Design And Development Of Image Stitching System For Forest Fire Monitoring

Forest fires are extremely harmful and seriously threaten people’s lives and property.The current forest fire early warning system is relatively sensitive to temperature and will immediately report an alarm when a fire is detected,so it will cause a high false alarm rate.Due to the complexity of the forest scene,there are factors such as changes in lighting and interference,and then it is difficult to meet the actual needs of fire detection based on the thermal infrared principle.This paper introduces the method of combining panoramic image mosaic and mosaic image rectangle,and designs an image mosaic system for forest fire monitoring.The generated panoramic image is used as the context and the background comparison is used to complete the detection of the suspected fire area,thereby reducing the forest.The high false alarm rate of the fire early warning system is as follows:(1)To meet the needs of forest fire monitoring,a panoramic stitching method for forest scenes is designed.The integrity of the forest image directly affects the accuracy of subsequent fire monitoring.In order to obtain a complete forest panoramic image,image pre-processing is first performed to remove the annotation information in the image,and then the commonly used feature point extraction algorithms are compared and analyzed to select the one suitable for the forest scene.algorithm.In the process of image deformation processing,the modification of parameters is used to eliminate the effect of parallax to make it more suitable for forest scenes.Then,for the mismatch of image registration,a Bayesian model with robust features is used to improve the matching accuracy.Finally,several images are fused.The algorithm performs experimental analysis and selects an algorithm that adapts to the forest scene to eliminate stitching traces.Experiments have shown that the generated panoramic forest image has a wide field of view and a large angle of view,clear and natural,and no stitching traces.(2)Aiming at the requirement of panoramic stitching of forest scenes,the effect of global optimization of the original energy function is improved by introducing a shape parameter.In this paper,using the generated forest panoramic image for rectangular processing can better restore the forest scene and eliminate as many suspected fire areas as possible.After acquiring the irregular forest panoramic image,first use the Seam Carving algorithm to rectangularize the irregular panoramic image,but the processed panoramic image will stretch and deform in local areas.Aiming at this problem,the idea of global optimization is adopted,and the shape parameters are introduced on the basis of the original energy function to make it more suitable for the rectangular output of the forest scene.Through experimental verification,the final panoramic image output conforms to the characteristics of the human visual system,which can complete the verification of the suspected fire area and improve the accuracy of forest fire detection.(3)Based on the research of image mosaic algorithm,a forest fire monitoring system was designed and developed on the basis of the above work.This system is mainly applied to specific forest scenes,which can reduce the false alarm rate of forest fires and has a large Practical application value.