The integration of rail transit stations and the surrounding underground space presents volume scale and functional complexity, and it has gradually formed a complex integrated transportation hub. Wayfinding inside its space is also more difficult, and the behavioral phenomena of passengers stopping, wandering, and getting lost are common. This paper uses station A of a large transportation hub as an example. It combines target detection technology and the optical flow method, designs the traffic space guidance system, tests how well passengers can find their way by moving their eyes, and then looks at how the visual space can be improved in terms of color, comfort, and subjective comfort after getting feedback. The data from the passenger eye-movement test reveals that during the sub-processes of information perception and elevator path selection, the logo is the primary focus, with attention ratios of 72.35% and 61.28%. Orange, yellow, and green were the three colors that received the highest ratings for subjective spatial comfort from passengers, with ratings of 0.58, 0.55, and 0.35, respectively, which are data-supported for visuospatial optimization.