Digital color classification for colorful cross stitch threads using RGB+Euclidean Distance and LAB+CIE94

This paper presents an approach of converting a digital image to a cross stitch pattern. The color space of the input image is quantized and classified into the nearest thread colors using k-NN classifier. Two different metrics are tested, i.e. CIE94 (which using LAB color space) and Euclidean Distance (which using RGB color space). The outputs are validated both quantitatively (using Color Coherent Vector/ CCV) and qualitatively (through human perception). The results have shown that CIE94 works best for grayscale and monochrome patterns, while Euclidean distance is more suitable for color patterns.