@article{ff80c6cd47fc407a931e4d5b4dc2c37e,
title = "A statistic approach to the detection of human faces in color nature scene",
abstract = "In this paper, a novel algorithm for oriental face detection is presented to locate multiple faces in color scenery images. A binary skin color map is first obtained by applying the skin/non-skin color classification algorithm. Then, color regions corresponding to the facial and non-facial areas in the color map are separated with a clustering-based splitting algorithm. Thereafter, an elliptic face model is devised to crop the real human faces through the shape location procedure. Last, local thresholding technique and a statistic-based verification procedure are utilized to confirm the human faces. The proposed detection algorithm combines both the color and shape properties of faces. In this work, the color span of human face can be expanded as wilder as possible to cover different faces by using the clustering-based splitting algorithm. Experimental results reveal the feasibility of our proposed approach in solving face detection problem.",
keywords = "Clustering-based splitting algorithm, Color classification, Face detection, Shape location",
author = "Hsieh, {Ing Sheen} and Fan, {Kuo Chin} and Chiunhsiun Lin",
note = "Funding Information: In this paper, a novel algorithm for oriental face detection is presented to locate multiple faces in color scenery images. The proposed detection algorithm combines the color and shape properties of faces. Experimental results reveal the feasibility of our proposed approach in solving face detection problem. During the past years, almost all detection methods were proposed to directly segment the faces embedded in the images by utilizing only a color classification algorithm. However, this will lead to the miss of some real faces because the color of face spans too wide to be covered by only some narrow color zones. To remedy this problem, we propose a skin color classification method with wider color span combining a clustering-based splitting algorithm in this paper. The color classification method is first applied to classify the skin-like and non-skin-like color regions in the original image. Here, the HSI color system is adopted to design the color classification algorithm because it is stable for skin color in different lighting conditions. Then, a clustering-based splitting algorithm is devised to separate the real human faces and other skin color regions. Thereafter, the real faces are located and labeled as “candidate face” by applying the model-based face location algorithm. Finally, the faces are verified by using the local thresholding technique and statistic-based verification procedure in the face verification module. In this paper, the color spans used in the color classification algorithm are selected as wider as possible to cover all the skin colors. Consequently, the color that does not belong to a face will also be covered in these spans. Hence, a splitting algorithm must be applied to separate the facial and non-facial color regions in the original image. It is the most important contribution of our work that other researchers have not done. Besides, an elliptic face model is devised to locate the real human faces by comparing the shape of face model and that of each color region obtained in the previous stage in the model-based face location algorithm. Though a feature-based approach can clearly depict the human face, it is unable to work well under different imaging conditions because the structure of facial features varies too much to be robustly detected. Hence, a statistic-based verification procedure is proposed to verify the candidate faces in our work. In summary, we have developed a face detection algorithm which can tolerate the wider lighting conditions in the scenery images. Besides, experiments were conducted on three different databases which contain images taken from digital camera, internet, and scanner. Experimental results have demonstrated the validity of the proposed algorithms. Especially, the clustering-based splitting algorithm can successfully separate the face from the neighboring background with the skin color. About the Author —ING-SHEEN HSIEH was born on 5 November 1957 in Changhua, Taiwan, Republic of China. He received his B.S. and M.S. degrees in electrical engineering from National Cheng-Kung University, Taiwan, in 1980 and 1987, respectively. From 1982 to 1987, he was a research assistant in the Chung-Sun Institute of Science and Technology (CSIST), Taiwan, where he has been an assistant researcher since 1987. In 2000, he received his Ph.D degree in Institute of Computer Science and Information Engineering from National Central University. His current research interests include pattern recognition and image analysis. About the Author —KUO-CHIN FAN was born on 21 June 1959 in Hsinchu, Taiwan, Republic of China. He received his B.S. degree from National Tsing-Hua University, Taiwan, in 1981, and the M.S. and Ph.D. degrees from the University of Florida, Gainesville, in 1985 and 1989, respectively, all in electrical engineering. In 1983, he was a Computer Engineer with the Electronic Research and Service Organization (ERSO), Taiwan. From 1984 to 1989 he was a Research Assistant with the Center for Information Research, University of Florida. In 1989, he joined the Department of Computer Science and Information Engineering, National Central University, Taiwan, where he became Professor in 1994. He is currently the Chairman of the Computer Center, National Central University. His current research interests include pattern recognition, image processing, computer vision and neural networks. Dr. Fan is a member of SPIE. About the Author —CHIUNHSIUN LIN was born on 5 December 1961 in Chia-yi, Taiwan, Republic of China. He received his M.S. degrees in Computer Science from DePaul University, Chicago in 1993. He is an assistant researcher in Committee for Planning and Organizing the National Taipei University since 1995. In 1996, he entered the Institute of Computer Science and Information Engineering at National Central University working toward his Ph.D. degree. His current research interests include pattern recognition, face detection, face recognition, image processing, and image analysis. ",
year = "2002",
month = jul,
doi = "10.1016/S0031-3203(01)00146-7",
language = "???core.languages.en_GB???",
volume = "35",
pages = "1583--1596",
journal = "Pattern Recognition",
issn = "0031-3203",
number = "7",
}