WU xudong 1 , LIU bingxiang 2 , ZHU Huiqiu 1 , FENG Luyuan 1 ,LI Yingxi 1
(1.Jiangxi Institute of Ceramics and Arts, Jingdezhen 333000, Jiangxi, China ; 2. Jiangdezhen Ceramic Institute , Jingdezhen 333403,
Jiangxi, China)
Abstract: In this study, the neural network algorithm is used for the discriminance of Longquan ceramics. Taking Longquan ceramic body as the research object, selecting 8 kinds of major and minor constituents, and using the software of matlab2012b, three algorithms are compiled for the experiment. The difference and the applicability of these algorithms are discussed, and also graphically and visually expressed. The experiment results show that the trained network can accurately tell the unknown Jingdezhen replica from the authentic Longquan ware.Through the detailed comparison of the advantages and application scopes of these three methods, a new approach is provided for the identification of ancient ceramics.
Key words: neural network; Longquan ceramics; identification.