摘要
提出一种参数自适应的差分进化算法(Adaptive differential evolution,ADE)拟合圆的方法对树干胸径进行测量。采用实验室自制胸径测量平台采集树干胸径数据,对采集到的数据用K-means算法进行聚类,分割出背景与目标。为提高测量精度,采用角度补偿法对树干表面点到激光传感器中心的偏角进行补偿。对补偿后的数据,以激光传感器中心为坐标原点,计算出树干表面点横纵坐标。最后对树干表面坐标点采用自适应差分进化算法拟合圆,拟合过程中,差分进化过程的种群数、进化策略、变异尺度以及交叉概率随着进化过程自适应变化。分别对校园内与人工林地各40棵不同大小的树进行了测量与拟合,其拟合时间为1.41 s,拟合结果与实际值比较分析,RMSE分别为4.996 mm与4.500 mm。并与Levenberg-Marquardt(LM)算法、粒子群优化算法(Particle swarm optimization,PSO)以及Hough变换拟合圆法测量树干胸径进行了分析与比较,实验结果表明所提方法能更有效准确地测量出树干胸径。
A parameter adaptive differential evolution algorithm was proposed to measure the diameter at breast height（DBH） of tree trunk by fitting circle. A DBH measuring platform made by laboratory was utilized to collect the data,and the K-means algorithm was first introduced to cluster the data and segment the background and target. In order to improve the fitting accuracy,the angle compensation method used arc length was employed to compensate the deflection angle of the trunk surface point to the laser sensor center before fitting. Then,the center of laser sensor was taken as the coordinate origin,and the horizontal and vertical coordinates of the trunk surface points were calculated. Finally,the adaptive differential evolution algorithm was proposed to fit the tree trunk surface. In the process of fitting,the number of population,evolutionary strategy,scaling factor and crossover rate of the differential evolution process were adaptively changed with the evolution process. Totally forty different sizes of trees in the campus and artificial forest were measured and fitted respectively,and the fitting time was 1. 41 s. The results of the proposed method were compared with the real values of DBH. The RMSE for DBH of trees in the campus and artificial forest were 4. 996 mm and 4. 500 mm,respectively. At the same time,the results were compared with the Levenberg-Marquardt（LM）,particle swarm optimization（PSO） and Hough transform. The experiments validated that the method proposed can measure the DBH more effectively and accurately.
作者
胡春华
李萍萍
金成磊
朱乐
HU Chunhua;LI Pingping;JIN Chenglei;ZHU Le(College of Information Science and Technology,Nanjing Forestry University,Nanjing 210037,China;College of Biology and Environment,Nanjing Forestry University,Nanjing 210037,China)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2018年第9期183-188,共6页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家重点研发计划项目(2017YFD0600905-1)
江苏高校优势学科建设工程项目(PAPD)
关键词
单木胸径
差分进化算法
激光传感器
参数自适应
individual tree diameter at breast height
differential evolution algorithm
laser sensor
parameter adaptive
