Multi-Fire Source Detection Algorithm Using Local Threshold Method In Smoky Circumstances

Multi-Fire Source Detection Algorithm Using Local Threshold Method In Smoky Circumstances

  In this paper, we propose a novel algorithm to detect multi-fire sources from images obtained by a near-IR camera in smoky circumstances. The near-IR camera we used has a spectrum between 400 ㎚ and 1100 ㎚. It can locate an object radiating infrared rays that a CCD camera can not detect. After obtaining an image from the near-IR camera, the image is divided into sub？blocks. A local threshold method based on Otsus algorithm is used to decide the optimal threshold for each sub-block. Finally, the absolute threshold is estimated using local thresholds. According to the absolute threshold, we separate each fire source if there are several fire sources in the image.