Jacquard Fabrics on Demand NTC Project F03-NS03s

Transcription

1 Jacquard Fabrics on Demand NTC Project F03-NS03s Project Team: Leader: Alan Members: Abdelfattah Seyam/NCSU/ Robert Barnhardt/NCSU/ David Hinks/NCSU/ Student: Kavita Mathur/ Project Website: Objective The goal of the proposed research is to develop an automated digital color imaging system to help streamline the design and production of Jacquard fabrics. The system will enable the Jacquard fabric designer to digitally select accurate color/weave patterns, receive rapid customer response prior to sample weaving, and thus allow delivery of accurate color reproduction in the final fabrics. This research will enhance the manufacturing process by contributing an automated tool to help Jacquard fabric producers create woven fabrics that are a very close match to the target image, thus providing a rapid response to customer needs that is both economical and efficient. Abstract Jacquard weaving provides the opportunity to design complex pictorial and other patterning effects from the combination of warp and filling colors and weaves. In the traditional fabric design process, the resultant visual perception of the design, using different colored yarns, can be attained only through the production of actual physical fabric samples, and this is a very time consuming process. No truly accurate digital color methodology is yet available to assist designers in the initial development of product samples. Currently, there is very poor correlation between the color that is shown on the screen and the actual weave structure created to simulate that color. The problem of achieving accurate color fidelity in the color reproduction of woven fabrics has been addressed, and work is being undertaken to achieve the highest possible color matching accuracy in Jacquard woven fabrics. This research is so far quite successful in creating a database of color/weave structures for the quality color reproduction of Jacquard fabrics. The research places emphasis on synthesizing computer generated color images that are perceptually close to the actual woven samples, with a high degree of color accuracy. Background and Literature review In the traditional fabric design process, colored yarn placement within a particular weave structure is planned-out on special graph paper (today, using a CAD system) -- then samples of the resultant fabric woven and assessed. Often, several iterations of this process are required,

2 and it is a very time consuming and costly process. In these applications, it is critical to simulate the color images with a high degree of color accuracy. For the textile and clothing industries in particular, it is generally accepted that pass and fail tolerances for colored goods fall within about 1.0 to 1.5 CIELAB color difference units. If a method of color matching cannot achieve this magnitude of color accuracy, it will affect the final samples and in turn will affect color quality judgment. Ultimately, the goal of a flexible electronic textile design system is to simulate the appearance of the final textile, including weave, color and fabric from any given external design source. Relatively little work has been done within the woven fabric area in the context of representing accurate color imagery of complex weave patterns [1,2,3,4,5]. Weaving involves the interlacement of two sets of yarns; one called the warp with a second, orthogonal set of yarns, referred to as weft (filling). Numerous descriptions of this process exist within the computer environment [6,7,8,9,10] addressing, algorithmically, the problems that arise when one attempts to harmonize visual pattern with the notational point paper diagrams of those warp and weft interlacings. Hoskins et al [1] developed an algorithm to analyze the color of woven structure, while Dawson [2] examined color and weave effects with small repeat sizes. He studied the basic effects of yarn color sequences over one weave repeat. For each weave family, the authors determined which of these effects can be obtained with one color of warp and two of filling and vice versa, and with two of each. Dawson s research was confined to small weave repeats up to 4x4. Grundler and Rolich [3] proposed an evolution algorithm to match weave and color. The software offers a wide range of fabric patterns and is also able to create new ones based on user s choice. The study emphasized matching the weave and yarn color, in order to have a predetermined idea of the appearance of the fabric to be produced. Adabala et al. [4] presented a technique for visualizing woven fabrics in real time, while optimizing the realism of the original image. However, in the above-mentioned works, consideration was not given to the actual color accuracy of the reproduced images. Osaki, however, [11,12] proposed a method for the development of high quality color reproduction on silk Jacquard textiles from digital color images. His study estimated the appropriate criteria for better evaluation of the quality of color attributes of woven textiles converted from digital color image. While use of colorimetry has helped to find a way to get better reproducibility and accuracy in textile wet processing, it is clear that no commercial system is available for accurate color imaging of Jacquard weave designs without first making the weave in prototype form. Our project seeks to address this problem. Approach One approach that has been used for on-screen color visualization of woven fabrics is to assume the color generated by a woven structure is a solid color. However, this is not likely to provide an accurate image rendition of the actual structure of complex Jacquard weaves, produced by mixing pre-colored yarns in the weft/warp. Hence, a more complex mathematical model and database is required. The process adapted in this research for woven image reproduction, with high color accuracy is shown below:

