Vision-Based Forest Fire Detection Using Machine Learning

Forest fire does great harm to the society. This paper studies the method of forest fire detection combining image processing and machine learning based on video sequences. The method consists of three parts: moving object detection, image feature extraction and classifier recognition. An optimal algorithm combination of the above three parts is found through comparative experiments. In this paper, the LBP feature is improved based on color information. Before the moving object detection, image segmentation step is added in a novel way to reduce the false positive rate. Experimental results show that the proposed method yields good performance.