Research On Augmented Reality And Expression Identify Learning System Based On Panoramic Annulus Imaging System

Augmented reality combines virtual scenes with real scenes.Real scenes provide visual,tactile,sound and taste experiences.Virtual scenes further provide more vivid animations and sound effects.Applying augmented reality to the learning field can increase users' interest in learning.However,the existing augmented reality systems applied in the learning field lack supervision of the learning state.The camera's field of view is limited and it cannot capture users at the same time,making it difficult to obtain the user's status.The panoramic annulus imaging system can solve this problem.In this paper,the augmented reality and expression recognition learning system based on the panoramic annular imaging system is developed,and corresponding solutions are proposed.First,the expansion of the original image is realized based on the model of the panoramic annulus imaging system and related calibration techniques.The 3D registration function based on natural features is realized through SURF feature extraction and Euclidean distance-based feature matching.The KD-TREE optimization algorithm is used to improve the matching speed,and high-quality matching points are selected by RANSAC to reduce mismatching in the matching process.The projection transformation of the 3D virtual model to the 2D image is calculated according to the matching results.And the interaction is realized through the keyboard and the mouse.Face detection is performed through the AdaBoost and MTCNN neural network.Expression recognition is performed through SVM and convolutional neural networks.Voice interaction is achieved based on the recognition results.This method uses the panoramic annular imaging system to solve the problem of small camera field of view in traditional augmented reality.It can provide more interaction and monitoring method by combining face detection and facial expression recognition.