基于双目视觉系统的幼龄格木生长因子测定  

Measurement of growth factors of juvenile Erythrophleum fordii based on a binocular vision system

作　　者：王鹏 王雪峰[1,2] 赵溪月 WANG Peng;WANG Xuefeng;ZHAO Xiyue(Institute of Forest Resource Information Techniques,Chinese Academy of Forestry,Beijing 100091,China;Key Laboratory of Forest Management and Growth Simulation,National Forestry and Grassland Administration,Beijing 100091,China;Jilin Forestry Science Research Institute,Jilin,Jilin 132000,China)

机构地区：[1]中国林业科学研究院资源信息研究所,北京100091 [2]国家林业和草原局森林经营与生长模拟重点实验室,北京100091 [3]吉林市林业科学研究院,吉林吉林132000

出　　处：《森林与环境学报》2024年第6期628-638,共11页Journal of Forest and Environment

基　　金：国家自然科学基金面上项目“林木对养分与水分需求的机器理解法”(32071761);中央级公益性科研院所基本科研业务费专项资金项目“珍贵树种健康状态的精准图像判定技术”(CAFYBB2021ZB002)。

摘　　要：为实现幼木参数的无损、高效测定,以3年生幼龄格木(Erythrophleum fordii)为研究对象,对幼木参数的双目测定方法进行研究。先用RealSense D415双目相机获取52株格木的侧面图像对,并采用伽马校正与限制对比度直方图均衡化(CLAHE)算法对图像进行预处理操作;再通过改进半全局块匹配(SGBM)算法对左右视图进行匹配,根据三角测量原理对像素点进行空间映射,生成三维点云数据;最后,通过提取点云中关键点的坐标信息,实现对格木树高、冠幅与地径的测量。结果表明:预处理后图像对比度显著增强,不同灰度区域的过渡更加平滑,视差计算更加准确,匹配精准度得到提升。改进的SGBM算法较传统SGBM算法具有更优的匹配效果,获取的视差图更加平滑,视差连续性更强,得到的点云密度更高且噪点较少,对树高、冠幅、地径的测量误差更小,平均相对误差(Emr)分别为1.981%、2.459%、2.942%,平均绝对误差(Ema)分别为1.492 cm、1.567 cm、0.044 cm,均方根误差(Erms)分别为1.843 cm、1.914 cm、0.060 cm。总体而言,基于改进的SGBM算法的双目测定方法显著提升了幼龄格木参数的测定精度。To achieve non-destructive and efficient identification of juvenile trees,the binocular measurement method of sapling parameters was studied using 3-year-old juvenile Erythrophleum fordii trees as the research object.Side image pairs of 52 E.fordii were obtained using a RealSense D415 binocular camera.The images were preprocessed using gamma correction and limited contrast histogram equalization(CLAHE)algorithm.Then,the left and right views were matched using an improved semi-global block matching(SGBM)algorithm.The pixels were spatially mapped according to the triangulation principle to generate three-dimensional point cloud data.By extracting the coordinate information of the key points in the point cloud,the height,crown width,and ground diameter of E.fordii were measured.The results show that the image′s contrast was significantly enhanced after preprocessing.The transition of different gray areas is smoother,which makes the disparity calculation more accurate and the matching accuracy improved.The improved SGBM algorithm had a better matching effect than the traditional SGBM algorithm.The obtained disparity map was smoother and the disparity continuity was stronger.The obtained point cloud density was higher and there was less noise.The measurement error from tree height,crown width,and ground diameter was smaller.The mean relative errors(E mr)were 1.981%,2.459%,and 2.942%,respectively.The mean absolute errors(E ma)were 1.492 cm,1.567 cm,and 0.044 cm,respectively.The root mean square errors(E rms)were 1.843 cm,1.914 cm,and 0.060 cm,respectively.In general,the binocular measurement method based on the improved SGBM algorithm significantly improved the measurement accuracy of juvenile E.fordii parameters,providing technical support for the precise cultivation and management of juvenile trees.

关 键 词：格木 生长因子 双目重建 立体匹配 三维点云 

分 类 号：S758.1[农业科学—森林经理学]