<p>Urban physical space environments are crucial for innovation activities. However, existing research predominantly focuses on the spatial distribution of innovation at medium and macro scales, the influence of economic and social factors, and the innovation network structures of urban agglomerations. To enhance scientific and technological innovation, it is essential to understand how urban physical space environments impact innovation performance at the micro scale. Leveraging patent data from Xi'an and street view images from Baidu Maps, this research introduces an interaction regression model framework. This framework aims to evaluate the combined influence of physical spatial scene elements and organizational structures on innovation performance. Through the integration of visual perception data and space syntax analysis, the study delves into the micro-scale spatial factors that determine innovation performance. The findings reveal that innovation performance correlates with the agglomeration patterns of scene-factor characteristics and the centrality of organizational structures. The combined perceptions of these elements by innovative talents significantly influence innovation performance. Specifically, factors such as green view index, walkability, road hardening ratio, sky view factor, and interface enclosure degree, along with centrality-based organizational structures, play vital roles. The inner city's physical space environment has a more pronounced impact on innovation performance than peripheral areas, contingent on environmental quality and spatial scale in the inner city, and on the intensity of the physical space in peripheral regions. A well-designed physical space can enhance physical activity and psychological well-being, boost knowledge spillovers, and improve innovation performance through spatial mapping effects.</p>