3 Original Image/ Artist Work Calibrated Color Display via Textile CAD system RGB Pixel Mathematical transformation of colorimetric image data, into weaving information Simulated Digital woven image Jacquard Loom Woven Jacquard Fabric Color/Weave Database Numerical and psychophysical color difference evaluation, compared to original digital image Colorimetric Evaluation The development of the database for textile color/weave combinations has led us to a new creative approach where a large number of colors can be reproduced with high precision on textiles, by weaving with only a limited number of dyed threads. By making use of the limited physical database, it is possible to create an imaging system that can display remarkably accurate reproduction of colors on fabrics that can be used to predict the final appearance of actual Jacquard weaves. To achieve this, colorimetric attributes for each pixel on the digitized original image are transformed into weaving data on the textile substrate by minimizing the color difference in CIEL*a*b* representation between actual weaves produced in the database and the simulated image/color of the woven fabric. The color difference used is the conventional formula in CIELAB color space and is defined by the difference between each attribute (e.g., lightness, redness-greenness, and yellowness-blueness) in the original image and that of the simulated woven fabric: E = ( L*) 2 + ( a*) 2 + ( b*) 2 Where, L* = Lightness a* = redness-greenness b* =yellowness-blueness. And, E is the permissible color difference between sample and specified color (Standard) L*= L* sample - L* standard + L* means sample is lighter than standard - L* means sample is darker than standard a*= a* sample - a* standard + a* means sample is redder than standard - a* means sample is greener than standard b*= b* sample - b* standard + b* means sample is yellower than standard - b* means sample is bluer than standard Psychophysical Evaluation A psychophysical experiment has been developed to determine the visual difference between the on-screen color and actual physical samples. A set of observers will be used to assess the

4 perceived differences in color for a representative set of actual weave structures. For this experiment: Colorimetric data has been calculated using the same light source as the viewing conditions. Viewing conditions will be highly controlled and standardized. The variability in electronic communication of color data to other locations will be tested. Color/Weave Database Development A scheme is developed to design the database tool, in order to make the color information accessible and easy to use. The database is designed by using information technology (here the information technology used is the database technology). This system provides a rapid application development tool which requires interface construction tools (Front end), database systems (Back end), and a way to get data into the database and to get data out of the database. In this research we would be constructing forms (as an interface between user and database) that will consist of all the required input fields by the user in order to get the closest possible color/weave match for the target image. On the Back end side database procedure and SQL (Sequential Query Language) queries will be written to process and retrieve information from the database as requested by the user. The color information will then be accessible with just a click of a mouse. The general outline of the database tool is shown below: Database User Interface Weave designs Database Engine Color image Color Attributes Color / Weave Database Weave structure Weaving Specifications Color/Weave related attributes Color information/l*a*b* values Color space model Yarn color Yarn type Weaving specifications As can be seen from the database outline, given the color specifications, the program tool will display those color images closest to the color selected in order to give designer a choice of colors along with the corresponding attributes of the image. With the help of the color/weave database, designer and producer will able to get all the possible color images which, in turn, will help them to achieve the best possible combination of weave structure and related attributes (as seen in the user interface).

5 Theoretical and Experimental Analysis Reproduction of color to exact specifications is a major concern of the textile industry. This is particularly true of the rapidly growing computer-aided-design (CAD) sector. Here designs are simulated on a computer screen in full color, and a variety of tools is provided for selecting and manipulating the color on the screen. The Color/Weave database generation is an attempt to simplify the way of selecting the required color for the woven fabric, within the gamut of the monitor. In order to predict a weave color of given specifications, we began by creating a comprehensive set of intricate colored woven designs fabricated from a four color rotational warp sequence (red, green, yellow and blue), with what is essentially a pick-and-pick, black and white filling. A wide variety of different color and weave interlacings were developed in order to obtain the largest possible color gamut through these warp and weft color combinations (Figure-2). The possible colors reproduced in the textile material were thoroughly investigated by checking various combinations of basic textile weaves. Color/weave structures are basis of all color imaging. From the Color/Weave database, it has become possible to successfully develop a new technique for converting the original digital image data (Figure-1) to fabric construction data for a Jacquard machine. Color Measurement (Spectrophotometer) Color Attributes L*=38.62 a*=10.41 b*=31.38 Color Reproduced (Color Calibrated Monitor) Figure-1: A weave structure (with microscopic image) developed from four warp color rotation and black and white in filling (left); Color reproduced on color calibrated monitor by using colorimetric data (right). Since, woven fabrics are inherently highly textured, the number of interlacings, in conjunction with the placement of the yarn colors determine the overall resultant color of the structure. The color of the yarn floats, dominant upon the surface of the weave, greatly contributes to the color effect produced.

6 Figure-2: Weave Blanket showing weave/color combination woven from four warp colors(green, red, yellow, blue) and black and white in weft. Since the last funded seed project in this area, a database of 218 color/weave structures has been generated for quality reproduction in Jacquard fabrics. The color of each structure was measured using a spectrophotometer (Datacolor SF600X) in daylight-simulated illumination with the CIE 10 degree standard observer. Hence, a dataset of reflectance curves for each weave was produced in order to obtain perceptually accurate color images on a high definition colorcalibrated monitor. The colorimetric data achieved from each of the 218 structures was then defined in the CIELAB color space model (Figure-3).

