Font Size: a A A

Research On Human-Computer Interaction System Based On Emotion Recognition

Posted on:2021-03-16Degree:MasterType:Thesis
Country:ChinaCandidate:L F ChenFull Text:PDF
GTID:2428330611488399Subject:Power engineering
Abstract/Summary:PDF Full Text Request
As the computing power of further ascension,and the application of all kinds of new technology,the computer is more intelligent face recognition technology is increasingly applied in today's world of applications and products in face recognition is to realize the nature of one of the basic means of human-computer interaction(HCI)emotional recognition is the key of the further research on human-computer interaction emotional complexity makes this task in the work of information collection is becoming more difficult in understand the face recognition is presented in this paper and the research status of emotion recognition,combined with laboratory existing NAO robot,designed a preliminary human-computer interaction system,the main research contents are as follows.Firstly,a method of image preprocessing using LBP operator is introduced.Face in order to obtain more accurate information for emotion recognition,we have applied to the face image to 3 d reconstruction,and the convolutional neural network model training when the input image preprocessing of LBP algorithm with the direct input RGB image training can effectively reduce concentration is different due to the interference of light factors,improve the accuracy of the model.Next,a multi-mode real-time face emotion recognition system is established.Face detection and emotion recognition are carried out through image pairs,and the success rate and accuracy of recognition are further improved by using sound as an auxiliary means.Using network(CNN)is a kind of improved VGG-16 network model,the whole system mainly consists of speech recognizer,face detector and emotional predictor,finally using FER2013 data sets and FER2013 + experiments have been carried out to verify data sets,compared to before using FER2013 dataset,FER2013 + data set a more accurate picture clearer,mark,which makes this article compared to before the system has better recognition efficiency and accuracy.Finally,a preliminary human-computer interaction system was designed.
Keywords/Search Tags:NAO robot, Face reconstruction, Emotion recognition, Humancomputer interaction
PDF Full Text Request
Related items