Guangxi Zhuang bronze drums are important material carriers of ethnic cultural heritage, yet their digital images often suffer from pattern wear, edge fracture, surface corrosion, color degradation, and local information loss caused by long-term preservation, environmental erosion, and acquisition limitations. To support the sustainable digital conservation and dissemination of bronze drum cultural heritage, this paper proposes an image restoration method that integrates degradation prior guidance with a damaged-region structure–texture collaborative constraint. Based on a generative adversarial network, the proposed method explicitly models complex degradation cues, including weathering, corrosion, wear, and color attenuation, on the generation side, thereby improving the perception and restoration of degraded cultural heritage image regions. On the discrimination side, local structural continuity and texture consistency are jointly constrained to enhance the visual authenticity of restored decorative patterns and metallic surface details. In addition, a Guangxi Zhuang bronze drum image restoration dataset is constructed to provide data support for intelligent cultural heritage conservation research. Experimental results show that the proposed method achieves an MSE of 27.318, an MAE of 2.431, a PSNR of 36.284, and an SSIM of 0.956, outperforming representative restoration methods in pixel fidelity, structural similarity, and visual coherence. The results demonstrate that the proposed method can provide an effective non-contact digital restoration tool for damaged bronze drum images, reduce reliance on repeated physical handling of cultural objects, and support the long-term preservation, digital archiving, visual interpretation, and public dissemination of ethnic cultural heritage within a sustainable development framework.