A study on the application of SAM classification algorithm in seal of calligraphy and painting based on hyperspectral technology

As an important part of Chinese painting and calligraphy, the seal not only has a high value of art, but also contains a lot of information about the author. Hyperspectral data has its unique advantages in the identification of the Chinese traditional painting and calligraphy seal. In this paper, the seal of WangXu collected by Capital Museum is taken as the research object. Firstly, hyperspectral image data of 400-1000nm wavelength range are collected by hyperspectral imaging instrument. Secondly, the average spectral curve is obtained by extracting the region of interest. The spectral angle mapping classification method is used to carry on the classification algorithm of the seal area. The threshold of spectral angle mapping should be focused, so as to achieve a better effect of mapping. With the comparison of the SAM classification effect of different thresholds in the seal of the painting and calligraphy, we have explored the spectral angle mapping rules of seal, and got the best classification threshold. The accuracy of classification and recognition of Chinese painting and calligraphy seal is greatly improved, and the hyperspectral imaging technology can be applied more effectively and efficiently to the identification and recognition of Chinese painting and calligraphy.