Intelligent Recognition of Colour and Contour from Ancient Chinese Embroidery Images

Abstract Ancient Chinese embroidery is an important intangible part of the cultural heritage of mankind. Its colours and contours are a major source of oriental inspiration and design elements for designers today. This study presents an effective intelligent recognition of colour and contour based on K-means++ clustering and the Canny operator for colour and contour application of ancient Chinese embroidery images and for digital inheritance and innovation. First, digital cameras and portable scanners were used in embroidery image acquisition. Second, colour level adjustment, sharpening and smoothing were specially added to the preprocessing, because of the ancient embroidery age or colour errors caused by the shooting angle. Third, K-means++ clustering was used for colour clustering. Fourth, the Canny operator was used for contour detection. After the addition of colour level adjustment and sharpening in the preprocessing, the colours and contours could be acquired accurately and more effectively from embroidery images and be read and edited independently.