主催: The Japanese Society for Artificial Intelligence
会議名: 2012年度人工知能学会全国大会(第26回)
回次: 26
開催地: 山口県山口市 山口県教育会館等
開催日: 2012/06/12 - 2012/06/15
Conventional image processing techniques are often too unsophisticated in their approaches to simulating the effects of aging and restoration of paintings. Current processing approaches fail to consider factors such as storage conditions, the status of pigments and surface materials, as well as historical trends in color selection and utilization. By neglecting these issues, realistic image manipulation of painted art is too often unsuccessful, both in terms of simulating weathering and age on contemporary works, and in attempting to restore or remove these same effects from older pieces. This paper proposes a knowledge-based approach to simulate and remove the effects of aging in traditional Chinese paintings. Related knowledge which concerning the degrees of color faded over time is discovered. The knowledge is computed by collecting, categorizing, and analyzing the images of paintings from different eras. In addition, relationships among color distribution histograms, the different subjects being depicted, the year or dynasty the painting was created, and the properties of different pigments are also tracked. By using our prototype system, both aging and reverse-aging processes can be simulated. Initial study results show that this approach has great potential in simulating aging effects on images of traditional Chinese paintings.