Training‐based head pose estimation under monocular vision

Abstract

Although many 3D head pose estimation methods based on monocular vision can achieve an accuracy of 5°, how to reduce the number of required training samples and how to not to use any hardware parameters as input features are still among the biggest challenges in the field of head pose estimation. To aim at these challenges, the authors propose an accurate head pose estimation method which can act as an extension to facial key point detection systems. The basic idea is to use the normalised distance between key points as input features, and to use ℓ1‐minimisation to select a set of sparse training samples which can reflect the mapping relationship between the feature vector space and the head pose space. The linear combination of the head poses corresponding to these samples represents the head pose of the test sample. The experiment results show that the authors’ method can achieve an accuracy of 2.6° without any extra hardware parameters or information of the subject. In addition, in the case of large head movement and varying illumination, the authors’ method is still able to estimate the head pose.

Pose Determination of a Robot Manipulator Based on Monocular Vision

“…If a feature point on the manipulator is selected, then the coordinates (x, y, z) of the point can be expressed as functions of joint angles of the manipulator through its kinematics equations. Thus, in order to solve (21) for three unknowns (x, y, z), the kinematics equations as shown in ( 8)-( 19) will be substituted into (21) for each feature point, so each feature point provides two equations in terms of six unknowns, which are five joint angles and the gripper gap of the manipulator. Furthermore, one defines four feature points to generate eight equations to be solved for the six unknowns, and the multivariate nonlinear regression will be performed to solve this problem.…”

“…Wu et al [20] developed a method to measure the position and attitude parameters of a weld stud using monocular vision, where the mathematical model related to the position and attitude of the weld stud is included. Guo et al [21] presented a 3D head pose estimation method based on monocular vision, and the method uses a linear combination of the head poses corresponding to some training feature sampling points. Sharma and D'Amico [22] assessed the performances of three initial pose estimation techniques for spacecraft to complete the formation-flying and on-orbit missions based on monocular vision, where the techniques uses a minimum number of features to estimate the spacecraft pose with respect to the camera.…”