FORM 2

THE PATENTS ACT, 1970
(39 of 1970)
&
THE PATENTS RULES, 2003

PROVISIONAL SPECIFICATION
(See section 10, rule 13)

“A METHOD OF EVALUATION AND GRADING OF TEXTILE OR FABRIC OR GARMENT APPEARANCE”

INDIAN INSTITUTE OF TECHNOLOGY, Delhi, Hauz Khas, New Delhi 110 016.

The following specification particularly describes the invention.
A METHOD OF EVALUATION AND GRADING OF TEXTILE OR FABRIC OR GARMENT APPEARANCE

FIELD OF INVENTION

The present invention relates to a method of evaluation and grading of textile or fabric or garment appearance. More particularly, the present invention relates to an objective method of quantifying seam pucker in stitched fabrics using phase shifting optical fringe profilometry using two dimensional 2D measurements. More particularly, in 2D analysis, a structured light in the form fringes is projected over fabric in dark room and image of the fabric is obtained by a camera positioned at an angle. The structure of light is extracted using digital image processing. The deviations of the line profiles from straight line are determined using various attributes and are correlated with AATCC subjective pucker grades. These attributes can be used to quantify pucker/appearance of textile materials.

BACKGROUND AND PRIOR ART OF THE INVENTION

The appearance of textile material/fabrics/garments includes pilling, wrinkles/creases and pucker in and around stitched lines. Of these the wrinkles and puckers are distortions of the plane of the otherwise smooth fabrics or stitched/sewn garments. The visual appeal of the garment is a principal factor deciding its value. Seam pucker, which is a wrinkled appearance along the seam, influences the appearance of garments/stitched fabrics to a considerable degree. Seam pucker appears when the material properties and sewing parameters are not suitably selected. The problem of seam pucker is receiving much attention in recent years with the development of micro fibre fabrics which are inherently prone to deformations during garment manufacture. Every garment manufacturer wants to eliminate this problem and it can only be accomplished if the evaluation is made and its causes being found.
Considering the importance of the problem, various associations/organizations concerned with the textile industry and trade proposed different standards, the most commonly used is the one proposed by AATCC (American Association of Textile Chemists and Colorists). According to this standard, seam puckers are divided into five discrete levels and the fabric is given a grade of the reference specimen which matches most nearly to the fabric specimen. The subjective evaluation suffers from limitations such as variability between experts’ judgment, high evaluation time, biased-ness toward a particular colour etc. There exists a need for objective evaluation of seam pucker. Also the five grades do not have equal intervals since it is not based on any quantitative measurement.
Many methods have been tried and reported that are based on non-imaging and imaging techniques to characterize wrinkles or pucker or undulations or distortion, in other words the appearance of textile materials. Yet no commercial system is available today quantify the appearance of textile materials.

SUMMARY OF THE INVENTION
The present invention relates to 2D IMAGE ANALYSIS, an objective method of quantifying seam pucker in stitched fabrics using phase shifting optical fringe profilometry. The fabric is illuminated by the vertical grid lines using a laser light source and grating in a dark room environment. A CCD camera captures these vertical grid lines at an angle of 45º to the line connecting light source and camera. The vertical grid lines in the puckered fabric image will be deformed and look like curved lines. The extent of this deformation of the vertical lines into curved lines is directly related to the severity of seam pucker. The captured image is digitally processed. The deviations of the line profiles from straight line were determined. The attributes ‘Mean deviation’ and ‘Variance’ of profile lines correlate reasonably with subjective grades.

OBJECTIVE OF THE INVENTION

The primary objective of the present invention is to provide a method of evaluation and grading of textile or fabric or garment appearance.

