Segmentation Of Fire Gun Jet Trajectory Based On Superpixel Random Walk

Automatic positioning fire water cannon system can automatically track and locate the fire site in the early fire,and automatically sprinkle,is an effective measure to deal with large space fire.However,in a large-scale outdoor fire scene,uncertain factors such as environmental changes and changes in the pressure of the water cannon will cause the accuracy of the water cannon to strike the fire target to decrease.Therefore,capturing and extracting the water cannon jet trajectory to form an effective closed-loop control has strong practical significance and application value.This paper presents an image segmentation method for water cannon jet trajectory.Due to the characteristics of uncertain illumination and complex background in natural scenes,the traditional image segmentation algorithm appears the phenomenon that the target boundary is not suitable or even segmentation error when dealing with the water cannon jet trajectory image.The image segmentation algorithm based on the random walk model is a semi-automatic segmentation method guided by manual marking.Compared with traditional image segmentation methods,it can achieve more accurate water cannon jet trajectory image segmentation.In this paper,the following research work is done on the shortcomings of the random walk model in the application of fire water monitor jet trajectory image segmentation :(1)The idea of super-pixel is introduced into the random walk model,and the super-pixel is used to replace the original pixel as the node of graph structure,which expands the minimum calculation unit of random walk and greatly improves the segmentation efficiency.At the same time,compared to the original pixels,super pixels have richer features,which improves the segmentation accuracy to a certain extent.(2)In order to solve the problem of automatic selection of mark points in random walk algorithm and realize automatic segmentation of water cannon jet trajectory image,an automatic segmentation algorithm of super-pixel random walk water cannon jet trajectory combined with saliency detection and pattern mining algorithm is proposed.First,extract the saliency map of the image,initially determine the target area of interest,that is,the jet trajectory part of the water cannon,then preliminarily select the mark points through the pattern mining algorithm,and then combine the saliency target position information to further classify the mark points into foreground mark points and background mark points.Finally,the superpixel random walk algorithm is used to achieve segmentation based on the mark points.Experimental results show that the proposed algorithm can automatically segment the jet trajectory accurately.