The Research And Implementation Of Face Detection Algorithm Based On The Skin Color Segmentation

With the development of information technology in contemporary society, the demands for security are becoming more and more important. Human face detection has essential functions in face recognition, biometrics, identification and other theories or applications. The automatic human face detection based on computer is one of the most important subjects in natural human-computer interaction technology. As part of multifunctional perception science, the goal of face detection is exactly detecting human face in different backgrounds, and giving the locations and sizes if the faces exist. Scientists and engineers have studied this subject for many years and made it became a very active research branch pattern recognition and computer vision application areas with its significant value in recent years.This paper researched the face detection in static and color pictures from a practical and easy way which based on the color of skin and eyes detection.Firstly, this paper introduces the concept of human face detection, background, research situation and some commonly used methods of face detection and evaluation ways. And then studied the algorithm based on using skin color information to separate candidate area from the background of images. After compared several common color space, this paper chooses YCbCr color space and uses Gaussian skin model to segment skin regions.Some of the skin areas detecting from skin color segmentation are the face regions, some are other parts of the body, such as arms and legs and some are even part of the background. So at beginning this paper uses the open operation and closed operation of the morphology to remove the small area that’s part of the not face region. Then uses area of parameters and aspect ratio to determine if the candidates are face area and remove the region that is not meet conditions given in the article. Eye detection to further checking if the left area is a really face, and locate it if it is a face.Finally, use MATLAB R2011a to program and achieve the proposed algorithm, and design a simple GUI interface to facilitate more intuitively the detection process of observation. And then a large number of images were tested. The experimental results show that, using the algorithm based on skin color detection and the verification of the human face detection, the detection effect of image following five people face is good, for detection rate is higher, error detection rate and control in a smaller range. While the residual rate and the rror detection rate relative on the higher side for the image with five to ten people face or small angle deflection face, but also can achieve good results.