Remote Sensing Monitoring Of Changes Of Forest Resources In Jiangle County Based On Small Class Objects

Forest resources are the mater:ial basis of forestry development,The investigation of changes of forest resources and the timely updating of forest resources databases are of great significance to the development of forestry and the formulation of forestry policies.In this paper,taking Jiangle County as an example,GF-1 and ZY-3 images,comb:ined with 1:10000 subcompartment data were used.Object oriented images analysis method was applied to research on monitoring the changes of forest resources and updateing forest subcompartment database.The main research contents and conclusions of this paper are as follows:(1)Construction of monitoring system for remote sensing changes of forest resources based on small class objects.It mainly involved data preprocessing,multi-level segmentation,feature selection,change information extraction.Multi-scale segmentation of preprocessed images,the ESP algorithm is used to analyze the images in the study area.TThe results show that the ESP algorithm can determine the best segmentation scale quickly in the study area.Using the mRMR feature selection algorithm to select the optimal feature subset,The SVM classifier was used to classify the change objects,determine the change types of each small class obj ect,and the accuracy of the classification results was evaluated.(2)By combining the characteristics of multi-temporal images,the classification accuracy of the change areas of small class objects can be effectively improved.Through the extraction of features of multi-temporal remote sensing imagery,the classification of the changes in small class objects was performed.The classification accuracy increased by 9.3%,which reduced the number of mixed woodland forest lands to unforested afforestation lands and felled lands.(3)Using the Monitoring System Constructed to Monitor the Change of Forest Resources in 13 Townships in Jiangle County.Use field ver:ification,cutting quotas,and other data to verify the accuracy of the results of the changes.The results show that the overall accuracy of remote sensing changes in forest resources in Jiangle County can reach 96%,which meets the requirements of the forestry resource management department.A small amount of manual correction of the results,Obtained monitoring results of changes in Jiangle County from 2014 to 2016,and updated the original subclass data.(4)Counting the area of each type of change in Jiangle County.The results of the study show that:From 2014 to 2016,the overall change area of Jiangle County was 18247.41 hectares,The increase in forest land was mainly due to the conversion of unforested afforestation land to forest land,which reached 5,540.37 hectares,accounting for 53%of the total area of change;The reduction in forest land is mainly due to the change of forest land to cutting forestland,which reached 3,479.25 hectares,accounting for 18%of the total area of change.The forest coverage of 13 townships in Jiangle County has been increasing from 2014 to 2016,and the forest coverage rate increased by 4.25%in Jiangle County,reached 83.25%.