CAJ | 학술논문

烟雾的检测在现代智能消防中有着重要的应用前景。针对当前烟雾检测算法中存在的易受疑似烟雾物体的干扰，对稀薄烟雾检测灵敏度不高的问题，提出一种基于暗通道优先规律的烟雾检测算法。首先，算法根据图像暗通道优先规律和烟雾形成的传递函数，去除图像中烟雾的成分；然后，对去烟雾前后图像进行差分并根据一定门限做二值化处理，由此得到烟雾的疑似区域；最后，根据烟雾的飘动特点，对二值化图像进行背景建模和前景检测得到烟雾区域。经过实验证明，该算法对于稀薄烟雾的检测灵敏度高。
연무적검측재현대지능소방중유착중요적응용전경。침대당전연무검측산법중존재적역수의사연무물체적간우，대희박연무검측령민도불고적문제，제출일충기우암통도우선규률적연무검측산법。수선，산법근거도상암통도우선규률화연무형성적전체함수，거제도상중연무적성분；연후，대거연무전후도상진행차분병근거일정문한주이치화처리，유차득도연무적의사구역；최후，근거연무적표동특점，대이치화도상진행배경건모화전경검측득도연무구역。경과실험증명，해산법대우희박연무적검측령민도고。
Smoke detection has important applied prospects in modern intelligent firefighting.Existing smoke detection algorithm is susceptible to the interference from suspected smoke object, and insensitive to thin smoke detection.Aiming at these problems, a smoke detection algorithm based on dark channel prior rule is proposed.According to dark channel prior rule and transfer function formed by the smoke, the algorithm first removes the smoke components in the image;Then, it performs differential on the images before and after smoke removal, and carries out binarisation on the image according to certain threshold, thereby obtains the suspected smoke area; Finally, according to the characteristics of smoke flowing, it obtains the smoke area by background modelling and foreground detection with binary image.It is proved by the experiments that the algorithm has high sensitivity for thin smoke detection.