ABSTRACT

This chapter is concerned with the efficient implementation of “low-level” morphological transformations. It is devoted to the most classical morphological algorithms, namely the parallel ones. Solving a moderately complex image analysis application by morphological methods often involves the concatenation of several tens or hundreds of low-level transformations. The performance of a morphological algorithm may be defined using three main criteria: speed, accuracy, and flexibility. Morphological algorithms are expected to avoid some of the aberrations associated with the use of discrete grids, like the “cone effect”. The skeleton transformation is widely used in morphological image processing. An algorithm should of course give results that are as accurate as possible. In fact, most of the time, the result is expected to be totally exact. Chains and loops are particularly efficient for the computation of binary morphological transformations and give rise to the interesting extensions.