基于核分析技术和深度学习的古陶瓷智能识别系统的开发

精美的中国古陶瓷在国内外一直深受喜爱，大多数被保存在博物馆中，博物馆承担着传播陶瓷文化的重要角色。近年来私人向博物馆捐赠陶瓷质文物的情况日益增多，但捐赠品真伪的问题引起了很大的争议，相关法规明确指出博物馆需要保证文物的真实性。此外私人收藏品真伪无法保障导致陶瓷文物市场的发展长期处于停滞状态。而现有鉴定手段中依靠外观、器型等特征具有较强的主观性，依靠仪器的成分检测则停留在对比范围的初级阶段，都无法实现高准确率的鉴定，因此迫切需要建立一种更加科学可靠的鉴定方法。本项目拟开展基于深度学习方法对古陶瓷相关研究结果的深度挖掘，从大量古陶瓷样本的微观结构图像、各类光谱中挖掘新特征，结合古陶瓷成分数据建立快速甄别古陶瓷窑口以及真伪的方法，开发智能识别系统实现准确的鉴定。项目所开展的核分析技术以及深度学习在古陶瓷鉴别中的应用，也可以为其他种类文物鉴别方法的建立提供参考，在保护文化遗产等方面发挥重要作用。

The exquisite ancient Chinese ceramics are always deeply loved all over the world. Most of them are preserved in museums, which play an important role in spreading ceramic culture. In recent years, private donations of ceramic cultural relics to museums have been increasing, but the issue of the authenticity of donations has caused a lot of controversy, and relevant regulations have clearly pointed out that museums need to ensure the authenticity of their collections. In addition, because the authenticity of the collection cannot be guaranteed, the development of the ceramic heritage market has been stagnant for a long time. At present, the existing identification methods are based on appearance, shape and other aspects, which are highly subjective. And the detection and identification of ancient ceramics based on the instrument stays in the primary stage of the comparison of element concentration, and it is impossible to achieve high accuracy identification. So it is urgent to establish a more scientific and reliable identification method. This project plans to carry out the deep mining of the research results related to ancient ceramics based on the deep learning method, so as to complete the extraction of the characteristics from the microstructure analysis of a large number of ancient ceramics and all kinds of spectral analysis, which are beneficial to the description and distinction of ancient ceramic kilns. The method to identify the kiln and the authenticity of ancient ceramics rapidly would be established by combining these characteristics with the ancient ceramic composition data, and the corresponding intelligent recognition system would be developed to achieve accurate identification. The application of nuclear analysis technology and deep learning in the identification of ancient ceramics carried out by the project can also provide reference for other kinds of cultural relics, and play an important role in the protection of cultural heritage.