【摘要】 纹饰提取对残损文物数字化修复具有重要意义。针对出土的西汉铜镜纹饰表面铜绿覆盖、边缘侵蚀模糊、轮廓特征难以提取等问题，提出了一种融合二维和三维特征的古铜镜纹饰提取方法。借助微距镜头相机和超景深显微镜采集古铜镜碎片的二维和三维影像，根据2种维度信息的不同特点进行图像预处理和二维、三维特征模型的构建。将稀疏表示技术与古铜镜表面纹饰的二维和三维特征相融合，提取完整的古铜镜纹饰轮廓，并通过形态学处理和轮廓平滑技术强化纹饰图像。结果表明，“高清相机+超景深显微镜”的原始纹饰图案采集模式结合基于稀疏表示的特征融合技术能够解决薄小碎片纹饰图案难以准确提取的问题，可以原真性还原古铜镜纹饰。更多还原

【Abstract】 Ornament extraction is of great significance to the digital restoration of damaged cultural relics. To extract ornament of unearthed bronze mirror with copper-green surface overlay, blurred edge erosion and difficulty in contour feature extraction in the Western Han Dynasty, a method of extracting ancient bronze mirror ornaments by fusing two-dimensional and three-dimensional features is proposed. The method collects two-dimensional and three-dimensional images of ancient bronze mirror fragments with the aid of a macro lens camera and a super depth-of-field microscope, respectively, pre-processes the images and constructs two-and three-dimensional feature models according to the different characteristics of the two dimensions of information, uses sparse representation techniques to fuse the two-dimensional and three-dimensional features of the ornament on the surface of ancient bronze mirrors, finally the complete outline of the ornament of ancient bronze mirrors is extracted, and the ornament is enhanced through morphological processing and contour smoothing techniques. The images are enhanced by morphological processing and contour smoothing. The experimental results show that the proposed “high-definition camera+super-depth-of-field microscope” original decoration pattern acquisition mode combined with the feature fusion technology based on sparse representation can solve the problem of difficult to accurately extract the thin and small fragments of ornament, and can restore the ornament of the ancient bronze mirror authentically.更多还原