To reveal the intrinsic relationship between visual saliency calculation and the visual saliency experience and visual quality of the main space, based on a profound understanding of the visual perception process of space through visual computing, quantitative analysis was conducted on the visual saliency elements that affect the visual quality of the underlying space. The method of combining visual saliency calculation with eye-tracking was adopted to collect corresponding data for analyzing the visual quality of the building's underlying space. Firstly, the basic principle of visual saliency calculation was elaborated. Secondly, virtual reality models of different schemes for the open space of the underlying public building at different stair positions were established, eye-tracking experiments were carried out, and the ITTI visual saliency calculation model was used to generate visual calculation saliency maps. Finally, combined with subjective questionnaire surveys, the correlation between the visual saliency of the building's underlying space and the visual quality evaluation was analyzed based on the data. The results show that the real-time visual experience eye-tracking hotspots map and the visual saliency map generated by visual saliency calculation have a high similarity, and the visual saliency values are positively correlated with the subjective visual quality evaluation. The stairs located in the middle of the underlying space and the spaces with complete geometric shapes have visual saliency and become important factors affecting the visual quality of the building's underlying space. Finally, it was verified that computer visual saliency calculation conforms to the visual perception laws of building spaces and can be used as an effective tool for evaluating the visual quality of building spaces, providing a quantitative basis for human-oriented architectural design.