Performance evaluation of various texture analysis techniques for machine vision-based characterisation of machined surfaces
by Ketaki Joshi; Bhushan Patil
International Journal of Computational Vision and Robotics (IJCVR), Vol. 10, No. 3, 2020

Abstract: Machine vision-based inspection of surface quality leverages the principle of surface-texture characterisation, capitalising on image data characteristics. Frequently, surface-texture analysis adopts statistical and filter-based techniques, for this purpose. For surface texture characterisation, traditionally researchers prefer – parameterised histograms, grey level co-occurrence matrices, discrete Fourier transforms as well as discrete wavelet transforms. Despite popular usage, extant literature features very little in terms of comparative analyses amongst these techniques. Accordingly, this paper evaluates comparative performance of these techniques, for characterisation of machined surfaces and also recommends a novel hybrid technique that leverages higher discriminating capability. This hybrid discriminant-analysis methodology is derived from characterisation of 532 images of multi-textured machined surfaces. The results prove that the proposed technique, provides superior performance with higher accuracy, while requiring reduced optimal set of parameters, for inspection of surface quality.

Online publication date: Mon, 11-May-2020

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Computational Vision and Robotics (IJCVR):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com