The robustness analysis on several kinds of morphological pattern spectrums

The initial form of the pattern spectrum came from the size distribution function presented by Serra (1982). The main difference between the size distribution function and the pattern spectrum is that the pattern spectrum denotes the difference information between two different sized transformations on an image. This paper presents a general representation of the pattern spectrum. Based on a typical noise model, we carry out a theoretical analysis on the robustness of several kinds of spectrums. Finally, some experiments are conducted.