The Growing Window Algorithm: a Sub-Optimal Strategy for Image Irregular Samples Selection

The new algorithm proposed in this paper is a novel method for sample selection based on a quasi-random search within an image. As far as the general sampling problem is concerned, a brief introduction on the well-known sampling algorithm is presented in order to introduce the most important parameters to be taken into account for the performances evaluation of the novel method here proposed. Then, the Growing Window Irregular Selection is described following few steps: obviously, the purpose of this irregular sampling strategy is to increase the sample density in the zones of strong luminance variance of the image. On the contrary, in zones with no variation almost no sample are allocated. Several tests have been performed on different images (e.g. geophysical and medical) and interesting results have been outlined.