Design Of Embedded Face Recognition System Based On ARM9

With the advent of the digital age,identity identification technology and biometric is showing its own value based on a technology as a research field of computer vision and pattern recognition for face recognition,in the background of rapid development of computer technology and pattern recognition technology under the strong impetus to the development and application of biometric technology in terms of security,detection etc..Application is very typical in intelligent access control,surveillance monitoring,identity authentication,confidentiality protection etc..In recent years,the embedded system with complete function,safe and reliable,low price,small size and low power consumption advantages,to life,with the characteristics of practical application as the core,more and more widely applied to various fields.The design for the image acquisition is easily affected by illumination,face rotation and other factors,the existing problems such as time-consuming calculation of the exact algorithm,is proposed to represent the face contour using the general defonmation model,stepwise refinement approach to local area detection,acquisition,processing,increase the anti-jamming performance,using a combination of global and local features which can further improve the recognition efficiency,better improve the design to transform the recognition image processing ability to deal with.Under normal circumstances,it is easy to be affected by illumination,distance,angle and so on,which greatly increases the difficulty of face recognition.The design in the image pre-processing using geometric image processing,the face parts can be precise positioning and introduction;using geometric normalization,image rotation,clipping to achieve uniform standards;gray scale processing,and set a reasonable threshold,making it easier to distinguish between the background image and face region the part,to speed up the processing speed;the distribution of gray histogram equalization technology is applied to adjust the image,improve the image quality;and median filtering is used to remove the image noise,and further ensure the quality of image recognition.After the pretreatment of the first link,enter the second stage:in the feature extraction step,through the comparison and selection of appropriate extraction method(for example:K-L transform method,wavelet transform method,Harr feature extraction method etc.);in the classification part,under the conditions set,the feature space is divided into types of space,and because the pattern the recognition system will need a learning process,the classification function automatically adjusted through change of sample characteristics,that is the training of the classifier.Therefore,in order to ensure the correct recognition rate,the corresponding decision rules are set up.And through the identification of the results,you need to experience a continuous input of a modified parameter of the positive feedback process,until the actual recognition requirements.In recognition algorithm,comparison of various mainstream recognition algorithms(e.g.,principal component analysis,Gabor transform etc.),the selected principal components of global and local features based on the analysis of the design as a recognition method.According to the actual needs of the application,the design of face recognition algorithm based on the ARM development board of embedded devices,using S3C2440A chip and Ubuntu operating system platform,identification of human face detection using ARM in embedded devices on the camera,the improved algorithm of the proposed design and implementation of classification training on face recognition classifier in PC on the use of network transmission mode to meet the practical requirements of the development board after training the classifier is transmitted to the ARM in the field test,the test results show that the design improves the recognition efficiency to a certain extent,has important practical significance.