Augmented Reality Recognition System Based On HoloLens

Augmented reality technology is a comprehensive technology based on computer graphics,artificial intelligence,sensing technology and human-computer interaction technology and other disciplines.This technology can deepen the user's perception of the real world by superimposing the virtual world into the real world,expand the real world,so that users can understand the real world more vividly and concretely.In recent years,augmented reality technology has been widely used in education,industry,entertainment,military and other fields.Its main research content includes video acquisition,image recognition,3D tracking registration,virtual-real fusion modules and so on,among which tracking and registration are core research contents.This paper focuses on the issues of augmented reality technology in different application scenarios,where identification information and virtual information need to be placed in advance,the display of virtual information is limited by the identification information,the efficiency of identification recognition is low,the augmented reality project is not universal,and the problem of high development costs.The core work is as follows:(1)In view of the problem that the identification information and virtual information in augmented reality need to be placed in advance,a model-free tracking dynamic loading method is proposed,which uses keyword query to dynamically load the model,interact with the real world with HoloLens' unique spatial mapping function,reduce the dependence of virtual information on identification information,and get rid of the tracking of identification information.Design and provide a variety of gaze,voice and gesture operation methods for human-computer interaction to control the loading,rotation,displacement and detailed display of virtual information.(2)Aiming at the problem of weak generalization ability and low recognition efficiency of identification information in augmented reality technology,a registration-free identification method is proposed,which uses transfer learning method to train convolutional neural network(CNN)to classify the identification of target information.First,pre-train the AlexNet network and initialize the network model parameters;second,collect sample data sets of application scenarios through the Internet,and expand the data set through methods such as rotation and mirror flip;retrain the fully connected layer of the network model to adapt to the application scenarios,Retain the parameters of the new fully connected layer;and finally fine-tune the initial network model and the new fully connected layer to complete the model training.(3)Augmented reality recognition system based on HoloLens is realized.The system uses the C/S(client-server)architecture to solve the problem of updating application scenarios and virtual information.The HoloLens device is used to complete the target information,dynamic model loading,spatial mapping and human-computer interaction functions for the client to realize the effect of augmented reality;according to the changes of the application scenario,the server uses the transfer learning method to train the network model,updates the target information and virtual information required by the system.This system uses the UDP protocol to transmit information,the UDP data is preprocessed to solve the packet loss disorder.Experimental results show that without changing the client application,the client can load new virtual information to adapt to new application scenarios,shorten the development cycle of the augmented reality system,and save labor and time costs.