Another objective of the present invention is to provide an objective method of quantifying seam pucker in stitched fabrics using phase shifting optical fringe profilometry using two dimensional 2D measurements

BRIEF DESCRIPTION OF THE FIGURES

Figure 1 shows fringe generating system.
Figure 2 illustrates the image of sample PA1 illuminated by vertical grid lines captured by the CCD camera at aforementioned conditions. The image reveals the non-uniformity of illumination. This is due to inherent constraints associated with poor quality of illumination.
Figure 3 shows the Binary image of sample PA1
Figure 4 Zoomed binary image of sample PA1 indicating the noise pixels
Figure 5 Clean binary image of sample PA1
Figure 6 Zoomed Binary image of sample PA1- after noise removal
Figure 7 Line detected image of sample PA1
Figure 8 Final image of sample PA1 after line linking
Figure 9 Final line image of Reference plane
Figure 10 Scatter plots of mean pucker grades and mean deviation of profile lines
Figure 11 Scatter plots of mean pucker grades and variance of profile lines.
Figure 12 scatter plots of mean pucker grade and mean deviation.
Figure 13 Scatter plots of mean pucker grade and variance.
Figure 14 Scatter plots of mean pucker grade and surface are ratio.
Figure 15 Scatter plots of mean pucker grade and power spectral density.
Figure 16 Normal probability plots of residuals
Figure 17 Scatter plots of subjective and objective pucker grades.
BRIEF DESCRIPTION OF TABLES
Table 1: Seam pucker grading by subjective evaluation
Table 2: Subjective and objective grades of fabrics

DETAIL DESCRIPTION OF INVENTION

Accordingly, the present invention relates to a method of evaluation and grading of textile or fabric or garment appearance. More particularly, the present invention relates to the 2D technique based on line deviation. Puckered or wrinkled fabrics have distorted planes in the form of projections and valleys that are called waves. The severity of these distortions is exemplified by both amplitudes of the waves (heights) and the frequency with which the waves appear on the fabric plane. Human perception of the appearance is essentially about the severity of the distortions, i.e. higher the amplitude of the wave and presence of too many waves, worse the human perception of the appearance. The line deviation attributes of distorted/puckered/wrinkled fabrics (irregular or stitched fabrics) from the reference plane (smooth) depends both on the amplitude and frequency of the waves present on the fabric. Colour, shade, pattern and texture of the fabrics influence intensity variations of the captured images of fabrics and hence the accuracy of the features of the images during digital image processing. However in the described method, only the patterns of lines are to be extracted and hence, the influences of color, shade, pattern and textures of the fabrics on the outcome of the results (errors) are masked to a greater extent. Previous methods based on measuring the intensity variations or areas of the shadows present on the captured images of fabrics are less accurate as they greatly suffer by the colors, shades, patterns and textures of the fabrics.

Accordingly, the present invention relates to a 2D method based on projection of structured light over the fabrics of appropriate sizes (stitched and unstitched ones) and capturing the images of the fabrics at an slant angle not necessarily at 45? (as we have demonstrated only at 45?). A method to fix the fabric on the fabric holder with the stitch line perpendicular to horizontal table. A fringe generating system comprising of expansion of laser beam from the laser source using a beam expander and spatial filter unit and collimation using a collimating lens; making the collimated light incident on a single optical element interferometer to make two sinusoidal wave fronts that interfere to produce fringes; and a method to change the number of fringes per mm. A method of projecting the structured light using projecting lens over reference plane and fabrics and capturing the grey images of the structured light by a CCD camera using imaging lens. Methods of processing the images of the fabrics for noise removal, technique to link the missing lines (which might be due to defects of uniform illumination, non-optimum slant angle of the camera, limitation due to the resolution of the CCD camera employed exist with current demonstration system) to extract features of lines. Evaluation of seam pucker/levels of undulation on fabrics/irregularity of fabric surface/appearance of fabrics/textile materials through various attributes/ characteristics of lines such as mean deviation, variance etc.

