Research On Implementation Method Of Face Detection On ZC702 Embedded System

The technology of face detection is an important research direction in the field of biological features recognition and computer vision,its purpose is to detect the face region accurately and quickly from the image or video.Although the traditional face detection technology is mature,to meet the high detection rate and high efficiency at the same time along with a low cost and a low power consumption has become a hot and difficult research.Now,the face detection system is being developed in the direction of miniaturization and convenience.Using the embedded system of the Zynq-7000 system which is based on the ARM and FPGA to achieve the face-detection technology can satisfy real-time,high-speed and efficient requirements.The face detection method and the implementation process of the face detection system on ZC702-based are being studied.The chief research works are as follows:(1)Research on the Adaboost-based face detection algorithm.Begin with the some common methods of face detection and the evaluation of the detection effect,it's proved that the Adaboost algorithm has obvious advantages in accuracy and speed to the face detection,so the Adaboost algorithm is chosen to construct the face detection system.The principle of Adaboost algorithm is further studied,and then the two concepts have been learned: Haar-like feature and integral graph.A cascade classifier has been created.(2)Improved PCNN(Pulse Coupled Neural Network)model and using it to enhance the image.Research on the ZVIK(Zynq-7000 Video and Imaging Kit)equipment to obtain the samples of the face detection system,using the improved PCNN algorithm to enhance the image when taking into account that the illumination difference of the images will have a impact on the face detection.A Adaboost-based face detection system is achieved by using VS2008 and Opencv.(3)The migration and implementation of the Linux system and the Opencv.Linux system migration and the Opencv migration are the most important works in the research design of the ZC702-based face detection system.So,the host(PC)and the target(ZC702demoboard)development environment of the Linux system is established.Through the system programming,modifying and compiling,the design of the face detection system is realized.Finally,the system migration is completed by using the Vivado design suit on the ZC702 demoboard.(4)Experimental testing and analysis of the face detection system based on the ZC702 demoboard.On the basis of system migration implementation,the face detection system is tested and verified its feasibility and accuracy,results show that each function of the system works normally and the whole system is of good effect,which achieves the purpose of the expected design.