Efficient retrieval of deformed and occluded shapes

摘要

We propose an approach for matching deformed and occluded shapesusing dynamic programming. Our algorithms handle noise and shapedistortions by allowing matching of merged sequences of consecutivesmall segments in a shape, with larger segments of another shape, whilebeing invariant to translation, scale and orientation transformations ofshapes. We illustrate the effectiveness of our algorithms in retrievalof shapes on two different two-dimensional datasets: one of static handgesture shapes and another of marine life shapes