Visual navigation is a fundamental capability ofmobile service robots, yet the onboard cameras required forsuch navigation can capture privacy-sensitive information andraise user privacy concerns. Existing approaches to privacy-preserving navigation-oriented visual perception have largelybeen driven by technical considerations, with limited groundingin user privacy preferences. In this work, we propose auser-centered approach to designing privacy-preserving visualperception for robot navigation. To investigate how user privacypreferences can inform such design, we conducted two userstudies. The results show that users prefer privacy-preservingvisual abstractions and capture-time low-resolution preserva-tion mechanisms: their preferred RGB resolution dependsboth on the desired privacy level and robot proximity duringnavigation. Based on these findings, we further derive a user-configurable distance-to-resolution privacy policy for privacy-preserving robot visual navigation.