Extraction Of Individual Tree Information Based On High Spatial Resolution Remote Sensing Images

In recent years,remote sensing technology have been widely used in the application of the agricultural environmental monitoring field.Forest is one of the basic material conditions of the agricultural environment,and also play an important role in agricultural production.In view of the traditional method of forest land information extraction has the defects of the time-consuming,heavy workload,low efficiency and timeliness and difficult to meet the monitoring requirements and so on,this paper based on the high spatial resolution unmanned aerial vehicle remote sensing images and the high spatial resolution satellite remote sensing image,through choose the artificial forest(including Pinus sylvestris forest and Pinus tabulaeformis forest)of eco-rehabilitation in the Daliuta coal mine which located in Shendong coal mining area as the study area,and conduct the detection of the individual tree and the delineation of the individual tree crown in the plantation area.To explore the feasibility and effectiveness of individual tree information extraction based on the high spatial resolution images from different remote sensing platforms.In order to provide technical support and scientific basis for agricultural environment protection.The main research contents and conclusions are as follows:(1)Based on the high spatial resolution UAV remote sensing images,a novel individual tree information extraction method based on the theory of image segmentation was proposed in this paper.This method taking full advantage of the characteristics of UAV remote sensing image,combining the characteristics of image spectrum and the spatial correlation between pixels,using the principle of image spectral enhancement,pixel clustering and image segmentation,conduct the detection of the individual tree and the delineation of the individual tree crown.Through the contrast and validate by the result of visual interpretation,the results show that the overall accuracy of the individual tree detection of the Pinus sylvestris forest and the Pinus tabulaeformis forest are 89.52%and 95.65%,and the producer’s accuracy and user accuracy of individual tree crown extraction of the Pinus sylvestris forest are 95.65%and 94.29%,the producer’s accuracy and user accuracy of individual tree crown extraction of the Pinus sylvestris forest are 81.90%and 83.50%,which has the advantages of applicability.In addition,the proposed method does not require a large number of manual intervention and prior knowledge input,greatly improving the degree of automation of the method.In order to verify the segmentation effect of this method,the multi-scale segmentation(Multiresolution Segmentation)method was used to segment the image.Through comparing the results,it shows that the extraction accuracy of the proposed method is higher,and the extraction effect of the tree crown outline is closer to the real situation.(2)Based on the high spatial resolution satellite remote sensing images,this paper use the WorldView-03 satellite remote sensing image of the Pinus tabulaeformis forest,using the local maximum method and marked watershed segmentation algorithm for individual tree detection and individual tree crown extraction.Through the comparison of the accuracy and analysis,the result shows that the overall accuracy,the commission error and the omission error of the individual tree detection are 95.45%,4.55%and 23.36%,respectively.And the producer’s accuracy and user accuracy of individual tree crown extraction of the Pinus sylvestris forest are 92.73%and 74.45%.In conclusion,it is feasible and effective for using the high spatial resolution UAV remote sensing images and satellite remote sensing image to extract the individual tree information.In view of the high spatial resolution UAV remote sensing images,the individual tree information extraction method based on the image segmentation theory which proposed in this paper can be applied in small scale area to extract more precision and rapid the forest information in the process of field forest monitoring and investigation.Based on the high spatial resolution satellite remote sensing image,it is also have a high accuracy when using the local maximum filtering and the marked watershed method to conduct the detection of individual tree and the delineation of individual tree crown,respectively.And can also extract the forest information effectively.