Hybrid 3D Dynamic Measurement by Particle Swarm Optimization and Photogrammetric Tracking

High-accuracy motion modeling in three dimensions via digital images has been increasingly the matter of interest in photogrammetry and computer vision communities. Although accurate sub-pixel image registration techniques are the key elements of measurement, they still demand enhanced intelligence, autonomy, and robustness. In this paper, a new correlationbased technique of stereovision is proposed to perform inter-frame feature tracking, inter-camera image registration, and to measure the 3D state vector of features simultaneously. The developed algorithm is founded on population-based intelligence (particle swarm optimization) and photogrammetric modeling. The proposed technique is mainly aimed at reducing the computational complexities of non-linear optimization methods of digital image registration for deformation measurement, and passing through 2D image correlation to 3D motion modeling. The preliminary results have illustrated the feasibility of this technique to detect and measure sub-millimeter deformations by performing accurate, sub-pixel image registration.