The Pain-expression Recognition And Application Study Of Human Being Based On Video

Pain is not only a sensation,but also the most valuable self-protection mechanism of the body.In the clinical environment,pain is difficult to assess and manage,and pain assessment is an important reference element in medical diagnosis and treatment.A good pain assessment can improve the quality of life of patients,reduce the fear of pain,avoid drug abuse,let medical personnel know more about patients' pain,use drugs in advance to achieve better efficacy,and many other benefits.Therefore,the research on pain facial expression recognition is of great significance and wide application prospects for medical clinical pain assessment,and it is gradually attracting more and more attention from medical and scientific researchers.The research of human facial expression recognition mainly includes the steps of acquiring the original data set(image / video),face detection,preprocessing,feature extraction and classification.Among them,the extraction and classification of facial expression features are the two most important steps.Accurate and effective feature extraction can greatly improve the recognition rate of pain expressions,and a suitable expression classification method can reduce the algorithm complexity and improve system performance.In view of the shortcomings of current facial expression recognition,this paper has carried out the following research based on the predecessors:1)This article proposes a feature extraction method based on the combination of Additive White Gaussian Noise(AWGN)and Discrete Cosine Transform(DCT).The Gaussian noise processing of the image is used to weaken the high-frequency features of the model.Influence,and then compress the spatial signal by discrete cosine transform to achieve more reasonable and accurate pain recognition.Through the experimental verification on the shoulder pain expression database of UNBC-Mc Master,the recognition rate of this method reaches 83.6%,which shows the effectiveness of this method.2)For the existing pain expression recognition research,mostly based on static images or single-frame images of video,the lack of motion information between adjacent frames and time domain information extraction.The Optical Flow Method can just make up for the lack of dynamic information in the time domain,and no researchers have yet used the optical flow method to study the classification of pain expression recognition.This paper proposes a pain expression classification method based on the combination of optical flow method and 3D Convolutionnal Neural Network(C3D): Active Appearance Model extracts C-APP appearance features and optical flow features togetherinto C3 D network,Realize the classification of pain expression.The effectiveness of the method proposed in this paper is proved through experiments comparing with other frontier methods.3)Based on the method mentioned above,combining Sina cloud server resources and We Chat public platform,by deploying preprocessing,feature extraction related code and trained convolutional neural network to Sina cloud server,and interface with We Chat public platform The design and development of the pain expression recognition system have realized the basic functional requirements for the classification of pain expressions by taking pictures on the mobile phone and calling the camera for real-time sampling.