Research On Key Technologies Of Forest Fire-Extinguishing Target Detection System Based On Dual-Light Images

Fire helicopter as one of the commonly used means of forest fire extinguishing,in the hang bucket fire operations mainly rely on the pilot naked eye observation and through their own experience and the location of the fire,but due to the forest fire environment complex,the pilot in the judgment easily affected by the external environment,such as: smoke shielding,burning boundary is difficult to determine and so on.These disturbances will lead to the pilot’s great deviation in the judgment of sprinkler timing,which will lead to problems such as empty sprinkling and missed sprinkling,which will affect the efficiency of helicopter bucket fire fighting operation and even endanger the safety of the aircraft itself.In purpose of solving the mentioned issues,this thesis presents an assisted forest fire suppression monitoring system based on dual-light images.The information of the infrared and visible images of the fire is collected by the built dual vision system,and uses the image registration and fusion technology to process and analyze it,so as to help the pilot quickly obtain the information about the edge,area and combustion degree of the fire site,and assist them in fire fighting operations.In the process of research,this thesis first calibrates the selected infrared and visible light cameras,and completes the construction of the binocular system and the image acquisition work.Then,the collected infrared and visible light images were preprocessed,including image correction and noise reduction,the foundation for subsequent image alignment and fusion is laid.Then the different principles of infrared and visible image imaging are addressed,resulting in the image matching feature matching difficulties,error rate,this thesis adopts a double-light image matching based on the edge contour feature algorithm,the algorithm using visible and infrared image of common edge contour features,to match the edges detected by the aligned image,and finally complete the feature points for infrared and visible image alignment.The experiment shows that the algorithm feature point matching accuracy is high,the infrared and visible image matching effect is good,and the infrared and visible image with the same field of view can be obtained.Finally,an improved color image fusion method based on non-subsampled contour wave transform(NSCT)is used in this thesis.The method combines NSCT and HSI transform to retain the color information in the visible image by HSI transform,and NSCT transform to achieve the fusion of useful information in the infrared and visible images of the fire scene,while using the pulse-coupled neural network model as the high-frequency fusion rule and the adaptive fuzzy logic algorithm as the low-frequency fusion rule to obtain higher quality color fusion images.And the experimental results are analyzed using the corresponding evaluation indexes,which show that the fused images of this paper’s algorithm better retain the color,edge and brightness information in the infrared and visible images,with good visual effects and meet the expected requirements.