The study of rough textured surface are generally made on grey level images. These studies suppose, that the variations of grey levels are representative of local variations of relief. This assumption, in the case of uniformly colored rough surfaces, finds its limit in the case when these rough surfaces present variations of color or aspect. The corresponding images will then present variations of grey levels which can be related to variations of color or to variations of surface relief or both. It becomes difficult in this case to work out criteria of roughness based on the analysis of grey levels of the surface. It is necessary to work out, before any study of roughness, a method of separation of information related to the color of that related to the relief in grey level images. We propose here to carry out this separation through frequential decomposition of the image starting from a wavelet decomposition. We have tested the effectiveness of our approach by comparing the evolutions of the criteria of roughness on various walls, without separating the information of color from that of relief, with the evolutions obtained after frequential separation of these two parameters. The results obtained show that the developed approach leads to a better discrimination by the criteria of roughness in the case of colored surfaces. In the absence of color, the method suggested does not affect the elaborated criteria of roughness.
展开▼
