Research on 3D Reconstruction based on Bionic Head-eye Cooperation

In order to solve the problem of difficult pose solution with a small amount of data input, a 3D reconstruction model of fixed pose is proposed based on bionic head-eye cooperation in this paper. Firstly, the model of the extreme positions of head-eye coordination is built and the Information Obtained Frequency (IOF) is obtained. By adopting the bionic method, the traditional 3D reconstruction method is improved, and the problem of solving poses is transformed into a fixed pose pre-processing. We obtain the fixed pose-matched RGB image and depth map through IOF, so that the bionic pose can be directly used for 3D reconstruction. This model is suitable for indoor mobile robots. On the premise of meeting the navigation requirements of the robot, a more accurate 3D reconstruction node model that conforms to human vision can be obtained.