Monitoring the carbon emissions of urban buildings is of great significance for assessing the benefits of emission reduction and promoting the development of energy-saving technologies in buildings. Visualizing building carbon emissions helps intuitively understand carbon intensity and enhances public awareness of environmental protection and their willingness to participate in carbon reduction. However, existing building carbon emission visualization methods are generally limited to data statistics and map visualization, lacking immersive visualization solutions from the first-person perspective and integration with street scenes. To address this issue, this paper proposes an augmented reality (AR) visualization prototype system based on urban 3D models, which achieves an immersive display of building carbon emission data by accurately aligning each frame of a street-view video from a mobile phone with the backend urban 3D model. For an arbitrary input street-view video, we first locate its position in the urban 3D model based on its location information. We then use a scene segmentation model based on deep learning to obtain segmentation results for buildings, roads, and the background in each frame of the video. Finally, using segmentation results as prior information, we optimize the photographic parameters of the frame images to achieve high-precision alignment between each frame and the urban 3D model, thereby realizing the AR display of carbon emission data in the street-view video. The prototype system developed in this paper is expected to be deployed on mobile phones or wearable devices such as virtual reality glasses, which can be considered a promising visualization technology for observing building carbon emissions from a first-person perspective.
Keywords building carbon emission, carbon emission visualization, urban 3D model, immersive display, image registration, augmented reality