A novel automatic image segmentation method for Chinese literati paintings using multi-view fuzzy clustering technology

Abstract

Under the background of contemporary cultural protection and dynamic inheritance, the interpretation and re-expression of the artistic connotations of Chinese literati paintings have become the main direction of heritage research. Digital technology and multimedia expression have become important means of cultural expression and transmission. Most Chinese literati paintings are ink paintings, and the particularity of ink painting makes it difficult to decompose and extract the screen content by simple means, which has caused difficulties in digitization, re-expression, and public interpretation to some extent. To solve this problem, a new robust multi-view (M-V) fuzzy clustering algorithm is proposed for image segmentation of Chinese literati paintings to achieve effective decomposition and extraction of ancient paintings. Through the effective decomposition and extraction of literati paintings, the electronic and digital transformation and preservation of literati paintings can be realized. This kind of preservation method, more than traditional scanning, can preserve the artistry of literati paintings, which is of great value for the re-expression and dissemination of cultural heritage. Experiments on noise-added Brodatz texture images show that the proposed algorithm is insensitive to noise and has good robustness. Experiments on real Chinese literati paintings show that the proposed algorithm can effectively segment literati paintings and further realize their decomposition and extraction.