Discovering Structured Variations via Coupled Template Matching

Understanding patterns of variation from raw measurement data remains a central goal of shape analysis. This understanding reveals which elements are exactly repeated, or how elements can be derived as structured variations from a common base element. We investigate this problem in the context of multi-view stereo reconstruction of buildings. Utilizing a set of template models, we establish geometric and semantic relationships among structural elements found in the data. Each template is equipped with a deformation model that defines correlated variations of a base geometry. Central to our algorithm is a coupled template matching and deformation analysis. This analysis simultaneously detects patterns across building elements by extracting similarities in the deformation modes of their matching templates. We evaluate our algorithm on several challenging datasets and demonstrate that such a coupled analysis can successfully detect structured variations even for noisy and incomplete data