7 Figure-3: The color spectrum of the actual color/weave samples, converted into a two-dimensional plot. Figure-3 shows a two dimensional slice of the color gamut, achieved using the physical samples. Most of the colors lie between ±20 on both axes, but some color samples are >20 (i.e., higher chroma) on the yellow axis. There are a majority of colors in yellow region and fewer in the blue region. The designs selected to date show variability in hue and chroma, and the achievable color gamut is reasonable. However, we plan to extend the database further, into the blue region. The resultant colors obtained from the initial set of samples enables prediction of colors that can be incorporated in order to achieve as wide a color gamut as possible. The data indicates the need to expand the weft yarn sequence to include the black and white yarns along with four other colors. These new colors consist of variations on red, yellow, blue and green -- with alterations in the underlying weave structures such that more blue colors are produced. We expect that the results gained will contribute greatly to an expanded palette; one exhibiting greater color clarity. Due to the 3-dimensional nature of fabric structure, the weaves tend to change color when observed from different positions. Colorimetric data were calculated using the DigiEye color imaging system that utilizes a calibrated camera and calibrated lighting conditions (Figure-4).

8 Figure-4: DigiEye Color imaging system The system can capture the image of 3D objects which are displayed on a color calibrated monitor. The illumination cabinet of the system has a daylight simulated environment, and hence the monitor correlated color temperature was also set to the Kelvin (i.e. Daylight color temperature) in order to ensure stable color results. Figure-5 shows the captured woven sample image along with its respective colorimetric data. Figure-5: Woven sample image (left), Reflectance spectra of the area selected (right), color appearance under different illumination conditions (top left).

9 The difference between the measurements made by spectrophotometer and Color Imaging system are exemplified in Table below. Reflectance Spectra Spectrophotometer DigiEye L* a* b* L* a* b* Color Tolerance E Table: Example of 10 color/weave structure showing reflectance spectra obtained from spectrophotometer and DigiEye color imaging system and the corresponding color tolerances E. The results from the color measurements show that there is significant difference in colorimetric data obtained by the two different methods. We are continuing to assess the reasons for the color differences, and it is likely that the texture of the designs is one of the reasons. However, ideally, in order to determine the correlation between the angular dependence of the reflectance properties of the 3D structures and human color perception, a goniospectrophotometer is required (one that can measure reflectance spectra from multiple angles of reflection). The best method of measurement will be determined by comparison with the visual data, as well as by assessing other color calibrated displays such as the Datacolor Colorite System. Once the color reproduction of the experimental weave structures is verified, then a color prediction algorithm for any colored yarn (warp or weft) will be developed. As a result, different filling colors can then be simulated to determine the maximum color gamut and design- -thereby reducing or eliminating the need for trial-and-error woven samples. Future Work Complete woven fabric physical sample dataset. Establish the scope and limitations (e.g. color gamuts attainable) of the color calibrated display and image capture system (example- DigiEye). Evaluate the percentage of each color contributing in a particular structure, using image processing and analysis technique. Update color / weave database with complete information. Numerical and psychophysical color difference evaluation, compared to original digital image. Incorporate database into an on-screen color computer program and integrate with the CAD software for the preparation of product design sampling.

10 References 1. Hoskins, J.A., Hoskins, W.D., May, J.L.W., Algorithms for color analysis, Proceedings of the 1985 ACM SIGSMALL symposium on Small systems, 1985, Dawson, R.M., Color and Weave effects with some small weave repeat sizes, Textile Research Journal, 72(10), 2002, Grundler, D., Rolich, T., Matching Weave and Color with the help of Evolution Algorithm, Textile Research Journal, 73(12), 2003, Adabala, N., Thalmann, N.M., Guangzheng, F., Real-time Rendering of Woven Clothes, Proceedings of the ACM symposium on Virtual reality software and technology, Xin, J.H., Shen, H.L., Computational model for color mapping on texture images, Journal of Electronic Imaging, 12(4), 2003, Hoskins, J.A., Hoskins, W.D., Algorithms for the design and analysis of woven textiles, Proceedings of the 1983 ACM Conference on Personal and small computers, Lourie, J.R., Loom-constrained designs: An algebraic solution, Proceedings of the ACM National Conference 1969, Lourie, J.R., Textile Graphics/Computer Aided, New York, Fairchild Publications Inc., Lourie, J.R., Bonin, A.M., Computer-controlled textile designing and weaving, Proceedings-IFIPS 1968, Lourie, J.R., Lornzo, J.J., Online Textile designing, Proceedings of the ACM National meeting 1966, Osaki, K., Reproduction of Various colors on Jacquard textiles by only eight kinds of color wefts, Proceedings of SPIE, Vol. 4421, April 2002, Osaki, K., High Quality Color Reproduction on Jacquard Silk Textile from Digital Color Images, AUTEX Research Journal, Vol. 3, No. 4, December Acknowledgments Datacolor International (Dan Randall, Color Manager, Textiles, North America), ScotWeave, Ltd., (Dave Kemp, Alan Watters, et al CAD programming, UK: Staubli Corporation (Yves Staubli, Technical Representative and Allen Thompson), DigiEye Plc (Sue Williams, UK).