict face like areas by Tamplet Matching, Finally step Fifth , Face Recognition. In this step we determined
the best face location in the face like area and identify the face using the principal component analysis (PCA). The
performance of system is tested by using more than 300 images with different background and illumination angle of
human faces. |
|
|
|
|
|
|
