Comments on "Classification of natural textures by means of two-dimensional orthogonal masks" by P. Cohen, V. Lacasse and C.T. DeDinh

The commenter discusses a texture classification technique by Cohen et al. described in the above-titled correspondence (see ibid., vol.37, no.1, p.125, 1989) that uses features derived from a set of four orthogonal masks applied to a 2*2 local neighborhood. The technique is an extension of the sum and difference histograms method previously published by the commenter. He points out that he has published a technique that is more general and subsumes the results of Cohen et al. and of his own earlier work. He also points out a mathematical error in the work by Cohen et al., which from a practical point of view, it is likely to have only minor effects.<
>