摘要: A parallel processing algorithm for shrinking binary patterns to obtain single isolated elements, one each pattern, is presented. This procedure may be used counting on a matrix, and hardware implementation of the using large scale integrated tecnology envisioned. The principal features this method are very small window employed (two-by-two elements), nature process, possibility any regardless complexity its configuration. Problems regarding merging disconnection during process as well determination maximum number steps necessary element from reviewed discussed. An analogy with neural network description, in terms McCulloch-pitts “neurons”