A topological characterization of thinning (Q1091154): Difference between revisions

This is a very interesting and fundamental paper proving a basic assumption underlying almost all thinning algorithms. Namely, that to remove one set D from another set F, with the desire to preserve the topology of F, one can do so pixel by pixel, testing only if the number of connected components in figure and background change. The paper neatly proceeds from the notion of preservation of digital topology, to bijectivity of mappings between components, to removal of components preserving the bijectivity (called 'strong-deletability'), to removal of a single pixel satisfying the conservation of components in figure and background ('deletability'). Well written, concise, and highly readable. A problem in reading is that 'S is k-connected' is used in a double sense: in one sense it means 'S consists of components determined by the fact that points are k-connected', in the second sense it means 'S consists of 1 component whose points are k-connected'. In all theorems and derivations the second sense is used, in some definitions (1 and 2), examples and figures 1 and 2 the first sense is apparently used. This is confusing, since application of the theorems to the examples produces contradictions. One should read '(strongly) deletable' and 'componentwise (strongly) deletable' to resolve these problems. This intermediate step, removing a set component by component, ought to have been indicated more clearly.