An Improved Adaptive Background Mixture Model for Realtime Tracking with Lightning Detection

An Improved Adaptive Background Mixture Model for Realtime Tracking with Lightning Detection

연구 분야: Infrastructure

학회: ICTCE '22: Proceedings of the 2022 5th International Conference on Telecommunications and Communication Engineering

Typical models of lightning detection are based on the acoustic, optical, and electromagnetic field properties of lightning, All of these lightning detection models require specialized instrumentation to implement. The image-based lightning detection model is more generalizable and scalable than traditional lightning detection models. The image-based lightning detection model just by Camera captures pictures to detect lightning. The images in a video can be divided into foreground and background, and lightning can be understood as the foreground of the video. The lightning images are separated from the video by judging the lightning foreground in the video with time index positioning. In order to quickly identify lightning from video, it is necessary to analyze lightning in terms of its physical properties and its digital image properties. Through the connection between physical properties and image properties, we propose the Adaptive Lightning detection for the task of background subtraction (ALD) algorithm by combining the properties of Background subtraction method and Mixture Gaussian method and adding Gaussian filter to remove noise. We used this method to analyze a total of ten hours of lightning video and compared it with the previously proposed algorithms.