The design of a digital photogrammetric metrology system for the semi-automated surveying and recording of pipe dimensions in industrial plants

Master Thesis


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title

University of Cape Town

This thesis reports on the design, development and testing of a semi-automated system to aid in the mapping of the interior of industrial plants. The system makes use of digital photogrammetry to assist an operator in locating and identifying components of the plants. All of the important photogrammetric theory is discussed in the text, and explained in detail in the appendices. Specifically, this system implements various algorithms used for camera calibration, object point intersection, and a method combining the two techniques. Considerable use is made of the iterative least squares method, which is the basis of many of the algorithms employed in this work. Image processing algorithms are implemented to enhance the digital images, and to ease the identification of objects in the images, and these are fully explained in the text. Adaptive least squares image matching is a method of matching corresponding points in different images and is used to ensure correspondence between points identified by the system operator. A weighted centre of gravity method is used to find the centre of target areas, and an algorithm is implemented to determine the radius, centre and direction of a pipe passing through a number of points. Various aspects of the system design are discussed and explained. In particular the requirements in terms of hardware and software are presented. In addition, the choices of the operating system and of the compiler are justified. Potential problems with the system, and possible enhancements of it are also described. Tests were performed to verify the correct operation of all of the algorithms used in the calibration of the cameras. Together with the point intersection routines, these tests calculated the position of various control points, the correct coordinates of which were previously known. The calculated point positions are compared to the known coordinates of the points to determine the accuracy of the various algorithms. Further tests were conducted to demonstrate and verify the ability of the system to measure distance in three dimensions. These tests illustrate that the accuracy achievable is approximately 0.053 of the total distance measured for an object occupying 803 of the width of the image. The system improves considerably on the method presently used in South Africa and in many industries worldwide which rely on analytical photogrammetry for the determination of object point locations. While the system suffers from reduced accuracy as a result of the use of digital cameras, this problem will become less important as technology and digital camera resolution improve. Possible enhancements include the use of more numerically efficient algorithms, and the introduction of techniques that would partially automate the identification of control points and pipes.

Bibliography: pages 51-53.