Research And Application Of Multimodal Expression Recognition In SEEE Model

The energy problem is one of the important problems which restrict thedevelopment of the national economy. Building energy is the main part of energyconsumption, how to solve the problem of building energy consumption andimplement building energy efficiency have a great significance to the development ofour country. With the improvement of people’s living standard, people’s demand forconstruction also gradually from function research、the performance research to thelevel of building comfort、health and research of building energy efficiency. Buildingenergy conservation should not be at the cost of the detriment of comfort and health,and SEEE model is realized the building energy smart in consideration of thearchitectural comfort environment and the condition of health to make sure thebuildings can adapt to changes in the environment, and access to healthy、comfortable and energy saving states in independent.Affective computing is to allow the computer to get the human’s emotions and tounderstand the emotional state of humans. Affection is just a kind of cognitive thatemotional subject obtains from the external environment condition which can directlyreflect the person’s subjective experience, and the level of comfort is just a experienceform the state of the building environment. Through the affective computing torepresent people’s perception of the building comfort is just an effective assessment toachieve building energy efficiency and the condition of the building comfort and alsocan be the theory of the SEEE model which realized the intelligent energy and thebuilding environment can be adaptive by itself. There is55%of the becomingemotional through the facial in the emotional expression. Facial expression can reflectaffection and affection can describe the state of comfort. Using the relationshipbetween the expressions and building comfort can directly reflect the awareness ofbuilding comfort by people.The purpose of this dissertation is just to study the effective method of facialrecognition and give effective evaluation through facial recognition on people’sperception of comfort and affective computing theory can be applied to SEEE modelto realized building energy conservation in the circumstances of the healthy andcomfortable. Provide effective theoretical support for building energy conservation.Firstly, this article obtain the expressions images, ECG data, and galvanic skin datathrough emotional induced from experimental subjects, analyze and screening thevalidity of data through the self evaluation questionnaire, the images can be intuitivereflect the affection of the subjects. By using the method of singular valuedecomposition to extract the feature of images；Access to the characteristics of ECGand GSR signals by using the method of wavelet decomposition. Although images candirectly reflect the affection, but also has the conceals nature’s and fraudulence whilethe ECG and electrical signals can objectively describe the emotional state ofparticipants, adopt the method of PCA and fusion of the characteristics of differentmodels in dimension reduction way, the fusion features included the intuition of visualimage features and the objectivity of physiological signals which can be moreeffective to characterize the participants’ emotional state. Finally using the BP neuralnetwork classifier to classify and recognize the features of single-mode characteristicsand multimodal fusion characteristics. BP neural network is easy to fall into the defectof local minimum. Genetic algorithm can optimize the BP neural network classifierand avoid the defects of BP network effectively then access to a more reasonablenetwork structure. Finally using the BP neural network classifier and the optimized one to recognize the characteristics of single-mode and fusion feature respectively,and analyzed the result of the experiment. Studies also show that characteristics ofmultimodal fusion expression improve the recognition rate more effectively.