Textile fabrics abrasion resistance – The instrumental method for end point assessment

2.1 Materials

The fabrics selected for this investigation were dedicated to garments, especially workwear, protective clothing, and uniforms, because of the important role played by abrasion resistance in user protection. Moreover, the selection relied on crucial woven fabric features that are essential when using image analysis tools, such as colour, texture, type of fabric surface, fabric density, and material composition. The compatibility between the elaborated end point criteria and the description in standard [19] is evident. In Table 1, the characteristics of woven fabrics were showcased, including the woven model structures subjected to the abrasion tests. Among them are raw materials’ composition (MC), mass per unit area (MA), threads density (TD) marked as wp – warp density and wf – weft density. In case of fabric 1, the antistatic thread positions are marked in yellow and black in the model.

Table 1

Fabric characteristics

No.	Fabric construction	Colour	MC	MA, g/m2	TD, 1/dm
1			AR 98%/antistatic fibre 2%	176	260 (wp)
					165 (wf)
2			AR 100%	170	260 (wp)
					165 (wf)
3			PES 48%/CO52%	118	450 (wp)
					260 (wf)
4			PES 48%/CO52%	128	330 (wp)
					260 (wf)
5			AR 100%	207	275 (wp)
					185 (wf)
6			PES50%/CO50%	251	255 (wp)
					225 (wf)

2.2 Methods

The main idea of abrasion resistance evaluation of textile fabrics using Martindale’s device shown in Figure 1 is presented in diagram (Figure 2). The samples, mounted on device holders, are subjected to abrasion cycles. The diameter of the textile samples fixed in the holder was 31.75 mm. After a certain number of cycles were finished, the visual assessment of the sample was conducted to detect fabric breakage defied in standard [19], being the two separate threads broken for woven fabric. If the end point of the test was not reached, the evaluation continued. Test ended when two broken threads finally occurred. The number of cycles previous to last was noted as a test result.

Figure 1

Holder with sample inside and table of Martindale device.

Figure 2

The conception of abrasion resistance test procedure.

The goal of the instrumental method abrasion resistance assessment was to supersede the purely organoleptic procedure of thread breakage identification by the instrumental evaluation based on sample image processing and recognition of broken threads. The instrumental method consists of several steps. First, the samples are mounted on holders and subjected to the abrasion cycles according to the standard [19]. This step is the same as in purely standard evaluation. In the next step, all tested samples, still placed on holders, are scanned using a flatbed scanner at one operation. Next the images of samples are captured. The image profiles are inspected for each vertical and horizontal line 20 pixels wide. The selected changes in the profile shape or amplitude covering the long enough part of profile are associated with places where the thread was missing in the fabrics structure. When one of these changes was found during the screening of the profiles, the broken thread is noticed. After at least two places meeting these criteria were found, the sample reaches the end point of the test. Figure 3 illustrates the idea of instrumental method in contrast with standard visual one. Primarily, the process of abrasion evaluation for both methods is the same until the assessment procedure came. The sample still mounted in the holder was scanned instead of being carefully examined by the operator under magnification. Then, the set of image profiles were checked to find the changes corresponding to established breakage criteria. Similar as in the standard, evaluation test ended, when breakage criteria were fulfilled.

Figure 3

An overview of the comparison between various methods for assessing abrasion resistance.

Determination of woven fabric breakage criteria in the image analysis domain was performed according to the following plan:

1. abrasion resistance test was performed according to conditions described in the International Organization for Standardization [19], load 12 kPa, normal climate atmosphere,

2. samples mounted in the holders were scanned using flatbed scanner Canon Canoscan 8800F, resolution 4,800 × 9,600 pixels with 48-bit colour depth after each abrasion cycle,

3. the brightness level profiles were plotted for image lines, with 20 mm (0.84 mm) each, a summary of the 37 profiles for both sample direction was obtained. Profiles were analysed to identify changes or disturbances caused by textile surface changes. Profile analysis was accomplished with the application of Fiji 1.53 [31],

4. critical analysis of changes in recorded profile had competed to find variations selective to the broken threads in fabric structure solely.

