Evaluation and Control Principle of the Crepe Effect on Fabrics*
Fabrics made up of both raw silk and other fibers can form a kind of peculiar visual texture, which is know as the crepe effect, on fabric surfaces. However, the level of crepe is difficult to control. We put forward a method using image analysis and fractal geometry to evaluate the visual effect of creping, and analyze the relationships between the bending shape of yarns and the visual effect of creped fabrics. The results indicate that it is possible to control the crepe level by adjusting the fabric breadth during the breadth drawing process.
The crepe effect on fabric surfaces is a kind of peculiar visual texture effect, which results from the shrinkage and torque of high-twist yarns in fabrics. This style originates from raw silk fabric. The crepe effect comes from the release of shrinkage energy and twist energy, which are stored in creped yarns after twisting and heat-setting. The energy is released through the release of the temporary heat-set. Taking crepe de Chine as an example, the crepe effect is formed through the processes shown in Figure 1.
To obtain the crepe effect, the belocking and release of deforming energy are two key factors. In raw silk fabrics, deforming energy coming from high-twist raw silk yarns is belocked by the sericin that covers the silk fibril. After weaving, the energy is released with the degumming of sericin. This causes the yarns in the fabric to distort and arch, which results in the rugged crepe effect on fabric surface .
Two important issues during manufacturing are how to form the crepe effect and how to control the level of crepe. Some researchers have investigated the mechanism of creping fabric. Studies have shown that yarn shrinkage strongly contributes to the fabric crinkling. Ohno and Kawamura  studied the crinkling mechanism of wool crepe fabric. They investigated the influence of yarn-twisting conditions on yarn torque, and indicated that shrinkage force and yarn torque resulted in fabric crinkling. Yamashita et al.  studied the influence of fabric structure and twisted yarn on cotton crepe fabric. The results showed that the number of ribs on the crepe fabric was related to the shrinkage; the shrinkage and the number of ribs were affected by the twist number, the density of weft, the count of weft and warp and the cloth cover factor. Wang et al.  presented the mechanism of PET crepe texturing, and designed PET crepe fabrics through two types of technique. Ishikura et al.  also studied the crinkling mechanism of crepe, assuming that the yarn is an elastic body.
Modern fabrics made up of both raw silk and other fibers can be creped. However, the level of crepe is difficult to control owing to the lack of an evaluation standard. Evaluation is a critical issue in controlling the crepe effect. The crepe effect on fabrics is generally assessed subjectively and there is no quantitative standard. Image analysis is a powerful method for gathering information and has been used widely. Fractal analysis may provide a method for the quantitative evaluation of structures that are difficult to accommodate by traditional methods [6, 7]. In the present work, we develop an image processing and fractal analysis method for quantifying the crepe effect. Furthermore, as the level of crepe is closely related to the spatial shape of yarns in the fabric, we propose a method for crepe level adjustment on the basis of analyzing the relationships between the fabric crepe effect and yarn shape.
Visual Evaluation of Crepe Effect
The essence of the crepe effect is the peculiar texture, which exhibits a concavo-convex fabric surface. If a fabric surface is unsmooth, when a beam of light with a certain angle of incidence irradiates the surface, the convex area will project shadows onto the fabric surface. At a given angle of incidence, the higher the convex is (i.e. the stronger the crepe effect), the larger the projection area. So the area and the distribution of projection can reflect the level and the character of the crepe effect. Therefore, the ratio of projection (shadow) area to total fabric area can be used to describe the level of crepe and the periodicity of projection block can be employed as an index of the delicacy level of the crepe .
Obtaining and Preprocessing the Creped Fabric Images
Eight crepe de Chine fabrics were selected. The specifications are shown in Table 1.
Eight-bit grayscale fabric images that have gray scales from 0 (black) to 255 (white) were captured by a video camera through a light microscope under a reflex light source. The system is illustrated in Figure 2. Each image corresponds to the real fabric of size 1.2 cm × 1.2 cm. To avoid the influence of uneven background, a sample image with fabric and background image without fabric were captured under the same light conditions, and a minus operation between the two images was performed. Thus an image avoiding uneven background was obtained, as shown in Figure 3.
Extracting the Eigenvalue of the Crepe Effect
Crepe Degree C
The crepe degree C is determined as
The projection area was calculated as follows: take a certain gray value (104 here) as a threshold to separate the projection area and the background area into a binary image. Pixels with gray values smaller than the threshold were set to black (projection) and the rest were set to white (background). In order to diminish the undulation caused by the interweaving between warp and weft, the morphological operations, namely erosion and dilation operations, were performed twice on the binary image. An erosion peels layers from objects (here the projections), removes extraneous pixels from the image and a dilation can return eroded objects to their original size. After preprocessing, the number of black pixels representing the projection area and the pixel number of the whole image representing the total fabric area were calculated. Fabrics with a strong crepe effect have larger C values (see Figure 3 and Table 2).
