Parallel Image Processing Primitives

The most common tasks of low level image processing are considered in order to define a set of primitives to be implemented in parallel image processing systems. In particular arithmetic primitives, global features (as grey level statistics), linear transforms and local computations are discussed taking care of the effectiveness of the implementations on array and pyramidal systems. The propagation capability supplied by flat arrays and pyramids is then presented in some extent with the purpose to highlight the performances of this kind of machines in well known low level tasks.