Abstract:

Objective In order to solve the problem that the existing color prediction methods for woven fabrics cannot meet the color prediction requirements of jacquard fabrics with high warp and weft density, a color prediction method is proposed, which can not only reflect the characteristics of the interwoven structure of the fabric, but would also take into consideration of the deformation of the yarn and the shaded relationship formed by the microscopic three-dimensional structure of the fabric.
Method Through the color separation of the color fabric images, four types of characteristic patterns were utilized to reflect each color weft, warp, shadows on the fabric surface, and the textures of the yarn materials were obtained, respectively. Based on calculating the area ratio of four characteristic patterns, the color reassignment was carried out, and then the color prediction model of the fabric was constructed by the Lab proportional addition method. The predicted and the measured values of fabric samples were calculated for brightness difference, red-green difference and yellow-blue difference, respectively, and then the ridge regression algorithm was adopted to correct the deviation of the three sets of data on color difference. The processed color difference value was taken as the dependent variable, the area rate of the four characteristic patterns was taken as the independent variable, and the linear regression model between the two was re-established to complete the optimization of the prediction model.
Results Three groups of red, yellow and blue gradient-changing fabric samples based full-color compound structure with triple-weft were designed and woven, and all 14 fabric samples prepared were measured by spectrophotometer to collect color data as the basis for testing and evaluating the prediction model. The color prediction accuracy of the basic prediction model and the optimized model was compared and analyzed, and it was found that the mean of the total color difference between the optimized predicted value and the measured value was decreased from 2.01, 2.28 and 2.49 to 0.56, 0.52 and 0.60, respectively, which proved that the optimization method had a significant improvement effect. The predicted value and the measured value were analyzed by item fitting, and the prediction accuracy of L value was found the most stable among the three parameters, while a value and b value were determined according to the specific situation. The red group had the smallest difference in a value, followed by L value and b value. A significant difference existed in a value in the yellow group, while there was a small difference in L value and b value. The fit degree of a value and L value in the blue group was better, and the difference in b value was significant. By comparing the above experiment results with the color values of the three color wefts, it was found that the prediction model was sensitive to the quantity of color components of the fabric samples, and the higher the quantity of a certain color component, the more accurate the prediction. Consequently, the color prediction data of the proposed method, the geometric model method and the image color averaging calculation method are analyzed and compared. The results showed that the proposed method was the most accurate in prediction for all of the fabric sample groups, followed by the color averaging method, while the worst was the geometric model method. Compared with the other two methods, the mean total color difference of the proposed color prediction method is reduced by 92.09% and 89.35%, respectively, demonstrating higher color prediction accuracy.
Conclusion The prediction model proposed establishes a clear correspondence with the fabric research object in terms of interweaving structural characteristics and yarn color. By overcoming the problem that the geometric model method cannot reflect the yarn deformation, and the low correlation between the color average calculation method and the structural characteristics of woven fabrics, this paper provides a reliable method for the color prediction of fabrics with high warp and weft density, and provides a new idea for the design and research of color prediction models.

Key words: full-color compound structure with triple-weft, color prediction, color gradient woven fabric, color fabric image, woven fabric design, color deviation