A transformation-invariant recursive subdivision method for shape analysis
Zhu Quiming, Poh Lay-Kheng · 2003
A method is presented for shape analysis by recursive subdivisions. The subdivisions are invariant with respect to translation, scaling, rotation, and intensity shifting. Eigenvectors of the second central movements are used to derive such subdivisions. A hierarchical shape description uses the invariants defined on the second central moments of the subdivisions. The approach emphasizes the local feature for recognizing object shapes as similar.>