Detection Of Human Face And Its Features Relevant Application

With the fast development of technology and gradual improvement of people awareness of safety, the demand for application on human face is increasingly urgent. Comparing with other human biological characteristics, face characteristics are prone to observe and direct, therefore detection of human face and its features and relevant application become the research focus of the field of image processing and has extensive application future. Face recognition system and fatigue detection system are the most important face applications which care about the safety problems related with human and have done great help to human safety, accordingly the research on them is of great importance. The face recognition system pays much attention to whether an input image is matched with the training images. The fatigue detection system works to detect and analyze whether a driver is tired or not based on real-time situation and sends out the warning if necessary. Both of the two systems need to detect the face firstly and then detect the critical features on human face, which is the basis of consequent work. So, how to detect the features accurately is the key to the face application and the emphasis of our research. In this dissertation, refinement of detection of human face, algorithms of facial features extraction, fatigue detection and relevant application were profoundly researched, based on which the classical algorithms, novel and creative improvement was proposed. The main discussion of the dissertation is listed as follows:1) The history and status of research on facial features extraction and fatigue detection system are systematically summarized. Detailed kinds of feature points extraction approaches, including grey-level-based algorithm, knowledge-based algorithm, Geometry-base algorithm, statistic-based algorithm, wavelet-base algorithm, and then analyze the different algorithm. Also, some methods to detect the fatigued status of human are introduced and analyzed.2) The algorithms of locating rough position of special facial organs are researched, based on which the detection of face is improved. Firstly, eyes locating methods using templates is introduced, and Clustering-based mouth locating method is advanced. According to the golden selection and experiences, correction of the frame of the face is made and the results before and after the correction are compared.3) Two classical algorithms of facial features detection are introduced: Active Shape Model (ASM) and Active Appearance Model (AAM). Experiments are made with comparing the two algorithms not only in theory but also in real use, and then we conclude the merits and weakness of them and propose the improving solution.4) A new method of facial features detection called Templates Matching Searching(TMS) based on some ideas of traditional Active Shape Model is presented. And it turned out that facial feature points can be located robustly and precisely using the method proposed. Experiments demonstrate that our algorithm for the facial feature extraction is precise and overcome the past defect that the features can be accurately extracted when searching the completely frontal face images.5) Eyes and mouth are detected in each frame of a video in a driving situation, and then their areas are computed, from which we can get the frequency of eyes'blinking. Analysis and decisions are made for different conditions, and our system will send the warning when finding abnormity of a driver.