Multi-ocular stereoscopic photogrammetry via PTV-based feature-point matching algorithm: a comparative study with DIC

Digital image correlation (DIC) is widely used in multi-ocular stereoscopic photogrammetry for non-contact three-dimensional (3D) surface measurement. Despite its flexibility and wide application, limitation related to spatial resolution remains challenging, particularly in measurement of complex surfaces with steep local gradients. This study proposes a feature-point matching scheme based on particle tracking velocimetry (PTV) to improve surface morphology measurement and quantitatively compares it to the traditional DIC-based approach. Unlike DIC, which matches subsets of texture on the surface, the PTV-based scheme matches individual feature points in different image planes with various viewing perspectives based on the geometric constraint, and then triangulates matched feature points to restore the 3D feature-point cloud. A wave-like surface was measured using both schemes, and a comprehensive performance comparison was conducted across a wide range of feature-point densities. The result demonstrates that the DIC-based scheme suffers from spatial averaging, introducing bias in depth and gradient measurements. As a contrast, PTV-based scheme is free from the spatial averaging effect, and it outperforms DIC-based scheme in terms of outlier ratio, depth error, and gradient resolution. Furthermore, its performance is less sensitive to the variation of feature-point densities, and the problem of selecting a proper subset size in DIC-based scheme is avoided. Since PTV-based feature-point matching scheme provides both high accuracy and high spatial resolution, it will serve as an ideal candidate for the DIC-based scheme for surface morphology measurement.