Figure 1 shows fringe generating system, wherein the top view of experimental setup used for this technique is shown in Fig.1. The fabric is fixed on the fabric holder with the stitch line perpendicular to horizontal table. A structured illumination using a laser source illuminates the fabric. The laser beam (2mm diameter) is first expanded using a beam expander and spatial filter unit. The expanded light is then collimated using a collimating lens (focal length 20cm and diameter 3cm). The collimated light is made incident on a shear plate interferometer (diameter 10cm and thickness 2cm and surface quality ?/4). This interferometer generates two wave fronts, one reflected from the front surface and another reflected from the bottom surface. Both the wave fronts interfere and produce fringes. This is a single optical element interferometer and hence is a common path and highly stable interferometer. Because of these advantages the interferometer can be used in robust environment. Further, spatial carrier frequency (number of fringes per mm) can be easily varied by means of changing the position of the collimating lens in the axial direction. Unlike the grating projection systems previously used in fabric characterization, the present interferometer system generates pure sinusoidal structured light which is essential for the measurement of the phase accurately. The structured light pattern is obtained projected on the fabric using a projection lens. A CCD camera kept at 45º to the line connecting light source captures the gray image of the structured light. Two images are captured, one using a reference plane and other projected on the fabric. The both reference plane and fabric are imaged onto the CCD camera using an imaging lens. Structured light pattern remains unaffected when projected on the reference plane while it is deviated from the straightness on projecting on the fabric. The degree of straightness varies according to the surface quality of the fabric. Software is developed to analyze the pattern of lines.

The next step in image processing is conversion of gray scale image to binary image using adaptive thresh-holding technique that was necessitated by the non-uniform illumination over the fabric sample. An image was segmented into a number of regions. Each region is considered as a separate image and thresh-holding is applied to that regional-image separately. Finally all the images are recombined to make original image. This results in a binary image as shown in Fig 3.
The binary image obtained by pre-processing contains some noise pixels. Noise pixel may be defined as an isolated white pixel in black background or black pixel in white back ground. Figure 4 reveals the presence of noise pixels in binary image. These pixels have to be made to blend with background, so that they do not cause problems for line extraction. A suitable computer program was written to perform this operation. The result of this operation can be seen in cleaned binary image (Figures 5 & 6).
The clear binary image obtained can be subjected to the process of line detection. The process of line detection is basically an edge detection process by which the transition points from black region to white region and vice-versa are enhanced. The code for line detection was written and image was subjected to this line detection process. The result of this operation is revealed by Fig. 7

The image reveals the breaks in detected lines, which are mainly due to the poor quality of illumination of the object by the light and can be minimized to a great extent by having uniform illumination over the object. These breaks have to be filled with white pixels to maintain the continuity of lines. A suitable code was written to perform this operation and the result of such an operation. Figure 8 shows the final image suitable for further analysis.
The aforementioned procedure was applied to image of reference plane. The reference plane refers to the fabric holder without any fabric on it. We used an aluminum plate for holding the fabric. The result of such an operation yielded the final image shown on Fig 9. The line detected from the reference image is parallel with very less deviations. The deviations can be attributed to non-uniform illumination, scope of image processing etc. Nevertheless the image contains sufficient information and accuracy and can be used for intended analysis. The profiles of the lines in the images captured from fabric samples are much different from the reference image. In general, the variability of the profile lines extracted from fabric images is higher than the reference image.
It can be seen that the lines extracted from the images captured from fabric samples deviates from straight line and assume curved shape. So the variability of these lines (using mean deviation and variance) can be used as attributes for objective evaluation. A code was written to perform this operation on various samples and reference image. These measures were correlated with mean pucker grade obtained using subjective evaluation by AATCC. Figure 10 illustrates scatter plot of mean pucker grade obtained from subjective evaluation and corresponding mean deviation of profile lines. When the mean pucker grade increases, the line profiles of fabric approaches becomes less distorted and approaches that of straight line. Consequently, the mean deviation of profile lines decreases while the mean pucker grade increases.
A simple linear regression analysis was performed to fit a regression model of the form and this yielded the equation MG = 4.8884*e-0.0038*MD and the coefficient of determination R2 = 0.647. The adequacy of the model is augmented by a reasonable coefficient of determination of 0.647. This regression equation can be used for objective evaluation of seam pucker. Figure 11 shows the scatter plot of Mean pucker grade of fabric samples and corresponding variance of profile lines. As mean pucker grade increases, (in terms of value) variance of profile lines decreases.
A simple linear regression model of the form was fitted to this data. The obtained equation is MG = -0.7748*ln (V) + 8.157and the coefficient of determination R2 = 0.67.
The lower values of coefficient of determination are the result of very low scanned area of stitched fabrics obtained in the study. The scanned area of fabric used in the analysis is only about 4 x 4 cm2. This area is limited due to the constraints associated with optics especially smaller aperture size of CCD camera used in this study. If the scanned area of fabric is increased considerably, an improvement in correlations between attributes and means pucker grades would be expected.

