Journal of Digital Information Management Jia Zhaoyang, Chen Guangxue 15 5 2017 http://dline.info/fpaper/jdim/v15i5/jdimv15i5_2.pdf Digital image inpainting method based on multispectral image texture synthesis is proposed for the digital archives and high fidelity replication of defective Painting Arts. Firstly, multi-channel images of Painting Arts are obtained through multispectral technology. Inpainting technique based on texture synthesis is adopted for the multispectral reconstruction image. In the process of inpainting, the improved Criminisi algorithm is adopted to calculate the image gradient based on IHS space. It avoids wrong information propagation caused by the difficulty of direction matching and inaccurate gradient calculation of the correlation of color components in the RGB space. It can reduce the mismatch of the color patches in the inpainting area and pseudo color, and improve the digital inpainting quality of multispectral images. It has been proved that MSE of multispectral images inpainting qualities is 5.4021 and PSNR of multispectral images inpainting qualities is 41.6320, so improved Criminisi algorithm is superior to traditional image inpainting algorithm. It provides a reliable basis for digital inpainting, digital archives and high fidelity replication of defective Painting Arts.