Non-contact estimation of surface roughness in turning using computer vision and Artificial Neural Networks. (26th March 2009)

• Record Type: Journal Article
• Authors: Shome, D.; Ray, P.K.; Mahanty, B.
• Is Part Of: International journal of industrial and systems engineering. Volume 4:Number 4(2009)
• Page Start: 349
• Page End: 367
• Publication Date: 2009-03-26
• Subjects: computer vision -- non-contact surface roughness estimation -- turning -- GLCM -- grey-level co-occurrence matrix -- ANNs -- artificial neural networks -- Bayesian regularisation -- image textural features; Systems engineering -- Periodicals; Industrial engineering -- Periodicals; 620.001171
• Languages: English
• ISSNs: 1748-5037

Abstract

Accurate non-contact estimation of surface roughness in turning operations plays an important role in the manufacturing industries. This paper investigates the effectiveness of using various surface image features, such as contrast, energy, homogeneity, entropy, range, and standard deviation for computer vision-based non-contact estimation of surface roughness in turning operations. A Bayesian Regularisation-aided Artificial Neural Network (ANN) model-based approach is proposed in this paper for accomplishing surface roughness estimation. Analyses of experimental data demonstrate that the proposed approach yields significant improvement in the accuracy level of computer vision-based non-contact estimation of surface roughness (without involving turning parameters) of turned workpieces.