Design Of Face Recognition System Based On Redis Cache

With the rapid development of face recognition technology,major universities and researchers have carried out research and analysis of face recognition technology.The domestic face recognition technology is mainly aimed at the automatic recognition of the positive face,and it obtains stable feature points through projection for face recognition.Some scientists reduce the computational complexity and improve the recognition efficiency by reducing the matrix dimension.However,this method can easily reduce the recognition rate.In the United States,feature vectors were originally generated by collecting the distances between multiple key points.However,the computational complexity of this method was too tedious.Therefore,a U.S.institute created a fully automatic algorithm to automatically calculate the distance between two eyes.Distance recognition,although it can improve certain recognition efficiency,its slightly lower recognition degree cannot be accepted for the time being.Today's face recognition systems are used in systems such as public security criminal detection,bank monitoring systems,and company security systems that do not require high recognition times and efficiency.However,when there are many people gathered,efficiency will become the most important factor in the face recognition system.This is one of the main reasons why the large-scale face recognition system cannot be used in schools.For example,if the efficiency of the face recognition system is too low during verification of the test,it will take a lot of time to perform face recognition operations,which has a certain influence on the time at which the test starts.Therefore,based on the above considerations,it is necessary to make certain improvements to the traditional face recognition system,improve its recognition efficiency,and increase its use rate in places where people gather at schools,for example.Because the face recognition system is identified by converting the faces into pictures,it is limited by the reading speed when the pictures are accessed subsequently,which has a certain influence on the recognition efficiency,so the picture access speed is improved.It will be an effective method to improve the recognition efficiency of face recognition.Increasing the redis cache to access the picture can effectively increase the access speed of the picture,thereby improving the recognition efficiency of the face recognition system.Therefore,the improvement of the recognition efficiency of the traditional face recognition system is to add a redis buffer to perform picture access operations.The design of the entire redis-based face recognition system is divided into the foreground display module design,background function module design and data storage module design.The design of the foreground display module is divided into page display module design and data interaction module design.The front page is written in html language.The data interaction mainly uses mysql database and redis cache to interact with data in the background.On this basis,the interaction with redis is supported.The data is analyzed in detail.The background functional module design is divided into face comparison function design,search function design,key point identification function design and detection function design.Before the function of the module is designed,the face recognition technology must first be theoretically studied and analyzed.It is a study of various methods based on local feature face recognition,various methods of face recognition based on global features,and various methods of face recognition based on hybrid features.Secondly,we need to set up the environment for each environment that the system needs to use,including tomcat server setup,redis cache platform setup,mysql database setup,eclipse development tool setup,and nginx reverse proxy platform environment setup.Finally,it is about the design and implementation of functional modules,including face comparison function,search function,key point identification function and detection function.The design and implementation of the face comparison function is divided into the face comparison function upload module design,the contrast function image transcoding module design and the comparative design.The design and implementation of face search function is divided into the face search function upload module design,search function image transcoding module design and search design.The upload module and image transcoding module of the face key point identification function are similar to the face search function,and the key point identification function is a display of the four faces of the human face.The face detection function upload module and the image transcoding module are similar to the face search function,and the face detection function detects the face and returns the age and gender.The core algorithm used for each functional design is the similarity percentage algorithm,which is the percentage of similarity based on autocorrelation and cross-correlation operations and a series of calculations.The data storage module is divided into a mysql data storage module and a redis data storage module.Redis data storage module as the core of the system's data storage needs its theoretical research and analysis,including the analysis of the cache type,the analysis of the mainstream cache technology and the analysis of redis role.At the same time,the advantages of redis caching in face recognition system are analyzed in detail,which is mainly reflected in the redis internal functional part,redis related features part and redis superior to other cache parts.The test of the entire redis-based face recognition system mainly includes face comparison function test,search function test,key point identification test,detection test,and redis efficiency improvement test for face recognition system.After the image upload test of each function,face comparison test,search test,key point identification test,and detection test were performed.Finally redis the efficiency improvement test of the face recognition system.The experiment completed the design,implementation and test of the redis-based face recognition system.At the same time,Redis completed the test of improving the efficiency of the face recognition system.The test results show that:(1)The face comparison function upload module can upload pictures normally,and the face comparison module can perform face comparison and return similar percentages.(2)The face search function upload module can normally upload pictures,and the face search module can search for faces and return 4 pictures with the highest similar percentage and corresponding similar percentages.(3)The upload function of the face key point identification function uploading module can be normally performed,and the face key point identification module can identify the four faces of the human face.(4)The face detection function upload module can normally upload pictures,and the face detection module can perform face detection and return age and gender information.(5)Redis has a certain improvement on the efficiency of face recognition system,and the lifting time is about 400 ms.Due to the existence of objective factors(such as when the system is running,the CPU is occupied),there is a certain error in the experimental data,and the error range is within 150 ms.The experimental results show that the redis-based face recognition system has higher recognition efficiency than the traditional face recognition system,and it has good application value when large-scale face recognition is needed,such as schools,train stations,etc.A large amount of people can carry out large-scale face recognition scenarios.