Local decomposition of gray-scale morphological templates

Template decomposition techniques can be useful for improving the efficiency of imageprocessing algorithms. The improved efficiency can be realized either by reorganizing a computation to fit a specialized structure, such as an image-processing pipeline, or by reducing the number of operations used. In this paper two techniques are described for decomposing templates into sequences of 3×3 templates with respect to gray-scale morphological operations. Both techniques use linear programming and are guaranteed to find a decomposition of one exists.