An image processing method for an explosion field fireball based on edge recursion

Abstract

To overcome the limitations caused by the complex fireball geometry in an explosion field and the smoke interference of fireball contours, a pre-processing strategy for fireball images in an explosion field, combining threshold segmentation, rectangular filtering, nearest-neighbor clustering and morphological processing, as well as an edge-recursive fireball image segmentation algorithm based on the polar coordinate system, is proposed. The overlap rate was used to compare and evaluate the segmentation effect of the proposed method and traditional image-processing methods on the measured explosion fireball image. The results demonstrate that the edge-recursive fireball image segmentation algorithm based on the polar coordinate system is excellent for identifying fireball image segmentation and fireball damage contour recognition under different smoke-obscured conditions in an explosion field, thereby facilitating the accurate recognition and measurement of the fireball image and contour under a smoke plume. This method improves the accuracy of measuring explosive fireball damage power and provides a method for effectively extracting the distribution information of explosive fireball damage.