On Facial Expression And Face Pose Recognition Of The Senior And Disable People Aid's Robot

In order to solve the problems of the senior and disable people aid's robot in human-computer interaction,a visual system with the ability of facial expression recognition and face pose recognition was designed,and was successfully applied to the robot.The system enables the senior and disable people aid's robot to perform corresponding actions according to their expressions and pose.The visual system studied in the thesis mainly includes two parts:facial expression recognition and face pose recognition.Firstly,the methods of face detection and face feature point alignment are introduced briefly.Secondly,in the facial expression recognition method,the Adaboost algorithm is firstly used to train the face detection classifier,and then the classifier is used to obtain the position of the face in the image,and finally the face image is obtained.For the obtained face image,the random forest method is used to extract the sparse local binary feature(i.e.LBF feature).Then the alignment of the face feature points is realized,and the outline description of the person's online facial features is obtained.At last,the support vector machine is used to classify different facial expressions.Thirdly,face pose recognition is studied on the basis of facial feature point alignment.Face pose recognition is a dynamic process of face recognition in video.Similar to facial expression recognition,it is necessary to locate the face first.For dynamic video,the thesis adopts a tracking method to realize the positioning and feature point alignment of the face part in the video.According to the extracted feature points,the feature vectors of face pose in video are constructed,and the recognition of face pose is realized by using support vector machine.After obtaining the recognition results of facial expressions and gestures,according to the different recognition results,the thesis establishes a mapping relationship between the robot action instructions and the facial expressions and pose recognition instructions,and constructs a set of instructions for the old and disabled robot.According to different recognition results,the robot is controlled to complete different tasks,it enables the senior and disable people aid's robot to complete relevant actions according to relevant visual instructions.In conclusion,through the above researches,the thesis studies the recognition of facial expression and posture based on the senior and disable people aid's robot,achieves good results on the built data set and the existing data set,and realizes the combination of vision system and the senior and disable people aid's robot.