Abstract
Urban greenery offers significant ecological and psychological advantages; however, widely used visibility metrics, such as the Green View Index (GVI), are limited to a single horizontal viewpoint. Modified metrics that encompass broader viewpoints often require specialized expertise and lack temporal and spatial adaptability. To overcome these limitations, we have developed a practical method that can be deployed in the field for assessing urban greenery visibility. This method uses a commercially available 360-degree camera and AI-based image segmentation. Panoramic images were transformed into upward- and downward-facing fisheye views with an equisolid angle projection. Vegetation was then extracted from the images using the SegNet model. The resulting metric, designated as the Spherical GVI (SGVI), was applied to two contrasting 1 km × 1 km urban areas in western Japan, namely Tomio (a residential area) and Esaka (a commercial area), to evaluate its usability and robustness. SGVI distributions differed between the two areas, reflecting their distinct spatial characteristics. To evaluate the robustness of the SGVI, it was compared with the conventional Panoramic GVI (PGVI), which is derived from an equirectangular projection. The results showed that SGVI was generally higher than PGVI, particularly in the commercial area where PGVI tends to underestimate the amount of greenery due to geometric distortion. These findings highlight the importance of projection-aware metrics in urban greenery assessment. This cost-effective, reproducible, and scalable approach requires little technical expertise or processing effort, making it appropriate for municipal green space surveys, policy monitoring and citizen science initiatives.