Abstract: This article analyzes the pain points of the traditional DSM, DEM, DOM image data processing production mode, after a lot of preliminary testing and research, the method based on HCI, GPU cloud computing and cloud desktop is used as the underlying system architecture of this research, builds a production resource pool for the multi-tenant model, in order to solve the shortcomings of the traditional mode. Through the test data comparison, the centralized processing method based on the cloud environment has a significant improvement in resource utilization and management methods, which can give full play to the capabilities of underlying hardware, further improve production efficiency. At the same time, it can effectively solve the problem of resource waste caused by repeated procurement and repeated deployment. The research results of this article will also be further promoted and applied in the actual project.

Key words: cloud environment, centralized processing, image data processing, multi-tenant, improved efficiency