Development of mesoscopic imaging system for surface inspection / Moe Win

Moe , Win (2018) Development of mesoscopic imaging system for surface inspection / Moe Win. PhD thesis, University of Malaya.

	PDF (The Candidate's Agreement) Restricted to Repository staff only Download (1511Kb)
Preview	PDF (Thesis PhD) Download (4Mb) | Preview

Abstract

The technology of mesoscopy brings into sight the microscopic level details of objects that are in dimensions of several meters, and the mesoscope is the device used for this purpose. The strength of this technology is that it can acquire ultrahigh-resolution images with micro to mesoscopic precision but with higher spatial resolution than a conventional microscope, and is able to solve the field of view limitations associated with optical microscopy. One of the main objectives of this is to develop and build a mesoscopic imaging system that can be used practically as an analytical tool for surface inspections. In particular, the focus is on the development of high-resolution images acquisition and advanced image processing algorithms which could be integrated with the surface inspection and diagnostic system. For this purpose, a high-resolution scanner consisting of a line CCD camera, lens, frame grabber and light source is used for acquiring high quality images of up to 5000 dpi (~5μm/pixel) with micro-mesoscopic spatial range. The post-processing system is mounted with image acquisition system, able to produce very sharp and high-resolution images. The high-resolution images of the samples are subjected to a manual inspection and automatic detection procedures. For the manual inspection, the images are processed and analyzed using raster graphics editors and the micro-meso scale defects are manually identified. The developed system is then utilised for inspection and defect detection of titanium coated surfaces, although appearing uniform, exhibits non-visible micro-cracks which remains hidden from naked eye observations. At high resolution range (above 600 dpi), non-uniformity and tiny defects could be easily detected by the developed mesoscopic imaging system. The results obtained are compared with those obtained from optical microscopy to verify the accuracy of detection. For the automatic inspection, defect detection algorithms based on image thresholding are developed and implemented as a statistical approach to identify possible defects. Existing thresholding-based and clustering-based methods are tested and compared to achieve faster and more efficient algorithms. Two new thresholding methods were developed. These are Contrast-Adjusted Otsu’s Method and Contrast-Adjusted Median-Based Otsu’s Method, which were then integrated into the automatic defect detection system. The proposed imaging system was shown to be capable of efficiently detecting microscopic surface features at any point of the scanned surface since the imaging system can provide ultra-high-resolution images. Also, as the design of the mesoscopic imaging system is built on line-scanning technology, measurements of colour fidelity and spatial accuracy is better than most other imaging systems. The significant findings of this research confirm that the proposed mesoscopic analytical imaging system is a novel approach in non-destructive measurement of surface coating quality.

Actions (For repository staff only : Login required)