A panoramic-image-based method for recognizing street shade conditions under direct and reflected solar radiation: Validation in central Tokyo

Source abstract

As urbanization progresses and urban overheating intensifies, pedestrian-level shade conditions are becoming increasingly important for characterizing solar exposure in dense urban streets. However, existing methods remain limited in their ability to recognize detailed shade conditions, particularly when solar reflections from glass facades are involved. To address this gap, this study develops and validates a panoramic-image-based method for recognizing pedestrian-level shade conditions in urban streets, including tree shade, building shade, and conditions affected by reflected solar radiation. The proposed method generates tree and building masks from panoramic images and predicts glass locations through a geometric transformation approach. The direct and reflected radiation components are then determined by combining sun path and reflection path analyses using the panoramic images. The method was validated using walking survey data and three-dimensional laser-scanning data collected in a high-density district in central Tokyo, Japan. Across the 113 non-reference evaluation locations, the proposed glass location prediction method achieved mean Precision, Recall, and F1 scores of 91.3%, 81.6%, and 86.1%, respectively. The proposed shade condition recognition method also achieved high overall performance, with Macro-F1 scores of 84.7% across all examined shade conditions and 82.9% across the three reflection-related shade conditions. These findings demonstrate that the proposed method can effectively distinguish detailed pedestrian-level shade conditions in complex urban streets, thereby providing useful information for shade management and street design in high-density street-canyon environments.