In order to improve the coloration efficiency of fashion design, an automatic color matching mechanism based on adaptive color clustering of image scenes is proposed for clothing patterns. Taking images of Sung porcelain as an example, 300 porcelain images from six different kilns were collected as testing samples. The porcelain area of each sample image was detected by image segmentation and denoising. The bipartite K-means clustering algorithm was utilized to adaptively extract the main colors of each sample. Also, the secondary clustering was carried out to obtain the main color values, proportions and co-occurrence ratios of each kiln’s image. A dynamic color matching mechanism integrating the number of color clusters, co-occurrence ratio, and structure characteristics of the target pattern, was designed and automatic color parsing and matching software was developed. The experiment results show that the color matching control parameters, source images, pattern shape, and other factors affect the main color selection sequence and the final matching effect. The time consumed for single sample color extraction and pattern automatic color matching are all less than 1 s, which can quickly realize the pattern color migration based on image scenes and further provide auxiliary decision-making for clothing pattern color design.
In order to fast identify and realistic simulate embroidery art, the respective texture characteristics of basic stitch in Simulation embroidery were extracted and preferred. Concretely, three feature extraction methods – gray co-occurrence matrix method (GLCM), Tamura method and gray difference statistics method (GDS), are combined to extract the features of embroidery needlework, and the best respected characteristics were compared and verified. Results shows that the best feature combination of needle texture in this paper is energy standard deviation, energy mean, standard deviation of moment of inertia, and mean of moment of inertia, which is defined as energy-moment of inertia feature. The proposed method effectively solves the inaccurate problem with single feature in the recognition of needle texture features, can be help for needlework recognition and virtual simulation.
In order to reduce the interference of stripe signal on the density automatic measurement of woven fabric and extend the the application scope, an adaptive Gaussian Notch Filtering (GNF) algorithm based on Fast Fourier transform is proposed. In the pre-proceesing stage, the Fourier transform was utilized to transform the woven fabric bitmap to spectral image, as to obtain the spectral characteristics of the striped fabric. The improved GNF algorithm was used to identify the peak of stripe signal, which determined the bandwidth radius and locate the stripe bright area. Based on the fused Fourier and GNF spectral map, interference frequency information was accurately removed and spatial map was restored to approximate pure color fabric image. Finally, the Morlet wavelet is utilized to analysis the density of the striped printed fabric. The experimental results show that the average subjective-objective consistency rate (AS-OCR) of the designed sample is 99.34%, and the range is 97.33%∼100%. The AS-OCR of the actual sample is 98.73%, ranging from 95.00% to 100%. The proposed method can effectively obtain the density of stripe printed woven fabric and expand the application scope of fabric density automatic detection.
Traditional Costume Color has significant value of transmission and regeneration in contemporary design. In this paper, we propose an automatic color parsing and transfer design system that assists designers to realize the creative design of the pattern color matching based on different traditional costume images and recommend sequence color schemes. The design system includes three modules: color extraction, automatic coloration, and color matching evaluation module. The adaptive bipartite K-means clustering algorithm is utilized to extract colors of each sample costume, while the genetic algorithm is applied to generate the optimal color matching scheme, and designers can participate in the design to obtain a satisfied scheme. The evaluation model combines the aesthetic measurement formula and the aesthetic measure of color harmony to evaluate the coloration quality. In the experiment, Jiangnan Han folk costumes are chosen as typical source samples to illustrate the effectiveness of the proposed method. Results show that different number of extracted colors and patterns can efficiently realize automatic coloration, and can provide auxiliary decision-making for pattern color design in fashion industry.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.