Preparing Images for Embroidery رسالة ماجستير

Abstract

Embroidery may now be seen on shirts, caps, jackets, and a variety of other items. Our eyes must view embroidery as a fully-meaningful image since it has been knitted with a range of thread colors. Because most colorful photographs contain a large number of different shades of color, we are unable to produce thread in the same color for each of these shades. To overcome this problem, we looked up thread manufacturers and discovered that DMC uses 454 colors based on the Munsell color model, which is similar to the HSV color system. In order to turn each segment into the nearest color in DMC color space, we wish to use the AB dimensions from the CIELAB color space in picture segmentation. Image colors in the CIELAB color space will be reduced to the complete DMC colors utilized in the colored thread manufacturing process. Our major goal is to divide the image into small segments, each of one color, and then convert it to the nearest DMC Color to receive the thread we need to embroider the canvas. To acquire the requisite number of DMC colors, a k-means clustering technique or a superpixel algorithm would be used. We aim to extract all of the colors that the brain can identify in each embroidered image in this thesis. To decrease the number of various colors that may be utilized on canvas, we employed the Delta function for the color difference, k-means clustering, and superpixel. Because k-means clustering with delta function has less distortion or noise than superpixel, and because the output is much better with high resolution than low resolution, we may utilize high-quality photos and subsequently scale them as needed for canvas.