Uncertainty coding and controlled data reduction using fuzzy-B-splines

In this paper, a method is described for the representation and reconstruction of single-valued surfaces given as sets of measured data, which may be uncertain as well as crisp. In the case of imprecise data, the fuzzy B-spline representation is able to keep track of uncertainty and provide tools for interrogating the model at prescribed presumption levels. In both cases, a very high degree of compression can be achieved through a procedure which defines, among spatially-clustered points, the most significant representative of the local neighbourhood. Experimental results are shown to prove the effectiveness of the proposed approach.