Design And Implementation Of Face Recognition Access Control System Based On Android

As one of the hot spots in the study of biological recognition technologies, face recognition is characterized by friendly, convenient use, non-contact and high concealment. It is widely used in the fields of entrance guard system, smart home and video monitoring. The entrance guard system has evolved from the original single mechanical locks into the electronic entrance guard systems based on chip cards or DSP, FPGA, ARM and other hardware platforms. With the advent of the era of mobile Internet, Android system on SOC has become one of the most popular intelligent hardware operating systems, and the face recognition under the Android system is gradually becoming a new research direction.Face recognition process mainly consists face detection, face image feature extraction and determination of identity. This thesis studied and optimized several key steps of the algorithm, and designed and implemented face recognition access control system based on Android. The main work is as follows:(1) The classic face detection algorithm is discussed; the advantages and deficiencies of various algorithms, as well as the limitations of implementation in concrete hardware equipment are summarized. The face detection part adopts face area detection based on the multi-feature method with the combination of face skin algorithm and Adaboost algorithm, and the skin color is modeled in the YCbCr color space. The experiment results reveal that, the method not only has short detection time, but also is not sensitive to the light, facial expressions and shooting angles of the input images and improves the detection rate.(2) The face recognition algorithm based on PCA is analyzed, the face recognition calculation process of PCA algorithm in entrance guard system is optimized, the complexity of calculations is reduced, and the face recognition speed is improved with the premise of unchanged face recognition rate.(3) The face recognition entrance guard system is realized on Android platform, the whole face recognition process from face image extraction to identity recognition is completed, and the system gives out the face recognition results and permissions to open entrance guard system.