Notations:

MG is Mean pucker grade
MD is Mean deviation of sample
V is Variance
AR is Surface area ratio
PSD is Power spectral density
Y is Response variable,
X 1, X 2, X 3, X 4 are Regressor variables,
ß0 , ß1, ß2, ß3 ,ß4 are Regression coefficients.

To objectively evaluate seam pucker, various attributes characterizing the seam pucker have to be determined from three dimensional data of fabric samples. The attributes essentially describes the variation in the shape of the fabric. Each attribute describes the variations in the height data (Z data) in a different way. In this study, four attributes namely Mean deviation, Variance, Surface area ratio and Power spectral density were utilized for describing the variations in height data and characterizing the seam pucker. The mean deviation (MD) and variance (Var) of n data points are:

The Fourier transform of a discrete function of two variables f(x, y),
x=0,1,2……M-1
y=0,1,2…....N-1 is defined as

for u=0, 1, 2…..M-1
v=0, 1, 2…..N-1

In general, the components of Fourier transform are complex numbers and they can be represented in polar quantities.

Where,

is called magnitude or spectrum of Fourier transform.

is called phase angle or phase spectrum of Fourier transform. Power spectral density is defined as square of Fourier spectrum.

R (u, v) and I (u, v) in above equations are called real and imaginary components of transform respectively.

Surface area ratio measures the extent to which the surface area of the puckered fabric deviates from that of the fabric without pucker. The surface area ratio is formally defined as the ratio of surface area of fabric to the projected area of fabric in XY plane.

Area ratio= Fabric surface area / Projected area

The attributes determined from the preprocessed data have to be correlated with present method of subjective evaluation. Various methods of subjective evaluation is available, the most widely used being AATCC standard 88B. According to this standard seam pucker was classified into five grades 1 to 5, grade of 1 refers to heavily puckered fabric and 5 refers to fabric with very little or no pucker at all. It should be noted that AATCC provides only reference photograph, not reference specimens. The sewn fabrics are compared with this reference photograph. The grade of fabric is the grade of the reference specimen which matches most nearly to sample fabric specimen. About 10 judges were asked to grade the seam pucker of 25 samples. The final pucker grade of each sample is the average of grades given by 10 judges. The results of subjective evaluation conducted are shown in Table 3. The mean pucker grade varies from grade 1 to grade 4.9. A pucker grade of 1 was assigned to sample PD1 and a mean pucker grade of 4.9 was assigned to sample PC5. Each of the attributes describing the seam pucker was correlated with mean pucker grade using regression analysis and a regression equation was obtained for each of them. The scatter plot of Mean deviation of samples and their assigned Mean pucker grades given by the judges is shown in Fig 12. The scatter plot of Variance against Mean pucker grade is shown in Fig. 13.

The linear relation between mean deviation and mean pucker grade is clearly seen. A linear regression equation correlating Mean pucker grade and Mean deviation was obtained using simple linear regression analysis. The obtained equation is MG = -2.2781*MD + 4.8998 and the coefficient of determination R2=0.8933. The adequacy of the fitted model is clearly evidenced by a high coefficient of determination.

