Please use this identifier to cite or link to this item:
Title: Development of an ultra-precision diamond-machined polar microstructure for computer vision-based precision positioning and nanoscale measurement
Authors: Zhao, Chenyang
Advisors: Cheung, C. F. (ISE)
Keywords: Measurement
Computer vision
Issue Date: 2019
Publisher: The Hong Kong Polytechnic University
Abstract: There are many measurement principles and methods for precision positioning measurement. The current challenges are generally due to the precise calibration of the key components and the strict environmental requirements, etc. Computer vision takes full advantage of the algorithms and software to reduce the requirements for hardware, which greatly reduces production costs. A new precision measurement idea has emerged, i.e. precision measurement integrated with computer vision technology. There is relatively little research on computer vision focused on the precision measurement area mainly because the current vision resolution is not satisfied. Microscopy can feasibly improve the resolution but most natural objective surfaces do not have easy-to-distinguish and regular textures under a microscope; therefore, it is difficult for pattern recognition and global matching to be completed by image processing. To address the above issue, designed surface pattern with a high-resolution of pattern recognition is necessary to be developed. Ultra-precision machining (UPM) is able to achieve a surface roughness of a few nanometers and form accuracy of sub micrometers. This study utilized UPM to machine a surface pattern named 'Polar microstructure'. The design principle and consideration requirements are proposed, and theoretical analysis based on mathematical relations is presented. Modeling and simulation of polar microstructure surface generation was undertaken as well as a theoretical and experimental investigation of machining parameters. The results show that the designed polar microstructure is able to achieve nanoscale measurement. A two-axis positioning system based on template matching theory was developed. The strategy was tentatively checked by the CMMs which possesses 1-nm-level resolution and the test outcomes demonstrated that the length vulnerability of the template-matching method can achieve 100-nm-level uncertainty. However, the template-matching method has a huge calculation cost, and its low robustness, orientation and scale variance also need to be resolved. To address the above issue caused by the template-matching method, an enhanced FRFP (fast and robust feature-based positioning) method was developed. The FRFP method guarantees that the measurement system is angle and scale-invariant. Another significant improvement of FRFP is its much faster calculation speed, especially when the pixel number of global images is large. The FRFP method also has strong robustness. Validation experiments were conducted and the experimental results all validated that the FRFP method is superior to the template-matching method. Hence, FRFP is chosen as the main theory for the IPMCV (integrated polar microstructure and computer vision-based) system. To address the specific application of IPMCV system, this study measured the repeatability of the machine tool using IPMCV and benchmarking with the measurement results with a laser interferometer. It was found that the IPMCV system has many advantages which greatly reduce the cost and improve the efficiency. The experimental results show that the IPMCV system enables the measurement ability to be extended to machine tools. This research significantly contributes to machining and measurement science and technology, especially in the field of computer vision-based measurement.
Description: ii, xvii, 216 pages : color illustrations
PolyU Library Call No.: [THS] LG51 .H577P ISE 2019 Zhao
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
991022214658403411_link.htmFor PolyU Users167 BHTMLView/Open
991022214658403411_pira.pdfFor All Users (Non-printable)9.08 MBAdobe PDFView/Open
Show full item record
PIRA download icon_1.1View/Download Contents

Page view(s)

Citations as of May 21, 2019


Citations as of May 21, 2019

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.