This application note describes how fibre parameters within the Morphologi software are calculated and presented, including the Fiber Total Length, Fiber Width, Fiber Elongation and Fiber Straightness.
The ability to calculate fiber parameters is included in Morphologi software version 8.10 onwards. Fiber parameters are calculated by skeletonizing the particle image. The skeletonization results in a single pixel line that represents the fiber.1,2,3,4 In the Morphologi software this is measured and reported as fiber total length. A pruning algorithm is performed to remove any superfluous branches, skeleton branches arising from small surface deviations. 4
Figure 1: Skeletonization to obtain fiber total length
The Fiber Total Length enables the calculation of Fiber Width, Fiber Elongation and Fiber Straightness as per the Application Note MRK1918.
Enabling Fiber parameters in the SOP
Fiber parameters must be switched on in the SOP before performing a measurement. Measurements cannot be reanalyzed with Fiber parameters if not selected in the original measurement. Figure 2 shows the toggle for switching on Fiber parameters. Under Analysis settings, tick the box next to 'Calculate fiber parameters'.
Figure 2: Location of fiber parameters 'switch'
By default, the skeleton will not be visible on the particle images, as shown in Figure 3a.
Figure 3a: A fiber without skeleton visible
Figure 3b: A fiber with skeleton showing
It is possible to view the skeleton, as shown in Figure 3b, for troubleshooting (or method development) by pressing the 'Show Skeleton' button in the particles tab as shown in Figure 4.
Figure 4: Location of Show Skeleton button.
Guidance when using the fiber parameters
Fiber parameters rely on having good quality images of the particles. With all Morphologi parameters, due to the binary nature of the threshold feature, image contrast and edge sharpness is important. Particles must be in good focus therefore choice of optic for the measurement is important. Since image contrast is critical for fiber parameters, measurements of fibers on filter membranes where contrast is not maximal are not recommended.
Figure 5 shows an image where part of the particle is raised up from the surface. This part of the particle is out of focus compared to the rest and is therefore poorly imaged. Poor image contrast in combination with thresholding leads to a 'fuzzy' edge which would result in extra branches on its skeleton. However, in the Morphologi software a pruning algorithm is automatically performed4 which removes these superfluous branches, or spurs. Figure 6 shows an image of the same particle where the pruning algorithm has removed the extra spurs.
Figure 5: A fiber with one end raised up from surface, the raised end is poorly imaged and has extra branches.
Figure 6: Particle image and skeleton after pruning.
The image shown in Figure 5 and Figure 6 may have been improved by choosing an optic with a larger depth of field, or using z-stacking, therefore keeping more of the particle in focus.
Fiber parameters are designed to work for fiber like particles and do not always work well for small, symmetrical none fiber-like particles. For such particles, small deviations on the perimeter can cause 'unexpected' skeletons if they are significant compared to the particle size; although these skeletons look odd they are mathematically correct. Examples are shown in Figure 7. In such cases, the caliper type length and width may be more representative than the fiber length/width.
Figure 7: Skeletons of small particles with small perimeter deviations.
When measuring fibers with loops in them as shown in Figure 8, hole filling should be disabled. This can be done in the analysis page of the SOP shown in Figure 2.
Figure 8: A looped fiber analysed with hole filling disabled.
However, as discussed earlier good image contrast is essential. Figure 9 shows a poorly imaged particle captured with hole filling disabled. The particle image has varying contrast and the thresholding is poorly set so the perimeter is not complete. The skeleton loops around the holes and the fiber total length is greatly oversized. Since the other fiber parameters are derived from the length they are also inaccurate.
Figure 9: An image with lower contrast areas/poorly thresholding showing the consequences of switching hole filling off.
On the computers supplied with Morphologi G3 orders as of 2013 no significant performance impact is expected. However, on older computers, which do not have a core i7 processor, large performance impacts have been observed. Measurements of mixed samples, containing both fibers and non-fibers, can take twice as long with fiber parameters activated.
The fiber parameters in the Morphologi software allow the measurement of fiber-like particles. When measuring these parameters, extra care should be taken to ensure good particle contrast and the appropriate settings should be applied.
- K. Saeed, M. Tabedzki, M. Rybnik, M. Adamski. 2, 2010, Int. J. Appl. Math. Comput. Sci., Vol. 20, pp. 317-335.
- M. Ahmed, R. Ward. 12, 2002, IEEE Trans. Pattern Analysis and machine Intelligence, Vol. 24, pp. 1672-1678.
- P. I. Rockett, 10, 2005, IEEE Trans. Pattern Analysis and Machine Intelligence, Vol. 27, pp. 1671-1674.
- F. Y. Shih, Image Processing and Pattern Recognition: Fundamentals and Techniques. Wiley, 2010, chapter 4.5.7-8