摘要
利用图像分析技术对农田对象进行分类,识别农田中不同植物和不同湿度土壤,为定点变量作业提供依据。首先,针对农田各类对象包含颜色信息的不同,采用不同因子实现农田图像的灰度化;然后,利用3种灰度图像对绿色植物、蓝色天空和褐色土壤的识别优势,分析比较阈值法和K均值聚类方法并实现了图像分割;最后,利用模糊聚类法对绿色植物和不同湿度的土壤进一步实现分类。实验结果表明,利用K均值聚类法对绿色植物的平均识别率可达92.5%,对不同湿度的3类土壤的平均识别率达95.6%。因此,本研究能够准确分割和识别不同类型的植物与土壤,为农田对象的识别提供了基础。
Aiming to provide the basis for site-specific operation, this paper classifies different types of plants and dif- ferent soil moisture of the farmland object using the image analysis techniques. Firstly, different gray indices are used to achieve gray scale image according to color information in different farmland objects. With the identification advantages of3 kinds of gray scale images for plants, sky and soil, and then threshold method and K-clustering method are analysed and compared to obtain image segmentation. Finally, the plants and different soil moisture achieved further classification using fuzzy clustering method. The experimental results show that the average identification rate of plants is 92.5 %, and the average identification rate of three types of soil is 95.6% based on K mean histogram method. Therefore, it can be applied to segment and identify different types of plants and soil accurately, and provides a foundation for the identifica- tion of farmland objects.
出处
《农机化研究》
北大核心
2015年第2期40-45,共6页
Journal of Agricultural Mechanization Research
基金
国家自然科学基金项目(31101075)
西北农林科技大学科研专项基金项目(QN2011069)
博士科研启动基金(2011BSJJ095)
大学生科创基金项目(1210712123
1201310712159)
关键词
图像分析
灰度化
图像滤波
图像分割
模糊聚类
农田
image analysis
image graying
image filtering
image segmentation
fuzzy clustering
form land