Projection Periodicity P
Most of the projection blocks lie along the warp direction and are approximately elliptical or rectangular. The projection periodicity P was employed to describe the delicacy of the crepe. The number of projection blocks appearing along the weft direction in the image was measured 10 times in different places randomly and the average value was calculated, indicating the repeating frequency of the projection blocks. The projection periodicity P is obtained according to the following equation:
Fabrics with a delicate surface have smaller P values (see Figure 3 and Table 2). However, when a fabric has very low crepe level, the P value will have a large calculation error; sample 1, whose P value was ignored here, is an example of such a fabric.
Fractal Dimension of Creped Texture D
Creped fabric has a concavo-convex, rough and irregular surface. Different fabrics have different crepe levels and different crepe distributions. These different morphologies have a direct influence on the fabric’s visual effect and other styles. The randomness and complexity are the main characters of creped fabrics. The fractal dimension is the parameter that quantitatively depicts the fractal character. The fractal characters of creped fabric surface images were analyzed by using a box-counting method. If an image is regarded as a set^l, the box-counting method is as follows. Let M × M be the size of an image. We can imagine grids of size S × S (S
For creped fabrics, a sample image was covered by boxes with e being 1/2,1/4,1/8,1/16,1/32,1/64 and 1/128, in turn. When a box contained at least one pixel whose gray value was smaller than or equal to a given threshold, it was counted. The number of counted boxes was N(ε). Here, the gray value 104 was taken as the threshold. The plots of In N(ε) versus ln(l/ε) are shown in Figure 4. Four points as e = 1/16, 1/32, 1/64 and 1/128 were in a linear section and were fitted by the least-squares method. The fractal dimension of the creped texture D shown in Table 3 was determined by the slope of the least-squares linear regression equation. Here R is the linear correlation coefficient.
It can be seen from Figure 4 that almost all of the data points lie along the straight lines in the ln-ln diagram. As R^sup 2^ is larger than 0.995 it is an indication that In N(ε) is well linearly correlated to ln(l/ε), which means that the creped texture of the fabric has a significant fractal character. The fabrics can be classified according to the value of D, i.e. samples 2 and 3 belong to the same category having a smooth surface as seen in Figure 3; samples 6 and 8 belong to the category with an undulated surface and coarse texture; and samples 4, 5 and 7 belong to another category with a distinct and exquisite texture. Sample 1 is not very creped and its fractal dimension, as well as its surface style, obviously differs from that of other samples. Referring to Table 1, it can be seen that the creped texture is much correlated to the fabric specification, especially to the weft twist. Excluding sample 1, the linear correlation coefficient between D and weft twist shown in Table 1 is -0.92882. This indicates that D may be a bridge between the visual style of the fabric product and the producing technique.
Extracting the Eigenvalue of the Creped Yarn Shape
According to the mechanism of forming the crepe effect, it is known that the strongly twisted weft yarn becomes an irregular spatial spiral owing to the release of energy. The undulated surface is a resulted of the spatial spirality. So the eigenvalues of yarn shape were extracted. Creped weft yarn images were captured under a transmission light source and thresholded to binary and skeletal images as shown in Figure 5. The real size of the fabric for each image was 0.38 cm × 0.38 cm.
Ten curves in each skeletal image were randomly selected to extract the mean peak valued and mean periodicity T of the curves according to the peak points. The peak points in each curve were determined by using the method illustrated in Figure 6, where mean is the average of the absolute amplitude of every point in the curve, max is the amplitude of the highest point and min is that of the lowest point. The black dots are determined peak points. At most only one extremum value was calculated in the period of every two white dots. The mean of absolute amplitude of peak values in the ith curve is denoted by A^sub t^ and T^sub t^ is the ratio of the width of image to the number of peak points . The mean peak valued A and mean periodicity T were calculated as follows:
The results are shown in Table 4.
Relationships between the Creped Yarn Shape and Crepe Effect and the Principle of Controlling the Crepe Effect
In a creped fabric, spatial spiralities of creped yarns are the foundation of forming the crepe effect. The results of a correlation analysis between the eigenvalue of the crepe effect and the creped yarn shape are shown in Table 5.
It can be seen from Table 5 that the crepe degree C is highly correlated with the mean peak value A and the projection periodicity P is highly correlated with the mean periodicity T. This indicates that the level of undulation and delicacy are highly dependent on the shape of the creped yarns. The more flexural the yarns are, the more undulant the fabric surface will be, which results in a strong crepe effect. The large periodicity of the yarn curve causes large projection periodicity on the fabric surface, which indicates a coarse fabric texture. The fractal dimension of the creped texture D is also correlated with the projection periodicity P and crepe degree C. This shows that D has some relationships with the fabric surface style. This is a point that deserves further investigation.
