| In the forest area,the terrain is obviously undulating.Due to the existence of terrain,the direct light and scattered light received from different slope and slope are obviously different,which leads to the different brightness of remote sensing image(Zhaoli Wang,2015).In the application of forestry remote sensing,this phenomenon seriously affects the accuracy of remote sensing image extraction,so it is necessary to carry out terrain correction.At the same time,the experimental area of this experiment is in Tianfeng Mountain Yunnan Province where has very serious plant diseases and insect pests.So it’s necessary to deal with this problem.Based on these two points,we can consider combining terrain correction with forest pest and disease information extraction.In this experiment,we use the method of field measurement and indoor satellite image processing.On the one hand,we use the three-dimensional model to explore the effect of different resolution DEM on the correetion effect,and use ZY-3 satellite data to verify the target.Visual judgment,DN Histogram,NDVI provide a method to evaluate the accuracy of terrain effect correction and try to answer the best DEM resolution.On the other hand,through the combination of terrain correction and pest classification,The maximum likelihood method is used to analyze the image classification results before and after terrain correction.The effect of terrain correction on the extraction of pest information is analyzed by visual judgment,mixed matrix and classification statistics.In this experiment,there are two innovations:one is three-dimensional model for terrain correction simulation analysis and the use of ZY-3 satellite data validation;the other is terrain correction and pest classification to explore the terrain correction of pest information extraction.Among these common resolutions(2.5m,5m,10m,15m,30m),5m resolution is the best corrected resolution.The accuracy of classification before calibration is 36.84%,and the post-calibration classification is(From 3.95%to 6.83%),the proportion of healthy forest decreased(from 29.08%to 7.28%),and the proportion of heavy forest was increased(from 3.95%to 6.83%).It can be concluded that the correction effect is the best when the DEM resolution is 5m among thees common DEM resolutions.The terrain correction has an effect on the extraction of pest information,which is mainly reflected in the improvement of the classification accuracy of the pest.The results of this study has a certain referring value and promoting effect. |
