Design And Implementation Of Embedded Face Recognition System Based On Cloud Service

With the rapid development of internet technology,and under the support of big data and deep learning technology,face recognition technology has obtained a deeper development in many fields,such as computer vision,artificial intelligence and so on,and the technology has been widely used in many scenes like smart robots and authentication.Under this background,this thesis designed an embedded face recognition system based on cloud computing technology to realize the face detection and recognition.After the introduction of fundamental image processing technologies related with face detection and face recognition,the framework of the embedded face recognition platform based on cloud service is presented.Modularity design technology is adopted by this thesis.The whole platform can be divided into five parts: the core processing module,power module,image acquisition and playback module,storage module and communication module.To meet the system's requirements,TMS320DM368 DSP produced by TI is selected as the core processor.The power module provides the stable and reliable voltages with various specifications for the entire system.OV7725 is used to acquire the input image,and LCD interface is provided to support image display and subsequent development of man-machine interface.Storage modules includes NAND FLASH and DDR memory,which are used to store the data of bootloader,kernel,file system,application and provide running space.RT3070 wireless adapter is selected to realize the network communications.The schematic and PCB of the platform is designed.The designed PCB is produced,set up and debugged.Based on the designed hardware,the embedded software platform is established and t application program is designed.The establishment of software platform includes construction of cross-compile environment,the cutting,compiling,and transplantation of UBL,U-Boot,Linux kernel,and the design of drivers for image data acquisition and playback module and Wi-Fi module.Application program includes image acquisition and playback,image preprocessing,accessing to the cloud server,and the face detection and recognition application.Finally,the designed system is tested under laboratory environment.The testing results show that the system meet the design requirements well,.Face recognition is implemented through the access of Microsoft Cognitive Services,which can be used in the scenes like smart robots with the ability of identification.