Face-camera Active Pose Coordination

Intelligent visual measurement platform based on synchronous belt motion controller and industrial camera has been widely used in indoor close-range photogrammetry,indoor video intelligent monitoring,real-time tracking and monitoring of moving face and other fields.The camera can not track and monitor the moving face in a fixed scene.In order to realize the real-time tracking and detection of moving face,this paper proposes a method of active pose collaboration between face and camera.The key problem is to establish the relative position relationship between the face and the camera,and to solve the coordinated pose between the face pose parameters and the camera.In this paper,the method based on three-dimensional model is used to calculate the face pose parameters,and the photogrammetric basis of the face pose solving model is deduced.The pose angle of moving face and the position of camera relative to face are solved by the single image space rear rendezvous.Intelligent vision measurement platform is built by industrial camera and synchronous belt motion controller.Mapping relationship among face coordinate system,camera coordinate system and motion controller is established.Collaborative pose of camera is calculated by using face attitude angle and relative position of camera at adjacent time.The calculation of face attitude angle and coordinated control of face and camera pose are realized.Aiming at the noise of pose estimation in the process of face-camera cooperation,Kalman filter is used to improve the accuracy of pose estimation.In the experiment,Adaboost method is used for face region detection and ERT method for face feature point detection.In order to verify the effectiveness of face pose collaboration,the experiments of face pose estimation and face pose collaboration are carried out.The experiments of pose coordination show that:(1)In the motion range of motion controller,the camera can coordinate with the moving face in real time and actively,and the position error of the camera on the motion axis is less than 10 mm.(2)When the pulse output of stepping motor is controlled at 3000PPU/S,the pose cooperative system has good stability and meets the needs of real-time face tracking and detection.