Local perturbations around contours strongly disturb the final result ofcomputer vision tasks. It is common to introduce a priori information in theestimation process. Improvement can be achieved via a deformable model such asthe snake model. In recent works, the deformable contour is modeled by means ofB-spline snakes which allows local control, concise representation, and the useof fewer parameters. The estimation of the sub-pixel edges using a globalB-spline model relies on the contour global determination according to amaximum likelihood framework and using the observed data likelihood. Thisprocedure guarantees that the noisiest data will be filtered out. The datalikelihood is computed as a consequence of the observation model which includesboth orientation and position information. Comparative experiments of thisalgorithm and the classical spline interpolation have shown that the proposedalgorithm outperforms the classical approach for Gaussian and Salt & Peppernoise.
展开▼
