A method of image segmentation of industrial fumes based on background modeling and feature matching is proposed

aiming at problem that industrial fumes is easily affected by motion interferent in background and the influence of different environments in image segmentation process. Construct accurate background image by setting up background model and updating the background timely. Figuring out fumes dynamic region by accumulating differences and use morphological filling method to get the coarse fumes region

through feature matching method to remove interference and get complete fumes segmentation region. Using the proposed method experiment is carried out through the actual steel industry fumes emissions video

and contrast test with other segmentation methods is carried out. It turns out that the proposed method can distinguish background interferent and fumes and the fumes region can be segmented completely. Besides

it is applicable as well as jamming-free

which also indicates that the proposed method has good practicability.