Design And Implementation Of Digital Photo Frame Based On ARM Processor Collection And Processing System

With the development of embedded technology,embedded platform plays a more and more important role in various industries.With the wisdom,safe city and high-speed development of AI,face recognition technology combined with the embedded platform has been gradually applied in intelligent security,intelligent transportation,attendance access control,identity authentication;ensure the banking system identification,etc.The installation and portability of embedded face recognition system based on Android can be widely used in the above industries,which have great significance for social security and economic development.This thesis is design and implementation a digital photo frame which based on Texas Instruments(TI)AM3358 processor.The digital photo frame is increased face recognition function,which can be used to categorize photos.This thesis consists of five parts: the demand analysis and scheme design,hardware design and implementation,the design and implementation of face recognition algorithm,the design and realization of software system,test plan and results analysis.The requirements analysis and design chapter proceeds as follows: in our life,digital photos ever-increasing quantity.The electronic devices,which have face sort search function and achieve the management of these pictures,will have commercial value.Starting from the demand combined with the existing technical conditions,determine AM3358 as the mainly processor,run Android face recognition system solutions,and puts forward the overall design scheme.The hardware platforms based on TI AM3358 processor include six parts: the processor core circuit,storage circuit,display circuit,network communication circuit,power circuit and the common interface circuit.The hardware system is low power dissipation and low cost,only 700 mw power consumption,is a more cost efficient solution,have good business value.Face detection algorithm using Haar feature based algorithm,in order to distinguish between face and non-faces,to improve image recognition degree,put forward the innovative non-face rejection algorithm is proposed.The similarity between the tested face and the standard face was compared;exclude a large number of non-faces which is caused by interference of the external environment.In order to improve operation Face recognition algorithm speed,using principal component analysis(PCA),the face projection data can be reduced to 300 dimensional coordinate spaces,linear analysis(LDA)is used to obtain the subspace characteristics better,in this coordinate space and optimize the face data,through some preprocessing algorithms to achieve the purpose of enhancing the recognition degree.Face clustering and ordering belongs to the content of data mining,the system innovatively applies it to face recognition field the function of face sorting is realized by using hierarchical clustering algorithm.The whole software system framework consists of three parts.Design the Java Native Interface(JNI)call Interface is designed to migrate the algorithm to the Android system in the form of a library,so that the upper application can be called.Set up the subsystem of face recognition algorithm is used to realize and verify face correlation function.Build a database system,create and update the original photo database,face photo database,contact person database,face class database used for face recognition and sort the input and output of a face,thus forming an embedded face recognition system.The test scheme and the result analysis part have conducted environmental reliability test,electromagnetic compatibility test,whole machine power consumption test and face recognition algorithm performance test for the equipment.The reliability of this scheme is fully verified.