A Hierarchical Surface Measurement Strategy Using Non-Repeating Patterns

Hierarchical approaches for automated photogrammetric surface measurement invariably employ image pyramids and image matching at multiple resolutions. Such an approach is also applicable in industrial photogrammetry where projected light patterns are used to provide texture on an otherwise texture-free surface. This paper describes a novel strategy for hierarchical surface measurement in which projected non-repeating target patterns are employed, rather than the lower-resolution levels of an image pyramid. Thus, although the final derived digital surface model is obtained by image matching, preliminary surface models which support the matching are obtained using the non-repeating patterns. These projected patterns allow a simplification of the triangulation phase since they offer the potential of a fast and unambiguous solution to the image point correspondence problem. Following an introduction to the non-repeating pattern strategy, integration of the approach into a multi-camera vision metrology system is described and some experimental measurements are summarized