The effect of sub-sampling in scale space texture classification using combined classifiers

Textures show multi-scale properties and hence multiresolution techniques are considered appropriate for texture classification. Recently, the authors proposed a multiresolution texture classification system based on scale space theory and combined classifiers. However, the use of multiresolution techniques increases the computational load and memory space required. Sub-sampling can help to reduce these side effects of multiresolution techniques. However, it may degrade the overall performance of the classification system. In this paper the effect of sub-sampling is investigated in scale space texture classification using combined classifiers. It is shown that sub-sampling can help to reduce both computational load and memory space required without compromising the performance of the system.