City Research Online

Abstract

Scanning systems for the optical inspection of flat surfaces, moving at high speed, have been developed and are in use in many industries. These systems produce an electrical signal describing the inspected surface, and incorporate signal processing capable of detecting many of the signals which arise from surface defects.

The work reported in this thesis is concerned with the possibility of identifying the defect type from an analysis of the profile of the signal generated by the defect as it is scanned.

The variation inherent in this signal, due to both the characteristics of the scanning systems and to variation between individual examples of the same defect, lead to the conclusion that the statistically based methods of feature space pattern recognition hold the most promise for defect identification. These techniques are reviewed, and those best suited to the system requirements are selected for further study. Most prominent among these requirements are those of fast processing and acceptable cost of implementation. The selected techniques are combined and extended, where necessary, into a set of programs for system design and comparative evaluation.

A data base from sheet tinplate is acquired on magnetic tape, using scanners developed by the SIRA Institute, and used to evaluate the selected techniques. With a suitable combination of these, 80% correct identification is achieved over five defect classes. However, this requires manual intervention in the processing chain so as
adequately to delineate (define the limits of) each defect signal, so that measurements can then be made upon it. The systems available for detecting the signals were found to be ineffective for their delineation. A system is therefore developed, based on a bank of matched filters, and shown to provide signal detection and delineation as good as, or better than, that achieved with manual intervention.

A hardware design for the preferred system is developed in detail.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science Q Science > QC Physics T Technology > TA Engineering (General). Civil engineering (General)
Departments:	School of Science & Technology > Department of Engineering School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads