Bi-dimensional Variational Mode Decomposition for Surface Texture Analysis
Zhuowei Li,Yuanping Xu,Tukun Li,Yajing Shi,Xiangqian Jiang,Yanlong Cao,Wenhan Zeng,Zhijie Xu,Chaolong Zhang,Jian Huang
DOI: https://doi.org/10.1016/j.procir.2022.10.006
2022-01-01
Procedia CIRP
Abstract:In surface metrology, filtration is one of the key operations which is used to decompose the components of the surface and extract a scale-limited surface for further assessment. Although many types of filters have been proposed and some of them have been standardized in ISO16610 serials, the classic surface topography filters often inherit the mode mixing and boundary distortion problems that may lead to surface texture verification failures. This study proposes a new filter algorithm for the surface texture analysis, and it is based on the extended bi-dimensional variational mode decomposition (BVMD), named EBVMD. BVMD is widely used in the field of image processing and it is probably first-time to be used in the field of surface texture analysis. It consists of three steps. Firstly, a coarse-grained parallel genetic algorithm is applied in the devised model to select of optimal penalty factor and decomposition number of the bi-dimensional variational mode, and the best set of modes is gained. Secondly, the instantaneous wavelength value of each mode is obtained by calculating the corresponding isotropic monogenic signals. Finally, the scale-limited surface is extracted by the mean instantaneous wavelength. Experiments are conducted to verify the feasibility of the proposed filter on areal surface texture feature extraction with the mode aliasing and shape distortion phenomenon remedies. The experimental results show that the mean square error of roughness reconstructed by the proposed filter compared with the benchmarking values is 6.37×10-5, and owing to high flexibility in practice, the devised EBVMD provides a promising solution to achieve a high accuracy and efficiency filter.