Research And Implementation Of Automatic Detection Algorithm For Forest Land Dynamic Change Based On Multi-source Remote Sensing Image

The forest is not only an important natural resource,but also the terrestrial ecosystem with the most extensive area distribution and the most complicated structure on the earth.With the rapid development of remote sensing technology,the combination of smart forestry and remote sensing technology has enabled forestry to develop better.In view of this,the rational use of multi-source remote sensing raster data and forestry vector data by forestry workers is of certain necessity and importance for the accurate detection of forest land change information.How to use the rich spectrum and spatial resolution of multi-source remote sensing images to effectively extract the forest land dynamic change detection information,draw out the forest land change information with the help of computer software development technology,and solve the problems of misjudgment and low efficiency caused by manual interpretation of subjectivity Become the key.This article has completed the research in the following aspects around the current process and actual work of forestry work:(1)Combining the difficulties encountered in the forest land dynamic change detection work at this stage,this paper focuses on the forest land dynamic change detection algorithm based on multi-source remote sensing images,using GF-1 remote sensing images and Landsat8 remote sensing images,combined with field survey of forests Supervise the manual interpretation of the data of the second class of survey maps and forest resources,integrate the seasonal characteristics and phenological characteristics of the study area,through the setting of different cluster window parameters and change thresholds,use the forestry auxiliary vector data to filter the filter conditions and extract the two The forest cand change characteristics of remote sensing images during the period,using multi-source remote sensing images to extract change information.(2)To study the relevant theories and algorithms for the extraction of forest land dynamic change information from different data sources,extract and plot the suspected change maps in the detection area and their spatial location distribution with high precision,aiming to study an economic,efficient and automated forest land The dynamic change information extraction method is intended to provide a reference for the government to quickly,accurately and timely grasp the status of forest land change information,assist the current complex and cumbersome manual interpretation work,improve the work efficiency of forestry personnel in China,and thus promote the development of smart forestry.(3)Based on multi-source remote sensing images,research an algorithm that can automatically extract the maps of forest land dynamic changes,and implement the algorithm in conjunction with the forest land dynamic change detection system.According to the spectral characteristics of the images of different remote sensing information sources,a normalized difference vegetation change index(NDVCI)was proposed.The experimental study was carried out in Pu County,Linfen City,Shanxi Province.The experimental results show that the algorithm can effectively improve the dynamic changes of forest land The accuracy of detection is 84%accuracy,88.04%accuracy,82.65%sensitivity and 85.71%specificity,which are better than the difference NDVI threshold classification algorithm with 76%accuracy,82.76%accuracy,72.73%sensitivity and specificity.Degree 80.26%.With the help of the forestland dynamic change detection system,the use of NDVCI threshold classification algorithm can reduce the workload of manual visual interpretation by 60%-70%,effectively assisting manual interpretation.In this paper,the experimental area of Pu County,Linfen City,Shanxi Province has a total land area of 151,100 square kilometers.The forest land dynamic change detection system and remote sensing technology can be used to extract the forest land change information in only two hours,which can quickly and efficiently promote the development of smart forestry.To sum up,the algorithm and development system platform studied in this paper have important reference value for the future normal forestland dynamic change detection and the updating of a forest map database.