International Bulletin of Mathematical Research Volume 02, Issue 1, 2015 Pages 203-208, ISSN: 2394-7802 Face Detection Using Skin Color Model Shalini Yadav 1 , Neeta Nain 2 and Tapas Badal 3 Department of Computer Science and Engineering, Malaviya National Institute of Technology, Jaipur-302017 (Rajasthan), INDIA. Email: 1 yadavshalini45@gmail.com, 2 nnain.cse@mnit.ac.in, 3 tapasbadal@gmail.com Abstract Face detection is a challenging task in image processing field. Now days face detection is used in many application which requires identification and verification. The major application areas of face detection are biometric field, crowd surveillance, photography and many security related areas. Issue with face detection is that it should perform well in any condition such as illumination, color variance, brightness, pose variations. This paper proposed a technique for face detection based on skin color model. For skin tone segmentation we are using YCbCr color model in algorithm. The reason behind using YCbCr is to remove illumination component which is represented by Y. We proposed a technique to detect face from image based on area and mark that region. The results of the algorithm is shown in figures. The proposed algorithm is tested on various dataset and gives better experimental results than traditional approaches. Keywords: Image Segmentation; Face Detection; Color Model; Median Filter 1 Introduction Face detection technique is widely used in the area of biometrics, time tracking service, outdoor surveillance camera service, smart captcha, secured access, video chat service etc. The main purpose of face detection is for authentication. This approach is very useful in biometrics. The physiological methods are more stable than methods in behavioural category. The reason behind this is that physiological methods are not altered. Various methods are proposed for face detection technique some are traditional and some are enhanced techniques. The problem with traditional technique is that it is affected in different lightning environment, identifying an individual from images of the face, so the result is erroneous. To differentiate the skin region and non-skin region is a major task in face detection system. Face detection technique is improved if skin area and non skin area is measured. Skin classification decrease the misclassification rate that’s why this paper proposed a technique face detection followed by skin classification. There are several new techniques and some enhanced techniques are proposed to overcome these problems. Several face detection approaches have been introduced [1]. Various methods are proposed for face detection i.e. the S-AdaBoost algorithm [2], Skin color based [3] [4], Neural Networks [5] [6], Bayes classifier [7], Chrominance based[8]. In this paper we proposed color model based face detection approach. Proposed approach is simple as it uses color component in RGB and YCbCr model to localize face region in an image and its computational complexity is low in comparison to other existing methods. This paper is organised as follows: Color model based face detection algorithm is described in Section 2, Our methodology is described in Section 3, Experimental results are discuss in Section 4, conclusion are given in Section 5. 2 Algorithm Description This section explains the basic steps used for face detection algorithm. Algorithm consist of following steps: A. Image retrieval B. Noise Removal C. Skin Classification D. Face Detection In first step images are acquired from database or from live cameras. As face detection are perform in various situations where images contain noise it is required to perform noise removal technique on image. In our method we use median filter for noise removal. There are also some other filtration technique used in literature such as low pass filter and FFT but median filter performs better for noise removal and smoothing the images.The framework of face detection algorithm used in this paper is shown in Figure 2.1.