An efficient and accurate 3D displacements tracking strategy for digital volume correlation
KAUST DepartmentPhysical Sciences and Engineering (PSE) Division
Mechanical Engineering Program
Composite and Heterogeneous Material Analysis and Simulation Laboratory (COHMAS)
Permanent link to this recordhttp://hdl.handle.net/10754/563001
MetadataShow full item record
AbstractOwing to its inherent computational complexity, practical implementation of digital volume correlation (DVC) for internal displacement and strain mapping faces important challenges in improving its computational efficiency. In this work, an efficient and accurate 3D displacement tracking strategy is proposed for fast DVC calculation. The efficiency advantage is achieved by using three improvements. First, to eliminate the need of updating Hessian matrix in each iteration, an efficient 3D inverse compositional Gauss-Newton (3D IC-GN) algorithm is introduced to replace existing forward additive algorithms for accurate sub-voxel displacement registration. Second, to ensure the 3D IC-GN algorithm that converges accurately and rapidly and avoid time-consuming integer-voxel displacement searching, a generalized reliability-guided displacement tracking strategy is designed to transfer accurate and complete initial guess of deformation for each calculation point from its computed neighbors. Third, to avoid the repeated computation of sub-voxel intensity interpolation coefficients, an interpolation coefficient lookup table is established for tricubic interpolation. The computational complexity of the proposed fast DVC and the existing typical DVC algorithms are first analyzed quantitatively according to necessary arithmetic operations. Then, numerical tests are performed to verify the performance of the fast DVC algorithm in terms of measurement accuracy and computational efficiency. The experimental results indicate that, compared with the existing DVC algorithm, the presented fast DVC algorithm produces similar precision and slightly higher accuracy at a substantially reduced computational cost. © 2014 Elsevier Ltd.
SponsorsThis work is supported by the National Natural Science Foundation of China (Grant nos. 11172026, 11272032, and 11322220), the Program for New Century Excellent Talents in University (Grant no. NCET-12-0023), the Science Fund of State Key Laboratory of Automotive Safety and Energy (Grant no. KF14032).
JournalOptics and Lasers in Engineering