A regression equation correlating Mean pucker grade and Variance was obtained using simple linear regression analysis. The equation obtained is MG= -0.8483*ln (V) + 2.67428998 and the Coefficient of determination R2=0.8907. The high value of R2 indicates the adequacy of the fitted regression equation. The scatter plot of Mean pucker grade and surface area ratio is shown in Fig 14. A regression equation relating Mean pucker grade and surface are ratio was obtained using simple linear regression analysis. The obtained equation is MG = 5E+08* e-18.473*AR and the coefficient of determination R2=0.9247. The coefficient of determination obtained in this case is higher than those obtained using attributes Mean deviation and Variance. Due to this, Surface area ratio is a better attribute for objective evaluation than Mean deviation and Variance.

The scatter plot of Mean pucker grade and Power spectral density is shown in Fig 15. A regression equation relating Mean pucker grade and Power spectral density was obtained using simple linear regression analysis. The obtained equation is MG = -0.9348*ln (PSD) + 5.2978 and the coefficient of determination R2 = 0.895. The adequacy of the model can be inferred from a high coefficient of determination.

Each of the attributes describing seam pucker and mean pucker grade are correlated to higher level. It is intuitive to correlate all the attributes simultaneously to mean pucker grade. A multiple linear regression analysis was performed to fit the regression model of the form, Y=ß0 + ß1 X 1+ ß2 X 2 + ß3 X 3+ ß4 X 4. The response variable of multiple linear regressions is Mean pucker grade and Regressor variables are seam pucker attributes namely mean deviation, variance, surface area ratio, and power spectral density. The multiple linear regression equation obtained is MG = 35.3 - 3.24 * MD+ 0.852 * V - 29.9 * AR - 0.00293 * PSD and the coefficient of determination R2=0.953. The adequacy of the fitted model can be verified using the residual analysis of fitted regression model. This is done by analyzing the normal probability plot of residuals. As can be seen from Fig 16, the calculated residuals lie on a straight line, indicating that the fitted multiple linear regression equation is adequate. The multiple regression equation can be used to objectively evaluate the seam pucker in fabrics. Table 4 shows the mean pucker grades of fabric samples obtained using subjective evaluation and objective pucker grade obtained using multiple regression equation.

Figure 17 illustrates the scatter plot of subjective and objective pucker grades. The subjective and objective grades are correlated in a linear fashion with a correlation coefficient of 0.9764. The high value of correlation coefficient indicates that the multiple regression equation can be used for subjective evaluation. This is expected because the developed model has a very high coefficient of determination of 0.953. Also the ambiguity in subjective evaluation can be eliminated by objective evaluation using this multiple regression equation.

Therefore, in 2D image analysis, the attributes ‘Mean deviation’ and ‘Variance’ of profile lines correlate moderately with subjective grades. The coefficient of determination R2 varies from 0.635 to 0.67. The lower values of R2 are attributed to smaller scanned area of fabric used in the study. There is further scope left to improve evaluation of seam pucker based on 2D image analysis by using considerably large scanned area of fabric than that used in this reported study and having uniform illumination (as the illumination of the present system was not uniform) of the fabric by having better light source.

The novelty of the present invention is based upon the 2D technique based on line deviation. Puckered or wrinkled fabrics have distorted planes in the form of projections and valleys that are called waves. The severity of these distortions is exemplified by both amplitudes of the waves (heights) and the frequency with which the waves appear on the fabric plane. Human perception of the appearance is essentially about the severity of the distortions, i.e. higher the amplitude of the wave and presence of too many waves, worse the human perception of the appearance. The line deviation attributes of distorted/puckered/wrinkled fabrics (irregular or stitched fabrics) from the reference plane (smooth) depends both on the amplitude and frequency of the waves present on the fabric. Colour, shade, pattern and texture of the fabrics influence intensity variations of the captured images of fabrics and hence the accuracy of the features of the images during digital image processing. However in the described method, only the patterns of lines are to be extracted and hence, the influences of color, shade, pattern and textures of the fabrics on the outcome of the results (errors) are masked to a greater extent. Previous methods based on measuring the intensity variations or areas of the shadows present on the captured images of fabrics are less accurate as they greatly suffer by the colors, shades, patterns and textures of the fabrics.

