Passive Measurement Method For Multiple Trees Attributes Based On Smartphone

The tree height,diameter at breast height and crown diameter are important measurement attributes in forest resource survey and management,which can be used to evaluate soil quality and forest growth status.At present,the measurement methods of tree attributes still have numerous of shortcomings such as low measurement efficiency,labor intensive and expensive equipment.Therefore,in this paper,we focus on the passive measurement method of tree attributes based on monocular vision camera of smartphone.We explore the use of camera calibration model,machine vision technology and close-range photogrammetry technology to study the non-contact passive measurement method of tree attributes.It mainly includes the following work:(1)A multi-dimensional feature adaptive MeanShift algorithm is used to segment the trees contour and extract the feature points of trees measurements.At the same time,an improved camera calibration model with nonlinear distortion terms is used to derive camera intrinsic parameters and nonlinear distortion parameters.Thus,it can be used to obtain more accurate feature point pixel values.(2)According to the principle of camera imaging system,we build a depth extraction model for calculating the depth of each tree in the image.(3)To calculate the height,diameter at breast height(DBH)and crown diameter of trees in the image,we established an adaptive feature coordinate system for multiple trees measurements,which can help us to calculate the rotation matrix and the translation matrix,and then study the conversion relationship and rigid body motion between the coordinate systems.(4)We construct the tree attribute measurement model,and substitute the pixel value of the tree attribute measurement feature point into the model to realize the measurement of tree DBH,height and crown width based on the smartphone.This paper propose a measurement method of the DBH,tree height and crown diameter based on the smartphone.Experimental results show that ranging by depth extraction model has a higher accuracy compare with others based on binocular vision system.The relative errors of the depth measurement is less than 2.64% when the distance is within 3 m.When it is 3 ~ 10 m,the biggest relative error is 5.01%.For trees with DBH values less than 20 cm,the absolute error is < 0.237 cm;for larger trees,the relative error is less than 1.272%.The relative errors of tree height measurements were less than 4.65%(MRE=2.414%);for crown diameter model,the relative error were less than9.6%(MRE=6.067%).Overall,the accuracy of this method falls within the requirements of the continuous inventory of Chinese national forest resources and provides a fast and automated measurement for tree height and crown diameter in areas of forest inventory.