Facial Expression Recognition Based On Convolution Neural Network

As an important part of the face detection technology,facial expression recognition is a new research topic in the field of artificial intelligence,which is an interdisciplinary topic since it relates to computational intelligence,pattern recognition,image processing,even physiology and psychology and so on.The.field of expression recognition research can promote the computer to the intelligent better development,has a very important practical significance and will have considerable social and economic benefits,the market prospects are bright.To develop and evaluate such applications,large collections of training and test data are needed.Since current databases for expression learning are usually achieved in experimental environments which are different from actual emotions,and the number of samples in each database is small,it is hard to train good expression classifiers with them,which leads to a low expression classification ability,especially when recognizing expressions in web images.In this thesis,a novel facial expression database construction method is proposed:firstly,a large-scale social label images are obtained through Google,Baidu and Bing web searches with the keywords of happiness,sadness,surprise,anger,disgust and fear,respectively;secondly,unrelated images are filtered as junk images interactively by the similarity measure and hyperbolic visualization technique;and then the expression database is constructed.All the images in the database are from real social networks,therefore,the database is more proper to train classifiers for recognizing expressions in web images.Based on convolution neural network,a new facial expression recognition method is proposed.First,the method constructs the complex CNN architecture by alternately superimposed convolution layers and pool layers;Then,the method of stochastic gradient descent is utilized to learn the model parameters;Finally,one layer with the Softmax classification method is used to identify six basic expressions.The method can be more adaptive of larger changes between different expressions,thus effectively improving the expression recognition.