The advantage of an instrumental method for abrasion resistance assessment is the shorter evaluation time compared to the standard visual assessment, where approximately 15 min is needed to investigate, under 8× magnification, the fabric structure in three samples to find potentially broken threads. In contrary, the time demand for scanning at least these three samples, capturing images, and analysing profiles to detect broken thread signs is estimated at approximately 3 min, thanks to the real-time assessment of profiles. In both cases, assessment was performed after a certain number of cycles, which could be 10–15 in one test, resulting in a significant reduction in time and labour, reaching up to 180 min. The longer the abrasion test, the better the resistance, and the higher the time efficiency for using the instrumental method. Furthermore, the instrumental method supports the test operator in the process of thread breakage identification, which is hard to recognize because of simultaneously existing other surface changes, such as fuzzing, pilling, and protruding fibres.

3.1 Results

Table 2 shows the test results for all four elementary samples (marked as s1 to s4), resented together with a description of changes existing on the woven fabrics surface but not directly related to the end point of the abrasion resistance test. Among the variety of such changes are pilling, fuzzing, or colour fading. The cycle variations that appeared in a specific instance were depicted as a range of rubs. The breakage of the first thread was recorded as well.

In the next step of work, the brightness level profiles were analysed for identification of changes, which correspond to the broken thread in the woven structure. The breakage of thread on the woven fabric structure shaped the profile. Some parts of the profile show different features than the rest of the profile, e.g. significant increase or decline in brightness level in part of profile longer than cause by level changes typical for fabric structure periodically. Figures 4–9 illustrate the various changes of profiles, with one being captured just prior to thread damage, while the subsequent profile includes a broken thread within its structure. It allows comparing the intensity of profile disturbances caused by reaching the thread breakage. The red arrows indicated part or parts of the profile representing the broken thread in the figures. The dotted black line figured the mean value of the brightness level of the profile.

Figure 4

The end point of the abrasion test for fabric 1.

Figure 4 shows profiles of the image sample, which reaches the end point of the abrasion test, meaning two threads were broken. Discontinuities of threads observed in the fabric structure were found in profile parts between 75 and 105 pixels and from 470 to 500 pixels. The above-mentioned profile sections exhibited a decrease in mean brightness values of 20 and 26 gray levels.

Figure 5 presents profiles recorded for the sample images of fabric no. 2, with changes matching the loss of one thread in the fabric structure. The broken thread-related profile length was 50 pixels, from 570 to 630, as was marked by the red arrow. The mean value of brightness for this segment was 19, in contrast to the 36 levels calculated for the whole profile length. The mean section with the damaged thread is 46% darker (lower brightness level) than the rest of the profile. Before the breakage occurred, the above-mentioned difference between profile areas was only 25%. This differences were caused by surface deteriorations when samples got more abraded and flattened, single threads thinner up to the breach.

Figure 5

Set of profiles taken for sample image of fabric no 2.

The next example of thread breakage identification in woven fabric structure was a part in profile, not shorter than 35 pixels, showing a significantly higher brightness level than the rest of the profile line that was visible in Figure 6. The brightness profile includes a sample area where the thread is interrupted, located at the beginning of the profile, between 12 and 47 pixels, with the mean brightness level of 202 to the mean gray level value of the rest of the profile of 186.

Figure 6

Brightness level profiles collected for sample image of fabric 3.

The profiles presented in Figure 7 show changes in the characteristics of gray level oscillations between 221 and 270 pixels caused by thread damage. The difference between the mean gray levels of the mentioned segment and the rest of the profile is 17. A similar area of image sample recorded one range of cycles before thread breakage exhibited the lower gray level difference of ten levels. This means that the described profile area covered by thread, which was partially disturbed and being further subjected to the abrasion finally broke, showed loss of particular fibres, which were broken, ultimately causing thread thinning.

Figure 7

Brightness level profiles of sample image of fabric 4.

The brightness level profile of 37 pixels in length, situated between 474 and 511 pixels and reflected by the existence of damaged thread in fabric is presented in Figure 8. The above-mentioned profile’s segment was lighter than the rest of the profile, up to 32 gray levels. Ranking with profiles recorded for a range of cycles before breakage occurred, this profile part was diverse from earlier collected profiles of 3 levels only. Despite the fact that the fabric 5 achieved high abrasion resistance, reaching 100,000 cycles, thread breakage befell quite suddenly. The profiles recorded previous to the thread damage did not exhibit any traces of fibre destruction and mass losses.

Figure 8

Profiles of a sample image of fabric 5 with one thread broken.

