On improving sub-pixel accuracy by means of B-spline

Local perturbations nearby contours strongly perturb the final result of processing remotely sensed images (RSI). It is common to establish a priori data to aid the estimation process. One can move some steps forward by means of a deformable model, for example, the snake model. In up to date research, the deformable contour is represented via B-spline snakes, which allows local control, concise depiction, and the use of fewer parameters. The estimation of edges with sub-pixel accuracy via a global B-spline depiction depends on determining the edge according to a Maximum Likelihood (ML) agenda and using the observed information likelihood. This practice guarantees that outliers present in data will be cleaned out. The data likelihood is calculated as a result of the observation model comprising both orientation and position data. Experiments where this procedure and the traditional spline interpolation have revealed that the algorithm introduced outperforms the conventional method for Gaussian as well as Salt and Pepper noise.