本文已被:浏览 1362次 下载 1015次 |
![本文二维码信息](/download_upload_file.aspx?file_name=/uploadfile/two_dimensional_code_img/zgnydxxb/2022/5/20220508.png) 码上扫一扫! |
基于颜色指数与阈值法的稻田图像分割 |
孙滨峰1,叶春1,2,李艳大1*,舒时富1,曹中盛1,吴罗发1,朱艳3,何勇4
|
|
(1.江西省农业科学院 农业工程研究所, 南昌 330200;2.南昌大学 机电工程学院, 南昌 330031;3.南京农业大学 国家信息农业工程技术中心, 南京 210095;4.浙江大学 生物系统工程与食品科学学院, 杭州 310029) |
|
摘要: |
为系统、全面地分析不同颜色指数对南方稻田图像分割的适应性,以分蘖期、拔节期稻田图像为研究对象,选择36种常用的颜色指数,采用Otsu阈值法开展基于颜色指数和阈值的图像分割研究,通过比较各颜色指数的分割结果,明确分蘖期和拔节期图像分割的主要干扰因素,筛选最适宜稻田图像分割的颜色指数。结果表明:水稻倒影、浮萍是分蘖期稻田图像分割的主要干扰因素,叶片镜面反射、浮萍和土壤阴影是拔节期稻田图像分割的主要干扰因素;组合指数COM2、MxEG、CIVE和GMR在分蘖期图像和拔节期图像均具有较好的分割精度。因此,基于颜色指数COM2、MxEG、CIVE、GMR和Otsu阈值的稻田图像分割方法对稻田图像分割的干扰要素具有较强的区分能力,分割精度较高,更适宜于南方稻田图像处理研究。 |
关键词: 颜色指数 图像分割 数字图像 Otsu阈值 |
DOI:10.11841/j.issn.1007-4333.2022.05.08 |
投稿时间:2021-09-16 |
基金项目:国家自然科学基金项目(41961048);江西现代农业科研协同创新专项(JXXTCXQN201904);“万人计划”青年拔尖人才项目;江西省“双千计划”项目;江西省重点研发计划项目(20192BBF60052,20202BBFL63046,20202BBFL63044,20212BBF61013,20212BBF63040) |
|
Paddy field image segmentation based on color indices and thresholding method |
SUN Binfeng1,YE Chun1,2,LI Yanda1*,SHU Shifu1,CAO Zhongsheng1,WU Luofa1,ZHU Yan3,HE Yong4
|
(1.Institute of Agricultural Engineering, Jiangxi Academy of Agricultural Sciences, Nanchang 330200, China;2.School of Mechanical and Electrical Engineering, Nanchang University, Nanchang 330031, China;3.National Engineering and Technology Center for Information Agriculture, Nanjing Agricultural University, Nanjing 210095, China;4.College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310029, China) |
Abstract: |
In order to systematically and comprehensively analyze the adaptability of different color indices for image segmentation of the paddy fields in southern China, 36 commonly used color indices were selected to study the images of paddy field at the tillering and jointing stages. Otsu thresholding method was used to carry out image segmentation research based on color indices and threshold values. By comparing the segmentation results of each color index, the main disturbing factors were selected and the most suitable color indices for paddy fields image segmentation were identified. The results showed that: The reflection of rice and duckweed were the main factors disturbing paddy fields image segmentation at tillering stage, and leaf specular reflection, duckweed and shadows in the ground were the main factors disturbing segmentation of images at jointing stage. Indices COM2, MxEG, CIVE and GMR presented the better image segmentations of rice canopy at tillering stage and jointing stage of rice. In conclusion, image segmentation based on Otsu, COM2, MxEG, CIVE, GMR with Otsu threshhold is able to to distinguish the interference elements of images segmentation, and more suitable for southern paddy image processing research. |
Key words: color indices image segmentation digital image Otsu thresholding |