The set of profiles, presented in Figure 9, includes thread breakage located between 19 and 77 pixels. The mean brightness level calculated for this profile’s part was 103, whereas the mean level for the rest of the profile rose to 122. The decline in brightness level reflected by thread breakage was caused by the sample in this part being getting darker. With regards to the profiles collected for the preceding test interval, the sample showed a similar drop in the brightness level, but less intense, of ten levels. This means that gradually progressive damage was observed to fabric 6 sample during the abrasion process, so the thread breach had not occurred instantly. The fabric 6 threads tended to deteriorate because of being thinner constantly after abrasion progressed.

Figure 9

Image profiles collected for fabric 6 abraded sample.

3.2 Discussion

As a summary, the presented investigation focused on finding the end point in abrasion tests using image analysis techniques and found that brightness level profile changes were the most suitable for precise and also exclusive determination of thread breakage in fabric samples, in test conditions. The goal of the investigation was to find an image analysis supporting method for assessment of abrasion resistance being consistent with the International organization of Standardization [19] in the scope of the end point criteria definition. Based on the survey presented, the two criteria for the end point of the abrasion resistance test of woven fabrics were identified:

1. The part of the profile, not shorter than 50 pixels, revealed a 30% change in mean brightness level in relation to the rest of the profile, as presented by relationships (1)–(5).

   (1) J = ∑ n = 1 m j n m ,

   (2) K = ∑ n = 1 k j n k ,

   (3) M = ∑ n = 1 m − k j n m − k ,

   (4) J = K + M ,

   (5) K = 1 , 3 M if K ≥ M 0 , 7 M if K ≤ M k ≥ 50 ,

   where

   J is the mean brightness level for a given profile line, K is the mean brightness level for profile’s part, which for the thread breakage point criteria, was achieved, M is the mean brightness level for profile’s part(s) not reaching the threads breakage point criteria

   n is the number of pixels for the sample’s image, m is the total number of pixels for the image’s gray level profile, k is the selection of pixels, for which the breakage point criteria were achieved.

   j _n is the brightness level of single n-pixel, presented in numbers between 0 and 256.

2. The profile section is not less than 30 pixels, which means brightness values differ from at least 15 gray levels to the rate of profile, as presented by relationships (6).

The values of brightness level were established as outer boundary lines for abrasion resistance test criteria, the difference of 15 levels or 30% compared to the rest of the profile came from observation images of samples and parallelly conducted evaluation of samples of the surface in scope of other features such as the pilling, fuzzing, and colour fading, which could influence the shape of the profile. Then, the decision was made about which profile features were caused by thread break exclusively to be a criterion in the assessment procedure, to distinguish them from other phenomena. The way that criteria were presented was connected with woven fabric colour. If criteria based on percentage deviation were used, the variety of gray levels for breakage point determination is raised with brighter colours (brightness level of high values). The second criterion, when a level divergence of 15 was set as breakage criteria, did not get influenced by sample colour, but was found as appropriate for shorter segments of profile. Mainly, the first criteria reflected the characteristics of changes in bright fabrics, when loss of thread continuity caused an intense drop of brightness level, because the sample background or foam mounted in the holder was exposed. For darker coloured woven fabrics, the decline in brightness level seemed to be stronger, the sample background differed less from the outer surface of fabric or foam. The length of profile segment in the identification of breakage criteria was also important. The parts with brightness levels changes shorter than 30 pixels (1.26 mm when a 600 point resolution scanner was used) were not considered as fulfilling breakage criteria, mainly being manifestations of pills or groups of entanglement fibres fixed on a fabric surface.

Both the thread breakage criteria described above were based on analysis of brightness level profiles of the sample image. This type of image analysis is widely known in the determination of surface roughness. The roughness test idea is very similar in its principle to the new method for textile abrasion assessment being the subject of this manuscript. The interrelationship between surface roughness and abrasion resistance shows that a mutual connection between these two methods exists. It is possible to assess abrasion-related damage on fabric surfaces with the help of brightness level profiles. Despite the differences between roughness terminated using a laser profilometer and abrasion resistance, the correlation between both surface parameters was proved in previous publication [28], giving a basis for further investigations in this area. Other research work focused on the kinds of abrasion resistance of textile fabrics [15,16,26] were restricted to dedicated kinds of fabrics, such as coated ones or light-weight nonwoven made of polypropylene fibres. In contrast to the presented methods, the instrumental method provided reliable results for different types of woven fabrics.