Monitoring changes in river water levels is essential for accurate flood prediction and disaster prevention. However, existing fixed water level measurement systems are often costly and inefficient for large-scale monitoring. This study presents a smartphone-based method for estimating water levels. The approach involves 3D reconstruction, visual localization, image-based water body segmentation, and Ray Casting. Our method estimates the camera’s pose and position from images, refining this through homography and IMU data. Rays are then cast along the segmented water body contours in the image to determine where they intersect with the 3D reconstructed area of interest. Water levels are calculated by averaging the heights of these intersection points. A field experiment at Sebyeong Bridge in the Oncheon-Cheon stream basin in Busan, Korea, demonstrated the system’s effectiveness, offering a reliable and scalable solution for flood monitoring. This method is expected to contribute to disaster management by minimizing flood damage and enabling rapid response.

E-mail: taeyun@pusan.ac.kr