文档介绍：
基于图像处理的田间水稻叶瘟病斑分割
方法#
吴露露1,郑志雄1,齐龙1,2,马旭1,2,邝健霞1,陈国锐1**
5
10
15
20
25
30
35
40
(1. 华南农业大学工程学院,广州 510642;
2. 华南农业大学南方农业机械与装备关键技术省部共建教育部重点实验室,广州 510642)
摘要:本文针对病斑在叶片上易形成封闭边缘的特性,采用色度学模型、边缘提取和形态学
等方法对田间稻叶瘟病斑进行分割。利用 2R-G 色差分量提取图像上的图斑;采用 Canny 算
法对 2R-G 色差分量图斑进行边缘检测,通过自定义边缘封闭修复模版对边缘进行修复;利
用 HIS 模型的 H 分量提取的叶片正常部位信息与修复后图像做掩膜运算,获得叶片范围内
的病斑边界,然后运用形态学运算剔除图斑中未闭合的边缘线;最后采用归一化绿蓝差值指
数(Normalized Difference Green and Blue Index, DNGBI)对封闭的非病斑区域进行阈值过滤,
提取出稻瘟病病斑。试验结果表明:对叶瘟病斑的正确识别率可达到 %。
关键词:色差模型;边缘检测;DNGBI;水稻叶瘟病;病斑
中图分类号:S24
Segmentation Method of Rice Leaf Blast Based on image
processing
WU Lulu1, ZHENG Zhixiong1, QI Long1,2, MA Xu1,2, KUANG Jianxia1,
CHEN Guorui1
(1. College of Engineering, South China Agricultural University, GuangZhou 510642;
2. Key Laboratory of Key Technology on Agriculture Machina and Equipment (South China
Agricultural University, Ministry of Education, GuangZhou 510642)
Abstract: The paper took the advantage of closed edge feature, colorimetric model, edge detection
and morphology methods were used to detect rice leaf blast. First,canny algorithm was used for
2R-ponent image to detect lesion edge. After that, the study proposed a marginal seal repair
template to repair edges. Mask operation was conducted using repaired image and green leaf
image which was extracted by ponent in HIS model, and the lesion boundary of leaf blast
we