[S-VC33] Monitoring and assessment of volcanic activities

[SVC33-04] Real-time volcano eruption detection with visual IoT system

*Kazutaka Kikuta¹, Ken T. Murata¹, Takeshi Nishimura² (1.National Institute of Information and Communications Technology, 2.Tohoku University)

We propose a volcanic plume detection method using a moving object detection algorithm on visual IoT technology. The advantage of using visual IoT is the availability of high-resolution, high-temporal-resolution videos. The cameras to be installed are commercially available IP cameras with high-cost performance. We developed a system capable of performing 24-hour monitoring of volcanoes and image processing on the edge side and the cloud side. The amount of communication can be reduced by splitting processing, allowing the system to be deployed in remote locations where data transmission is performed via mobile communication. The edge-side PC extracts potential volcanic plume candidates through lightweight image difference detection processing and forwards them to the cloud-side server. On the cloud side, detection is achieved through motion vector evaluation using optical flow, which results in efficient processing and a high detection rate.