Early Fire Smoke Image Segmentation in a Complex Large Space

Aiming at the problem that the fire smoke image segmentation algorithm cannot obtain integral smoke information, the adaptive background updating smoke foreground object extraction algorithm based on block segmentation is proposed. According to the characteristics that the smoke internal pixels continuously roll in the heat of the driver and the pixel concentration from near the fire source to edge decreases in turn, the method for different blocks using different difference threshold is adopted to ensure the integrity of the extracted foreground target. Under the difference between the current frame and the background frame, block difference thresholds are updated to let the differential threshold adjustment with continuous adaptive monitoring. In the new weighted average updated background image corrected by the original background image, the interference suspicious target edge pixels are removed. Simulation experimental results show that the method is able to extract more complete information of smoke in a suspected area of the edge, and eliminate the interference of light and pedestrian in complex space.

This creative work has been published under the Creative Commons Attribution 4.0 International (CC-BY 4.0) license which allows copying, and redistribution as well as adaptation of the original work provided appropriate credit is given to the original author and the conditions of the license are met.