Research And Implementation Of Suspect Identification Algorithm In Public Traffic Management System

The biometrics technology is the use of some of the physiological characteristics inherent in the human body,Such as iris,fingerprint,face,retina,palm print,DNA,etc.,which have unique and unchangeable characteristics)and human behavioral characteristics,Techniques such as handwriting,sound,etc.,which also have unique and difficult to change characteristics,use image processing means and pattern recognition methods to identify personal identities.Compared to other biometric technologies,face recognition has unique features such as unique,stable,non-replicable,and non-counterfeit.It is more secure than traditional methods for identity authentication.Therefore,face recognition technology is regarded as the most ideal biometric recognition technology so far.Face recognition technology is used in public traffic management system to compare passengers with people in pickpocketing database,so as to identify suspects.Face images are usually large images.If the images are directly input into the network,it will bring an inestimable amount of computation.The most common processing method is to compress the image,which can reduce the amount of computation,but at the same time it will lose a lot of useful information for recognition.In order to avoid the above problems better,under the background of public transport management system,the suspect identification based on Adaboost algorithm and bidirectional 2DPCA algorithm is proposed as the preliminary study of the system.Adaboost is the most representative lifting algorithm in Boosting method.The core idea of Adaboost is to train weak classifiers by increasing the weight of the wrong samples,and through each iteration to improve one by one,finally all these classifiers are constructed into a more powerful classifier through linear combination.The bidirectional 2DPCA algorithm is improved and upgraded according to the shortcomings of the traditional principal component analysis(PCA)face recognition algorithm.The main idea of the algorithm is to extract features by reducing the data dimension.Adding bidirectional 2DPCA algorithm to Adaboost algorithm can not only locate the face more accurately,but also improve the speed of suspect face recognition.Face recognition algorithm is implemented in embedded system.The system is installed in public transport passenger access channel,and image data is updated regularly through wireless network.