Accordingly, the present invention relates to a 2D method based on projection of structured light over the fabrics of appropriate sizes (stitched and unstitched ones) and capturing the images of the fabrics at an slant angle not necessarily at 45? (as we have demonstrated only at 45?). A method to fix the fabric on the fabric holder with the stitch line perpendicular to horizontal table. A fringe generating system comprising of expansion of laser beam from the laser source using a beam expander and spatial filter unit and collimation using a collimating lens; making the collimated light incident on a single optical element interferometer to make two sinusoidal wave fronts that interfere to produce fringes; and a method to change the number of fringes per mm. A method of projecting the structured light using projecting lens over reference plane and fabrics and capturing the grey images of the structured light by a CCD camera using imaging lens. Methods of processing the images of the fabrics for noise removal, technique to link the missing lines (which might be due to defects of uniform illumination, non-optimum slant angle of the camera, limitation due to the resolution of the CCD camera employed exist with current demonstration system) to extract features of lines. Evaluation of seam pucker/levels of undulation on fabrics/irregularity of fabric surface/appearance of fabrics/textile materials through various attributes/ characteristics of lines such as mean deviation, variance etc.

Table 1 Seam pucker grading by subjective evaluation

Samples J1 J2 J3 J4 J5 J6 J7 J8 J9 J10 Mean rank
PA1 1 1 1 2 1 1 1 1 1 1 1.1
PA2 2 2 1 2 1 1 2 2 2 2 1.7
PA3 3 3 2 3 2 2 2 2 3 2 2.4
PA4 4 4 3 4 3 4 2 2 3 3 3.2
PA5 5 5 4 4 4 5 3 3 4 4 4.1
PB1 1 1 1 1 1 1 1 1 1 1 1
PB2 2 2 2 3 1 1 1 1 1 1 1.5
PB3 3 3 3 3 3 3 3 1 3 2 2.7
PB4 4 4 4 4 3 4 3 3 4 4 3.7
PB5 4 5 5 5 5 5 5 5 4 5 4.8
PC1 1 2 1 3 2 2 2 2 2 1 1.8
PC2 2 2 2 4 2 1 1 2 3 2 2.1
PC3 3 3 3 3 2 2 4 3 3 3 2.9
PC4 4 4 4 5 3 3 4 4 4 4 3.9
PC5 5 5 5 5 5 5 5 5 4 5 4.9
PD1 1 1 1 1 1 1 1 1 1 1 1
PD2 2 1 1 1 1 1 1 1 1 1 1.1
PD3 3 3 2 2 1 2 2 2 2 2 2.1
PD4 4 4 4 4 2 3 3 3 4 3 3.4
PD5 5 5 5 5 4 5 5 4 4 4 4.6
PE1 1 1 1 2 1 1 2 1 1 1 1.2
PE2 2 2 1 3 2 1 2 2 3 1 1.9
PE3 4 3 3 5 3 3 4 2 4 4 3.5
PE4 4 4 4 5 3 3 4 4 4 4 3.9
PE5 5 5 5 5 4 4 5 4 4 5 4.6

Table 2 Subjective and objective grades of fabrics
Sample AATCC grade Fitted grade Sample AATCC grade Fitted grade
PA1 1.10 1.43 PC4 3.90 4.02
PA2 1.70 1.91 PC5 4.90 4.59
PA3 2.40 2.57 PD1 1.00 0.61
PA4 3.20 3.56 PD2 1.10 1.35
PA5 4.10 4.67 PD3 2.10 2.38
PB1 1.00 1.31 PD4 3.40 3.26
PB2 1.50 1.81 PD5 4.60 4.59
PB3 2.70 2.75 PE1 1.20 1.13
PB4 3.70 3.45 PE2 1.90 1.50
PB5 4.80 4.92 PE3 3.50 3.14
PC1 1.80 1.76 PE4 3.90 3.48
PC2 2.10 1.88 PE5 4.60 4.43
PC3 2.90 2.69