Key words
face detection, validation, false positive, cascading approach
articles
The author:
Zeno B. Published in
№ 1(73)
27 february 2018 year
Rubric: Models and althorithms
Rubric: Models and althorithms
Face validation using skin, eyes and mouth detection
In unconstrained facial images, large visual variations such as those due to pose, scale, presence
of occlusions, expressions and lighting cause difficulties in discriminating faces from the
background accurately, so as a result, there are non-face regions that are recognized as faces
(false positive), whereas the effectiveness of face detection algorithms is characterized by low
false positive (FP) rate, high detection rate and high speed of processing. So, to reduce these
non-face regions, instead of developing accurate face detection algorithm that needs much time
for processing, face validation step will be added after the detection. In this paper, new fast face
validation method is proposed. It consists of two steps, the first one is skin detection using YCbCr
color method. The second step is eyes and mouth detection using Cascading approach; In this
step, region of candidate face is divided into two overlapping regions, one for the eye detection
model and the other for mouth detection model. For evaluation our method, SVM face detection
algorithm is used as a baseline validation algorithm. The experimental results on FDDB dataset
showed a better performance of our proposed method (2 ms validation time compared to 500
ms in the SVM algorithm) and a similar number of rejected FP.
