Research on three-dimensional fusion method of surveillance video in smart park

The three-dimensional video fusion technology is increasingly becoming an important part of smart parks. In previous security work, surveillance videos are typically displayed in a multi-screen layout, making it difficult to directly analyze the spatial positions of scenes and objects in the monitoring footage. Based on this, a real-time fusion rendering method for RTSP video streams and three-dimensional park models is proposed, which uses texture projection to integrate real videos into virtual three-dimensional models, achieving the fusion of multiple video sources. For scenarios with multiple videos at the same location, a viewpoint-based weighted method is adopted to achieve texture overlay rendering. Finally, a smart park surveillance video fusion platform is built based on the Three.js framework, supporting real-time rendering on the web. The system also provides various interaction methods, including roaming and viewpoint switching, making it convenient for personnel to monitor key areas. Simultaneously, it achieves the effect of roaming in virtual three-dimensional scenes, enhancing the stereoscopic presentation of real scenes in surveillance videos. The system demonstration proves that effectively combining real-time surveillance videos with three-dimensional models results in better spatial presentation of videos and enhances the ability to discern the actual positions of scenes within the smart park.