基于改进共生生物搜索算法的植物冠层图像分割研究  

PLANT CANOPY IMAGE SEGMENTATION BASED ON IMPROVED SYMBIOTIC ORGANISMS SEARCH ALGORITHM

作　　者：王帅[1] 贾鹤鸣[2] Wang Shuai;Jia Heming(Department of Basic Education and Research,Changchun Guanghua University,Changchun 130033,Jilin,China;College of Mechanical and Electrical Engineering,Northeast Forestry University,Harbin 150040,Heilongjiang,China)

机构地区：[1]长春光华学院基础教研部,吉林长春130033 [2]东北林业大学机电工程学院,黑龙江哈尔滨150040

出　　处：《计算机应用与软件》2020年第9期152-159,182,共9页Computer Applications and Software

基　　金：中央高校基本科研业务费专项资金项目(2572019BF04);黑龙江省教育科学“十三五”规划重点课题(GJB1319007);东北林业大学横向课题项目(43217002,43217005,43219002)。

摘　　要：针对植物冠层图像背景复杂干扰目标多,经典Kapur熵阈值化技术效率低的问题,利用改进共生生物搜索算法(Symbiotic Organisms Search,SOS)对最佳阈值组合的选取过程进行优化,有效消除局部最优,进一步利用莱维飞行与自适应权重因子提高算法的优化能力。通过对植物冠层进行分割实验,结果证明,该算法可以获得准确的分割阈值,进而精确地分析植物冠层生长面积与其基因组学间的关系。Aiming at the problem that the background of plant canopy image interferes with many complex targets and the efficiency of classical Kapur entropy thresholding technology is low,this paper optimizes the selection process of the optimal threshold combination by using the improved symbiotic organisms search(SOS),which can effectively eliminate the local optimum,and further improves the optimization performance of the algorithm by using Levy flight and adaptive weight factor.Through segmentation experiments on the plant canopy,it is verified that our algorithm can obtain accurate segmentation thresholds,and then accurately analyze the relationship between the growth area of the plant canopy and its genomics.

关 键 词：植物冠层 多阈值图像分割 共生生物搜索 莱维飞行 自适应权重 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]