Fine-grained spatiotemporal estimation of tourism flows leveraging cross-video collaborative perception

Abstract

Microscale research on tourism flows is crucial for controlling such flows, analyzing resource-carrying capacity, and promoting sustainable development in the tourism industry. Current fine-grained monitoring of tourism flows using location-based big data faces challenges, such as inadequate user representation, data acquisition difficulties and spatiotemporal uncertainty. This study presents a method for the spatiotemporal modeling and estimation of regional tourism flows based on the collaborative perception of discrete surveillance videos. The method employed bridges the gap between physical and video image scenes by establishing collaborative perception relationships among multiple devices, thereby enabling the precise modeling and estimation of the dynamic spatiotemporal processes of population movement in the region. Empirical studies in real scenic areas confirm the adaptability of this technology to diverse geographical scenes and ensure the accuracy of the spatiotemporal flow estimation. This study addresses the challenges of high sampling costs and low spatiotemporal accuracy in regional fine-grained crowd estimation and offers technical support for near real-time dynamic crowd modeling and monitoring. The experimental results have the potential to assist in applications, such as tourism flow management, dynamic regulation and the risk analysis of group activities in scenic areas.