Attitude Parameters Measurement Of The Human Head

Attitude Parameters Measurement of The Human Head is related to a research topic in the area of computer vision, human computer interaction, gaze tracking, fatigue testing. The study has important significance in theory and practical application. Focused on the head position parameter detection, this paper proposed a binocular vision measuring method based on the summary of research status in china and abroad.Based on the camera calibration, the paper described the principles of how to implement parameter measuring by binocular vision system. Proposed the method of how to complete the head position and attitude parameters by use binocular vision of perspective projection model: Use left and right images in binocular parallax and image coordinates of the point location to obtain the three dimensional coordinates of destination points and the position parameters, to get the positon parameters of X,Y and Z; using the rotation matrix to solve the angle of rotation to get the head angle parameters roll, pith and yaw.The main focus on camera calibration, binocular vision and stereo matching, face detection and facial feature extraction of these four parts, This paper introduced each of these key technologies, and at every subsequent chapters have carried out the appropriate experiments: Obtain the camera parameters by the calibration of camera; Get the left and right images with the structural subjects of binocular system, using stereo matching for image corresponds to find the feature; Face detection is determined by the position of human face images to determine the location of the head, face detection method of Adaboost algorithm is used in this paper; Facial feature extraction is realized using AAM (Active Appearance Mode)feature to locate and match. Face shape models and textures models are included in AAM, which can locate the facial feature perfectly, then reverse method is used to complete the combination of human face images and model.Finally,complete the subject head attitude measurement by coordinate the corresponding image points and the detected feature points, and get the head position and orientated parameters X,Y,Z and the attitude parameters of roll, pitch and yaw.