It is possible to remove the traditional need for at least two subiterations of an iterative thinning algorithm [1] by expanding the window for each pixel to include edge information about its neighbors. The improved algorithm has been implemented on a 64 × 64 distributed array processor (ICL DAP) and the basic improvement in time is 40 percent, partly due to the reduced time per iteration and partly due to the reduced number of iterations. Further refinements are also proposed in the algorithm (noise reduction and staircase elimination) to yield cleaner skeletons, though a time penalty must be paid. The implementation of the algorithm is straightforward on an ICL DAP-compatible array processor. The reader will not require any specific background nor familiarity with related works.