中国书画艺术作品印章真伪检测算法研究

Research on Seal Authenticity Detection Algorithm for Chinese Painting and Calligraphy Art Works

摘要: 当前中国书画艺术作品印章真伪鉴别以人工检测方式为主，存在主观性强、效率低且容易误判的问题，为此，本文提出采用计算视觉技术的书画印章检测算法。通过对印章图像进行特征区块提取、图像纠正、特征匹配得到待测印章图像的相似度评分，再使用双曲线函数对评分进行优化得到真、假印章的区别。经过在宣纸及普通打印纸上的印章图像进行实验结果分析，算法对真假印章的检测准确率可达98%以上。基于算法开发的印章鉴别检测软件，在书画印章鉴别检测领域具有较好的应用前景。

Abstract: The current identification of the authenticity of Chinese calligraphy and painting works mainly relies on manual detection, which is subjective, inefficient, and prone to misjudgment. Therefore, this paper proposes a seal detection algorithm using computer vision technology. By extracting feature blocks from the seal image, correcting the image, and matching features, a similarity score for the test seal image is obtained. Then, a hyperbolic function is used to optimize the score to distinguish between real and fake seals. After analyzing the experimental results of seal images on xuan paper and ordinary printing paper, the detection accuracy of the algorithm for real and fake seals can reach more than 98%. The development of the seal identification detection software based on the algorithm has good application prospects in the field of calligraphy and painting seal identification detection.