In fact, the routine technology of forming the crepe effect is to configure more energy to the creped yarns than required. After degumming, a much stronger crepe effect than expected is formed on the fabric surface. The crepe effect is then weakened to an expected level by drawing the fabric breadth and heat-setting.
If we model a creped weft yarn as an ideal three-dimensional spiral line, consider A as an approximation of the spiral radius and T as an approximation of the distance between two spirals (see Figure 7), then the length of one spiral (l) can be calculated as
Assume that L is the length of creped weft yarn, W is the breadth of fabric and n is the wave number of the spiral along the weft direction, then
Ignoring the tensile deformation of the creped weft yarn during the breadth drawing process, L is unchanged. From equations (1) and (2), we know that there is a certain relationship between A and W, as well as between T and W. This means that if the fabric breadth is adjusted, A and T will change, which results in a change in the crepe effect on the fabric. This indicates that, to some extent, the crepe effect is can possibly be adjusted during the breadth drawing process.
By using an image analysis technique and fractal theory, the eigenvalues of creped fabric surface morphology have been extracted. The crepe degree C is a measure of the level of the crepe effect. The projection periodicity P is a measure of the delicacy level of the creped fabric texture. The fractal dimension of the creped texture D is correlated with the fabric surface style and fabrics with different textures can be classified according to the value of D.
The eigenvalues of the creped yarn shape, i.e. the mean peak value A and mean periodicity T, have been extracted from creped weft yarn images captured under a transmission light source. The crepe degree C is highly correlated with the mean peak valued and the projection periodicity P is highly correlated with the mean periodicity T.
The visual effect of creped fabric results from the fabric surface morphology, and the fabric surface morphology depends on the spatial shape of the creped yarns in the fabric, while the creped yarn shape is closely related to the fabric breadth in certain fabrics. Therefore, it is possible to control the level of the crepe effect by adjusting the fabric breadth during the breadth drawing process.
Further work may be necessary to understand the relationship between D and other parameters.
* This paper was presented at the 17th IMACS World Congress.
1. Li, D, Analysis of Crepe Effect of Crepe de Chine, Journal of the Suzhou Institute of Silk Textile Technology, 14(4), 87-96 (1994).
2. Ohno, H., and Kawamura, H., Effects of Yarn Twist on the Surface Appearance of Worsted Fabric Part 1: Crinkling Mechanism of Wool Crepe Fabric and Yarn Twisting Conditions, Journal of the Textile Machinery Society of Japan, 50(9), 251-257 (1997).
3. Yamashita, S., Takatera, M., and Shinohara, A., Influence of Fabric Structure and Twisted Yarn on Cotton Crepe Design Part 2: Effects of Weaving Condition on Shrinkage, Rib Shapes and Number of Ribs, Journal of the Textile Machinery Society of Japan, 50(6), 77-85 (1997).
4. Wang, S., Gao, Y., Zhang, Z., and Zeng, X., J. China Textil. Univ., 12(1), 43-48 (1995).
5. Ishikura, H., Kase, S., and Nakajima, M., Journal of the Textile Machinery Society of Japan, 44(12), 62-69 (1991).
6. Pearce, N. R. L., Guild, F. J., and Summerscales, J., The Use of Automated Image Analysis for the Investigation of Fabric Architecture on the Processing and Properties of Fibre-reinforced Composites Produced by RTM, Compos. Appl. Sci. Manuf, 29(7), 829-837 (1998).
7. Conci, A., and Proença, C. B., A Fractal Image Analysis System for Fabric Inspection Based on a Box-counting Method, Computer Networks and ISDN Systems, 30(20-21), 1887-1895 (1998).
8. Wang, Z., and Li, D., Evaluation of Crepe de Chine with Computer Vision Technique, Journal of Suzhou Institute of Silk Textile Technology, 14(4), 110-117 (1994).
9. Bisoi, A. K., and Mishra, J., On Calculation of Fractal Dimension of Images, Pattern Recogn. Lett., 22, 631-637 (2001).
10. Ying, H., “Analysis of Wave Shape and Spectrum and Processing of Random Data”, China Railway Publisher, Beijing, 1985.
Xu-Hong Yang1 and Dong-Gao Li
College of Material Engineering, Suzhou University,
No. 178, East Ganjiang Road, Suzhou City, Suzhou,
Jiangsu Province 215021, People s Republic of China
1 Corresponding author: e-mail: firstname.lastname@example.org
Copyright Textile Research Institute Oct 2007
Provided by ProQuest Information and Learning Company. All rights Reserved