Designing indices to measure surface roughness based on the color distribution statistical matrix (CDSM). (June 2018)

• Record Type: Journal Article
• Authors: Lu, Enhui; Liu, Jian; Gao, Rongyu; Yi, Huaian; Wang, Weifang; Suo, Xinyu
• Is Part Of: Tribology international. Volume 122(2018)
• Page Start: 96
• Page End: 107
• Publication Date: 2018-06
• Subjects: Surface roughness -- CDSM -- Index design -- Index evaluation; Tribology -- Periodicals; 621.89
• DOI: 10.1016/j.triboint.2018.02.033
• Languages: English
• ISSNs: 0301-679X

Abstract

Abstract: At present, in machine vision-based roughness measurement methods, indices are mainly designed based on grayscale images without considering the advantages of multidimensional information of color images, which makes the corresponding algorithms complex and inaccuracy. Meanwhile, performance evaluation methods of the image-based indices mainly focus on the prediction accuracy, ignoring the monotonicity, stability and efficiency of the indices. To address these problems, a color image-based indices design and evaluation method is proposed in this article. Two specific indices are designed based on the proposed method. The experimental results prove the feasibility, efficiency and universality of the proposed indices. Besides, the index performance evaluation method can quantitatively and comprehensively characterize the performance of different indices. Highlights: A methodology to design CDSM-based indices concerning clear physical meanings is firstly proposed. A scientific roughness-correlated indices performance evaluation methodology is firstly introduced. Two specific indices with clear physical meanings and practical engineering values are designed.