Movement computation of an articulated body using coordinate transformation

Abstract

AbstractIn this paper, we propose a new formula for calculating the motion of an articulated rigid body using images obtained with a moving camera with an autofocus ability. Each of the input image scenes consists of moving joints with their own position and orientation unknown. Based on the computed coordinate system of the previous joint, the triangulation method finds the position of the feature point, and the computed feature point establishes the coordinate system for the next adjacent joint. First, the external and internal parameters of a camera are calculated using the optimization technique using the feature points in fixed environmental objects. The first joint is defined as the joint in contact with the environment. Based on the first joint, the movements of subsequent joints are computed in order. Afterwards, we calculate the motion for the connected joint with respect to the previous joint in a tree structure. To verify the accuracy of the movement of a model, a three-dimensional reconstruction was performed by using computed joint movements. The proposed formula is verified by the result of 3D reconstruction.