Aiming atthe existing fresco restoration methods based on generative adversarial networks, their generated samples lack diversity and are prone to large-scale feature loss and other problems. A virtual restoration method for fresco images based on bi-generator generative adversarial network (BGGAN) is proposed. Firstly, sample generation from two random directions ensures the diversity ofgenerated samples. Secondly, forthe Dilate U-NetKares generator model, the inflated convolutional expansion rate in the downsampling stage is improved and the pooling operation is eliminated. Finally, the loss function is designed to combine the MSE loss with the adversarial loss, and the feature gradientofthe generated samples is constrained byλG. Restoration tests are performed on the collected mural dataset, and the testresults are compared with multiple imagerestoration methods. The results show thatthe imagerestoration results obtained by the proposed algorithm have clearer details. The peak signal-to-noise ratio (PSNR) of the restored image is improved byabout1.12dB on averagecompared to thecomparison model, and thestructuralsimilarity(SSIM) is improved by about0.047 on average.