Design Of Face Recognition System And Implementation On Android Platform

In recent years,as people pay more and more attention to information security issues,bioidentification technology has been used a lot in identity authentication field because of their own advantages of traditional identity technology.As a kind of important biological characteristics,the face is unique and has the advantage of carrying easily.At the same time,the face image acquisition conditions are relatively loose so that face recognition technology is used a lot in access control,security and other related fields.In this paper,the face recognition system is studied.For the face detection phase,two methods of face detection are studied.The use of skin color detection method could detect face faster,but the detection range is not accurate and often easy to mistake some other parts which contain skin color for the face.On the other hand if using the AdaBoost algorithm for face detection,the detection range is accurate but the detection speed is slow and the AdaBoost algorithm may detect some areas as faces by mistake.According to the shortcomings of these two algorithms,this paper combines the skin color detection method and the AdaBoost algorithm to obtain the face from the background.It chooses the YCbCr color space to establish the Gaussian model to obtain the face candidate region and gets the complete regions through the morphological means.In the obtained candidate region,AdaBoost algorithm is used to further detect the face to achieve more accurate positioning.Experiments show that the detection method can reduce the false detection rate of the two methods alone.In the aspect of preprocessing,the face image is aligned with the method of detecting the position of the eyes and the image is interpolated and normalized to the same size which makes the face more standardized.At the same time,in order to reduce the influence of light,the face image is processed based on image enhancement method.The preprocessing process for face images is provided to provide more efficient input for subsequent feature extraction.In the aspect of feature extraction,the feature extraction algorithm of local binary patterns is selected to extract the feature.By dividing the face image,the image histogram of each block is connected as the final face image features to match.Finally,the face recognition system which contains different modules is implemented on the Android platform.The system includes mobile phone camera to obtain face images and can switch the front camera and the rear camera to control image input.The face is detected by the skin color detection method combined with the AdaBoost algorithm and then the human face is normalized and the light is pretreated.The processed face is stored and the feature vector is extracted.And by computing the similarity between the collected face and the existed samples from the face database to achieve the function of matching recognition.Finally this paper verify that the system can meet the real-time and accuracy requirements through computing the running time of recognition time of the system and testing the actual recognition rate on the system.