Content-based lunar image retrieval provides a convenient and efficient way for accessing relevant lunar exploration images by their visual contents. To increase the efficiency, the process of content-based lunar image retrieval was analyzed and modeled using Petri nets, and a parallel mechanism was designed based on the model. A parallel architecture was then proposed for the content based retrieval of lunar exploration images. According to the architecture, an experimental system was implemented. Experiments upon real datasets including Chang’e lunar exploration images confirm that the proposed parallel architecture can effectively improve the constructive and retrieval efficiency.