Carlos Lopez-Franco1, Javier Gomez-Avila2, Nancy Arana-Daniel3, Alma Y. Alanis
Intelligent Automation & Soft Computing, Vol.24, No.2, pp. 431-442, 2018, DOI:10.31209/2018.100000000
Abstract This paper presents a method for 3D pose estimation using visual information
and a soft-computing algorithm. The algorithm uses quaternions to represent
rotations, and Particle Swarm Optimization to estimate such quaternion. The
rotation estimation problem is cast as a minimization problem, which finds the
best quaternion for the given data using the PSO algorithm. With this
technique, the algorithm always returns a valid quaternion, and therefore a
valid rotation. During the estimation process, the algorithm is able to detect
and reject outliers. The simulations and experimental results show the
robustness of algorithm against noise and outliers. More >