Lightning Image Segmentation Based on Contour Detection and Curvature Analysis

Lightning, as a powerful natural discharge phenomenon, is harmful to human production and life, natural ecology, railroad transportation, etc. The study of lightning is of great significance, and image segmentation is the key link in lightning research. In this paper, a method of lightning image segmentation based on contour detection and curvature analysis is proposed. This method can accurately segment the lightning channel and locate the coordinates of the lightning strike point, which is of great significance for the in-depth study of lightning. Meanwhile, this project collects lightning images from multiple sources to construct a dataset. It realizes lightning channel extraction and strike point localization through the steps of image preprocessing, thresholding and contour detection, curvature analysis, and lightning strike point recognition. The experimental results show that the method is effective in single-branch and multi-branch lightning and cloud flash image segmentation, with obvious advantages over traditional methods, and the strike point recognition algorithm has high reliability.