Abstract: In recent years, the research on color restoration of mural image has become a hot spot in the field of mural cultural relics protection and display. Aiming at the problems that the overall feature information of mural image color restoration is difficult to extract and maintain effectively, and local color restoration is prone to generate false color phenomenon and color spill, a multi-level color restoration method of mural image based on gated positional encoding is proposed. Firstly, this method constructed an encoder network based on global feature constraints, and established feature pyramid of mural image through an improved multi-kernel multi-value pooling algorithm to reduce the overall feature loss of mural image in the feature encoding process. Secondly, in order to accurately restore the local color information of mural image, a color transfer module based on gating positional encoding was designed to restrict the learning of similarity kernel between content features and color features in spatial domain, to reduce false color phenomenon and color spill in restored image. Experimental analysis shows that the proposed method has good performances in restore the color information of mural image and maintain the global structural texture characteristics of the mural image to be restored.

Key words: mural image, color restoration, encoder-decoder network, global feature, positional encoding

摘要： 近年来，围绕壁画图像的色彩还原研究已成为壁画文物保护和展示领域的一个热点。针对壁画色彩还原面临的整体特征信息难以有效提取和保持、局部色彩还原易出现假色现象与色彩溢出等问题，提出基于门控位置编码的壁画色彩多级还原方法。该方法首先构建基于全局特征约束的编码器网络，通过改进的多核多值池化算法建立壁画图像特征金字塔，以减少壁画图像在特征编码过程中的整体特征损失；其次，为准确还原壁画图像的局部色彩信息，设计基于门控位置编码的色彩迁移模块，约束空间域中内容特征与色彩特征之间相似性核的学习，以减少还原图像中的假色现象与色彩溢出。实验分析表明，此方法能有效还原壁画色彩信息并保持待还原壁画图像的整体结构纹理特征。

关键词: 壁画图像, 色彩还原, 编码器-解码器网络, 全局特征, 位置编码