Fiber Wall Thickness Measurement by Machine VisionJuhani Hirvonen, VTT Automation
Automatic measurement system based on image analysis was constructed to characterize pulp quality by fiber wall thickness. It provides an accurate system for controlling pulp quality in the manufacturing of high-quality printing papers
In the refining, pulp fibers are applied mechanical treatment so that their physical, and also chemical, structure is changed, which leads to a significant improvement in their paper-making potential. In the refining, fiber is less or more delaminated and fibrillated, depending on the energy applied and refining conditions. Delamination and fibrillation are essential phenomena for fibers, affecting their paper-making properties.
Especially in the case of mechanical pulps, the control of fiber wall thickness is and the fibrillation are favorable for producing high-quality printing papers. In addition, it has been shown that the quality of the fiber fraction reflects the quality of the whole pulp and further of the end-product. Thus, measurement of fiber wall thickness could also be used for controlling the whole pulp quality and paper-making process.
Better pulp quality control also leads to a more stable paper-making process with a decreased number of web breaks on the paper machine. This automatically means a decrease in the energy consumption, savings in the raw materials and a decrease in the effluent discharge from the paper mill. Thus, the better fiber quality control has multiplied effects on several factors in the society in terms of savings in the natural resources and of environmental load.
The fiber wall thickness measurement can be applied for chemical, mechanical and recycling fibers.
Microscopy and image analysis have been applied on measuring pulp quality in research laboratories. Mostly the studies have concentrated on measuring some specific feature and efforts have been made to correlate these measurement with some process data or paper quality. The results have not been very convincing which due to the small amount of data obtainable by ordinary microscopy. A step towards to on-line measurements was taken by VTT Automation, Laboratory of Paper Technology at Helsinki University of Technology, Valmet Automation Kajaani Ltd., and UPM-Kymmene in years 1995-1996, when a novel image analysis equipment and method for characterizing mechanical pulp fines were developed. The systems can, in principle, be installed on-line and produce fines quality measures every ten minutes.
Today there are no feasible methods for on-line measurement of fiber wall thickness available. Some laboratory scale measurement devices have been developed for measuring, e.g., fiber length, which is not enough for characterizing fiber quality. Furthermore, some methods are based on manual fiber imaging, which is laborious and do not give a sufficient number of images for statistical consideration in a reasonable time. Thus, it is obvious that an automatic imaging device should be developed. In addition, the image treatment and calculation must be fast enough so that the apparatus can offer real-time information (i.e. 5-15 min). An automatic device for measuring fiber wall thickness would be a valuable tool for controlling the whole paper-making process, including pulp treatment, energy cost and end product quality.
Image acquisition and processing.
A sample of the pulp is diluted into water with some detergent to avoid flocking. The diluted sample flows trough a cyvette of two plates of class. The distance between the plates is 0.5 mm. The sample is illuminated by a flash through the cyvette The captured images are compared against the estimated background image and the images containing fibers are stored for later analysis.
The image processing sequence is the following: First the uneven illumination is corrected, secondly fiber is extracted and the images containing air bubbles and excessive fibrils are rejected. The following step is the formation of the skeleton of the fiber and determination of the area of interest shown in Figure 1. The found area of interest is used in the final wall thickness measurement. Typical steps of image processing is shown in Figure 1.
Figure 1: From top to bottom: found area of interest and measured points, and a partial zoom of previous
The fibre wall thickness
measurements obtained by machine vision for five samples are compared to
manual laboratory measurements in the following figure 3. It shows that
there is quite a good correlation with automatic and manual measurements.
It can also be noticed that the scale in wall thickness seen by machine
vision is smaller. There are at least two possible reasons for this phenomena.
First the definition of wall thickness is different, manually the thickness
is measured at one point whereas machine vision measures the fibre at up
to more than hundred points. Careful inspection of the analysed images
reveals that the wall thickness measurements from obtained images appears
to be statistically correct.
In the following figure 4
fibre wall thickness distributions obtained by machine vision are compared
to manual measurements. There seems to quite good resemblance with the
distributions, although machine vision seems to favour medium wall thickness.
This phenomena is most likely due to the different definitions of wall
Today, paper-making process is largely automated, but the pulp production and especially its quality control are still lacking of good on-line measurement equipment. Each step forward that leads to a better fiber quality control, leads also not only to the better pulp and paper quality, but also more efficient paper-making economy, and paper-making process with a decreased number of web breaks on the paper machine. This automatically means a decrease in the energy consumption, savings in the raw materials and a decrease in the effluent discharge from the paper mill.