Use of the sigma probability in Tomita's filter

An image smoothing algorithm, the Sig-Tomita algorithm, based on the sigma probability used in the sigma filter and the directional method used in the Tomita's filter is proposed. The sigma probability used in sigma filter is integrated with Tomita's (1977) directional smoothing process. In this approach, the geometric shape of the subregion selected is varied instead of a square shape as used in most edge-preserving algorithms. It is shown that a more homogeneous neighbourhood will be found for smoothing with the proposed algorithm. This method is effective in smoothing complex-shaped images such as remotely sensed data. The regularity of the masks used in the algorithm is simple to implement. The comparison with Tomita's and sigma filters is illustrated.