Characterization of Surface Topography from Small Images

Abstract

Detailed characterization of 3D engineering surface topographies is still an unresolved problem. The reasons are that the majority of the real surfaces are anisotropic and multi-scale, i.e. their directionality and roughness change with the measurement scales. To solve this problem, a variance orientation transform (VOT) method was developed. It calculates fractal dimensions at individual scales, i.e. it calculates the fractal signature (FS) in all possible directions, addressing, in this way, the problems of surfaces' multi-scale and anisotropic nature. However, the VOT method is not suited for the analysis of image sizes that are smaller than 48 × 48 pixels (e.g. images of wear particles surfaces, small surface defects, etc.). To redress this problem the VOT method was augmented so that it can calculate FSs for all images including those with small sizes. Previous study showed that the augmented VOT (AVOT) method is accurate in the analysis of hand x-ray images where the bone texture images are small (20 × 20 pixels). However, its usefulness in analysing small images of engineering surfaces has not yet been investigated. In the current study, we use range-images of different sizes (20 × 20 and 30 × 30 pixels) of polished (isotropic) and ground (anisotropic) steel plates. When applied to images of steel surfaces of different topography, the AVOT method has detected minute changes at different scales, undetectable by other commonly used surface characterization methods, between the surfaces. The results show that the method can be a valuable tool in characterization of small images of 3D engineering surfaces.