Abstract
M.A.
This dissertation comprises an in-depth investigation into the domain of computer vision, with specific reference to the textile industry. The study consists of three main sections.
In the first section, the computer-vision process is scrutinised in its entirety. Attention
is given in this regard to what computer vision is, where it originated from, how it compares with human vision and what the motivations are for its implementation.
Following, the computer-vision process is divided into four main components, namely
image acquisition, image processing, image analysis and image interpretation.
Subsequently, each component is discussed in greater detail, as well as aspects such as
the hardware used in the course of the process, the algorithms that are implemented
and the specific applications used for the process of analysis. In the second section, the focus is shifted to the textile industry, where our main focus lies. In this regard, examples are examined of the successful implementation of computer-vision technologies in the textile industry. In the third section, an investigation is launched into the specific problem for which a solution needs to be found in the present study, namely to determine whether
computer vision constitutes a cost-effective way in which to locate broken thread during the spinning process. A wide range of algorithms has been applied for this purpose, whereafter the results of these experiments are announced.