Extraction Of Tree Crown Parameters And Model Developmentbased On UAV Image

Crown width is an important characteristic factor of canopy structure,which directly affects the productivity and vitality of trees.The forest canopy density is one of the important indexes to reflect forest canopy structure and density and to evaluate forest management and logging intensity.Given advantages of Unmanned Aerial Vehicle(UAV)that it can fly under cloud and is easy to obtain images with high accuracy and low cost,this paper discovers approaches that can effectively use them to extract canopy index.When these approaches are systematized,the results of accurate and efficient monitoring and detection of forest resources are thereby achieved.This paper takes Plum Blossom Valley,located in Jiangle Forest Farm,Fujian,as the experiment area.By using UAV image as the remote sensing data source,the high-precious Digital Orthophoto Map(DOM)is produced through PIX4 D mapper application.Combined with some of the ground measurement data,using methods of multi-scale segmentation technique and watershed algorithm separately can achieve the extraction of individual tree's canopy.After comparing these two methods in terms of their accuracy that extracts the canopy of high-density Chinese fir plantation,this paper chooses the high-accuracy method to estimate the canopy density and construct the model of aviation standing volume,The key findings are as follows:1.UAV satisfies the demand of high degree of accuracy when extracting the forest index.The combination of UAV and high-resolution camera can easily obtain a range of forest data very quickly and meet the requirements of aerial photogrammetry mapping.Use 64 high resolution images acquired by UAV.In the UAV image processing platform-PIX4 D mapper,generate high precision DOM.2.Under the condition of high-density plantation(the canopy density is 0.7785),the multi-scale segmentation technique achieves higher segmentation effect than watershed algorithm.Watershed algorithm can't segment effectively,multi-scale segmentation technique has better effect.The R2 of extracted area of canopy and actual area of canopy achieves 0.8068.3.According to the sample data collected in the experiment area,the relevance analysis of DBH,volume,canopy area was conducted.There are two basic models: one is diameter of Chinese fir-volume model,and another is canopy area – diameter model,then the final model of volume-canopy area model is produced through simultaneous equations.Using this model for estimating the timber accumulation of 112 independent samples of fir trees outside the modeling,the average accuracy was 67.75% compared to the results calculated using the local fir standard volume table.In conclusion,the UAV aerial image can be used in the forest research,the object-oriented multi-scale segmentation technique can effectively extract the canopy area of an individual tree from the produced Orthphoto Map,so that the overall accuracy of canopy density is above 75%,establish the volume model of individual Chinese fir single volume table.This research provides contributions to scholarship of using high-resolution UAV image to extract the canopy area of individual Chinse fir and estimating the amount of growing stock.This research provides new approach for forest resource investigation and monitoring.