Evaluation of suitable color model for human face detection

Mohamad, Fatma Susilawati and Abdulganiyu , Abdu Yusuf and Sufyanu, Zahraddeen (2015) Evaluation of suitable color model for human face detection. International Journal of Advanced and Applied Sciences, 12 (2). pp. 46-50. ISSN 2313-626X (Printed) 2313-3724 (Online)

[img] Text
Evaluation of suitable color model for human face detection.pdf
Restricted to Registered users only

Download (453Kb) | Request a copy
Official URL: http://www.science-gate.com/IJAAS/


Skin color as a technique for face detection is a technology applied in many applications such as pornography filtering, human-computer interaction, and face recognition systems. A human face detection approach is presented in this paper using the skin color segmentation algorithm. Choosing appropriate color model is very important to increase algorithm efficiency. Yet, there is no conclusion about which color space is the best fit for skin color detection. The motivation is to provide a platform to decide which color model is the best to build efficient skin color detector that can improve the overall face detection system. The proposed technique is based on finding the maximum energy of histogram signal for skin which is limited to the ranges for each component of the color space under study. Different parameters such as energy of the histogram of each component of the color space, the limit of skin range in each color space and the maximum energy of the color spaces are used to evaluate the result. The result indicates that YCbCr provide better performance compared to RGB, YUV, HSV and CMYK color model. A detection rate of 97.51% was obtained using PICS database.

Item Type: Article
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Faculty / Institute: Faculty of Informatics & Computing
Depositing User: Dr Fatma Susilawati Mohamad
Date Deposited: 08 Sep 2016 04:01
Last Modified: 08 Sep 2016 04:01
URI: http://erep.unisza.edu.my/id/eprint/4986

Actions (login required)

View Item View Item