The automatic sizing of fibres

Natural fibres have been used in textile industries for a hundred years. In recent times it has become very important in many industrial applications. Amongst these are textiles, cement, cellulose pulp, paper and fibre reinforced materials. An important factor in the industrial use of fibres is the relative size of individual fibre diameters. For example, knowing the fibre diameters enables a marked improvement in the texture of the textile or paper product. This requires a fibre sizing scheme. On the other hand, it is recognized that fibres are also a health hazard. Whenever fibre material is used or processed, a fine fibrous dust is released into the atmosphere. Exposure to this dust can cause serious lung disease. Controlling this exposure requires a fibre sizing scheme. It follows that fibre sizing is of great industrial and toxicological importance. Manual sizing methods have proved to be tedious, unreliable and time-consuming, while the automation of this task offers significant advantages. These advantages are reliability, speed and reduced cost. An automatic fibre sizing scheme is proposed in this thesis. The hardware of the scheme is based on the use of a picture processing machine. This comprises a video input device (Vidicon camera/ Scanning Electron Microscope), a frame store digitiser and a computer. The video signal is routed via the frame store digitiser to the computer. Algorithms are developed to handle the sizing task automatically. The algorithms developed are thresholding, skeletonising, tracing and measuring algorithms. A statistical prediction procedure is also developed. This is mainly to predict the dimensions of the fibres whose lengths exceed the diameter of the input micrograph view of the electron microscope. The accuracy and limitation of this scheme is investigated.