基于机器学习的烟雾检测算法去除固定干扰

Remove of Fix interference Based on Machine Learning in Smoke Detection

烟雾是早期火灾最为突出的视觉表现,因此检测火灾烟雾在日常防火应用中具有重要的意义;针对目前视频烟雾检测算法中误报率高,适应性差等特点,在实施基本烟雾检测算法并在其基础上提出基于机器学习的视频烟雾去干扰方法,能很大程度上去除固定干扰物的干扰,提高了烟雾识别的正确率;实验证明,该算法可以较好、较快地检测出烟雾,并做出早期的预警工作,且具有检测精确等优点,很方便在现实中推广使用。

Smoke is the most prominent of the fire detect, so detect the smoke has an important significance in daily lives. In view of the current smoke detection has the high error rate and poor adaptability. Therefore, this paper presents Remove the fix interference which based on machine learning in the smoke detection and it can remove the fixed interference as soon as possible, and it can improve the correct rate of smoke recognition. Experiments show that the algorithm has a high speed and good effect on the smoke detection and make the early alarm work, And has accurate detection etc, so it＇ s convenient to be widely used in reality.