The Tree DBH Measurement Method Based On Smartphone And Machine Vision Technology

This paper analysed the key problems of the existing DBH(Diameter at Breast Height)measurement methods.We based on the smart phone,using the techniques of camera calibration,machine vision and image processing,researching an intelligent non-contact DBH measurement method.The main tasks are as follows,(1)With the structural characteristics of the smart phone’s camera,we designed the model and algorithm of smart phones’ camera calibration and distortion correction.And calculating the internal and external parameters by the reference of Zhang Zhengyou’s calibration method.(2)With the camera model,reconstructe the physical size of the unit pixel in the 3D world coordinates under different distance conditions.(3)With the image features of tree in natural environment,we designed the tree contour detection and segmentation algorithm.(4)Using the improved algorithm of minimum bounding rectangle to obtained the minimum circumscribed rectangle,and determine the height and DBH direction.(5)Building the height positioning algorithm,and extract the height of 1.30 meters diameter,with the result(2)to calculate the DBH.In this paper,we designed the DBH measurement experiments based on Android platform,which has widest uses and good campatibility.In the calibration experiments of accuracy and efficiency verification,we finded that the average reprojection error is 0.12 pixel,and the iteration time of the internal and external parameters is 0.063s(0.234 s at the smartphones),which has higher accuracy and efficiency.In the experiments of unit pixel reconstruction precision to 3D world physical size,the relative error can be controlled within 2.5% at the range of 1.0-3.0 meters,which has high accuracy.In the experiments of tree contour detection,we compared with OTSU,K-means and Itti segmentation methods in different environments.The average misclassification error of the proposed method is 6.27%,which can overcome the influence of illumination and complex background noise and effectively segment the contour of standing tree trunk.In the DBH measuring accuracy experiments,the relative error of measurement results were less than 3.20%.The experimental data showed that the proposed method has a high precision,which can be applied to the forestry resources survey.The innovation of this study are as follows,(1)With the camera calibration technology,the improved method eliminating the distortion of camera lens exists,and reducing the system error.(2)With machine vision technology and image processing technology,this method realize the rapid DBH measurement in forestry resources survey.(3)With camera calibration,visual measurement and 3D reconstruction technique,this method Significantly improve the measurement efficiency and measurement accuracy.