Kathleen Iwancio, North Carolina State University


Noise Tolerant Planar Curve Matching Using Invariants


Abstract: Curve matching has a variety of applications in computer image processing and image recognition. Two curves are equivalent if they lie in the same orbit. Two curves in R² are equivalent under the action of the Euclidean group if one curve can be mapped to the other by a combination of rotations, reflections, and translations. Differential invariants for the Euclidean group are well known and can be used to solve the curve matching problem. The use of differential invariants is problematic though because derivatives are senstitive to noise. We instead use integral invariants, which are much less sensitive to noise.

Advisor: Irina Kogan (NCSU)