In order to detect the defects in a random surface, the local fractal dimension(LFD) of the random surface image and its anisotropy are studied in this paper. An algorithm to estimate the anisotropy of local fractal dimensions based on moments of the increments is obtained, and the ratio of horizontal and vertical differential of LFD which can characterize the anisotropy of local fractal dimension is proposed. The parameter overcomes the local textual irregularities by representing the region stationary of a non-stationary rough surface. Images of variant random surfaces and defects are investigated. The results of experiments show that the method and parameter proposed are robust, which extract local morphologic features of roughness anisotropy in random surfaces distinctly. By measuring the rupture joint of the ratio, the defects in random surfaces can be detected and located sensitively.

Date Published: 30 November 2009
PDF: 9 pages
Proc. SPIE 7506, 2009 International Conference on Optical Instruments and Technology: Optical Systems and Modern Optoelectronic Instruments, 75062E (30 November 2009); doi: 10.1117/12.837230

Published in SPIE Proceedings Vol. 7506:
2009 International Conference on Optical Instruments and Technology: Optical Systems and Modern Optoelectronic Instruments
Yongtian Wang; Yunlong Sheng; Kimio Tatsuno, Editor(s)