Research On Object-oriented Remote Monitoring Of Forest Fire

Forest fires seriously affect our forest ecosystems and the forest fire prevention and monitoring work which is of great practical significance to the protection of forest resources in China.Satellite remote sensing technology has become an indispensable means of global fire monitoring,how to use remote sensing data to achieve more accurate fire point information is a focus of scientific research in recent years.Object oriented remote sensing image fire information analysis technology to break the shackles of fire recognition technology based on the idea of the traditional pixel,fire point object based on image segmentation,can greatly improve the accuracy of information extraction.Appearance of the remote sensing image classification technology information of the object oriented broke the shackles of pixel information recognition technology based on the idea of the traditional image segmentation,it is based on the object information,can greatly improve the accuracy of information extraction,provides new ideas for the research of forest fire points.Based on the support of the forest fire point monitoring and warning project of Central South University of Forestry and Technology and Tianlitai Project,this paper chooses Hunan Province as the research area,explore the forest fire of the object-oriented information extraction method of fire information extraction research on MODIS LIB 1km resolution data.The main results are as follows:(1)The data preprocessing of the Aqua satellite MODIS is achieved.The normalized difference vegetation index(NDVI)data of the image and thermal infrared bands B31,B32,B22,B21 band brightness temperature data are obtained by band analysis and operation.(2)Analyzing the band characteristics and correlation characteristics of bands B1,B2,B3 and B4,the optimal segmentation band combination of segmentation is determined,and the different scale The segmentation effect of the region is the best segmentation scale.,and the multi-scale segmentation of the image of the study area is carried out,and the segmentation object is obvious and the segmentation effect is well.(3)Through the comparative analysis of characteristics of their respective methods of identifying fire three channel method,absolute fire recognition algorithm,context recognition algorithm,the absolute spectral characteristics determine the fire quantization point recognition algorithm of forest fires,and the establishment of the rule set combined with the fire characteristics and fire point data NDVI texture feature.(4)According to the NDVI data,the division of vegetation and non-vegetation was realized,and the information of forest land was extracted.And then use the fire point information knowledge base to extract the information of forest fire,and obtain the vector data of forest fire.(5)From the two aspects of coordinate area and fire point,different object oriented method of fire point and forest fire prevention information network Chinese fire point,through the data analysis of the results,It is found that the fire point is more in line with the actual fire point in the two aspects of the fire point area and fire point coordinates.