Stability of Hausdorff-based Distance Measures

M.D. Shapiro and M.B. Blaschko (USA)


Hausdorff, Object Recognition, Spatially Coherent Match ing


A number of Hausdorff-based algorithms have been pro posed for finding objects in images. We evaluate different measures and argue that the Hausdorff Average distance measure outperforms other variants for model detection. This method has improved robustness properties with re spect to noise. We discuss the algorithms with respect to typical classes of noise, and we illustrate their relative per formances through an example edge-based matching task. We show that this method produces a maximum a poste riori estimate. Furthermore, we argue for improved com putational efficiency using the Hausdorff Average distance measure over other variants.

Important Links:

Go Back