This thesis mainly proposes a system that can automatically acquire the region and features of faces from color image in complex background, and obtain the result of face detection and thus providing more information for future studies in image processing based on this research. The face detection system firstly detects the skin color form the image by using lighting compensation and normalization. The image is transferred into YCbCr color space to improve its effectiveness, and then proceeds to image segmentation with Opening operation of Morphology. The human face ratio, mouth and eyes will be captured from these skin regions and excludes any non-human face region one by one in features of faces. Thus the specific human face region can be acquired successfully. As a result, the information gathered form this proposed system can be used in many applications such as human face recognition, facial expression recognition and pedestrian detection.