Textile Texture Image Segmentation And Colorization

This article discusses two important problems, which are the segmentation and colorization of the texture images of textile, from an application perspective. Although there are many related algorithms, proposed algorithm is superior and innovative.The reflectance of textile fabric changes with location because of its construction features. Photometer can smooth the reflectance and get a reliable result. However, it can’t get the reflectance of small area, because the photometer only achieves the average reflectance of a fixed-size circular area which is larger than the detail area. Our system calculates the average reflectance of an area grouped by pixels with the same color. Segmentation is used for finding these areas. In order to get the accurate result of segmentation, we present a seed region growing algorithm for image segmentation. Human-computer interaction is necessary to assign the seeds of each color region. We propose an algorithm based on the open operation and close operation of image morphology so as to expand the seed and segment the detail areas correctly.Colorization is the process of adjusting image colors. It can adjust the color, the brightness and the contrast. During the design process of textile fabric, designers look for dozens of colors to match the pattern. In the traditional way, a lot of fabric is printed in order to find the best one. It’s costly and time consuming. We prefer to simulate the result by our algorithm if it can create real effect. Firstly, the proposed algorithm uses the result of segmentation and Gaussian probability to identify the area to be colored. Secondly, we look for proper model and color space. Lastly, we colorize the edge pixels based on optimization.