Indrazno Siradjuddin, Gillang Al Azhar, Cahya Rahmad, Fatkhur Rohman, Indah Martha Fitriani, Erlillah Rizqi Kusuma Pradani, Erfan Rohadi, Ferdian Ronilaya, Rosa Andrie Asmara
Rigid transformation estimation problem appears often in robotics and computer vision. The problem is also referred to absolute orientation. Current developed algorithms for determining rigid transformation rely on factorisation of matrices, for instance, singular value decomposition or eigenvalue and eigenvector. This technique is relatively computationally expensive, thus it may be a burden on low-end hardware when the absolute orientation needs to be computed continuously. This work presents a closed-form solution to the least squares problem focussing on three or more pairs of point vector in 2D space. The proposed algorithm has also considered the scaling factor as a part of the rigid transformation. The simulation results shows the effectiveness and the numerical accuracy of the proposed method. © 2018 IEEE.
Electrical Engineering Department, State Polytechnic of Malang, Malang, Indonesia; Informatics Engineering Department, State Polytechnic of Malang, Malang, Indonesia; Mechanical Engineering Department, State Polytechnic of Malang, Malang, Indonesia