中国农业科学院农业信息研究所 中国农学会
Agricultural Information Institute of CAAS China Association of Agricultural Science Societies
En
Ch
中国农业科学院农业信息研究所 中国农学会 主办

AI检索

订阅

返回顶部
当前位置: 首页 > 文章 > 基于轮廓固性与分水岭算法的叶片粘连雾滴图像分割 中国农机化学报 2023 (6) 182-191
Position: Home > Articles > 基于轮廓固性与分水岭算法的叶片粘连雾滴图像分割 Journal of Chinese Agricultural Mechanization 2023 (6) 182-191

基于轮廓固性与分水岭算法的叶片粘连雾滴图像分割

作  者：
葛翔;陆军;曹冬林;金天澍
单  位：
上海电机学院机械学院;浙江省慈溪市观海卫镇农业农村办公室;樱田农机科技(泰州)有限公司
关键词：
雾滴覆盖率;雾滴沉积密度;粘连雾滴图像分割;轮廓固性;分水岭算法
摘  要：
为精准辨识叶片粘连雾滴，提高叶面雾滴沉积分布检测的准确性，利用智能手机采集田间植保作业雾滴图像，选取9组喷雾检测叶片样本，使用以灰度级形态学为基础的处理算法进行叶片区域和雾滴区域特征提取。提出基于轮廓固性的雾滴类型判别方法，并与传统的基于圆形度和基于形状因子的方法进行对比；以粘连雾滴迭代开运算的结果作为标记，对传统距离变换分水岭图像分割算法进行改进。结果表明：经改进的分水岭检测算法，雾滴覆盖率比OTSU算法（最大类间方差法）与分块阈值算法分别高31.93%和33.43%,雾滴提取最低正确率为91.19%,平均正确率为97.23%;基于固性的雾滴类型判别方法正确率为97.5%,较基于圆形度和基于形状因子的方法分别提高21%和5%;沉积密度与人工统计的平均相对误差为6.61%,粘连雾滴分割最低正确率为91.03%,平均正确率为94.74%。该算法可为植保器械田间作业质量图像检测提供技术支持。
作  者：
Ge Xiang;Lu Jun;Cao Donglin;Jin Tianshu;College of Mechanical Engineering, Shanghai Dianji University;Yingtian Agricultural Machinery Technology (Taizhou) Co., Ltd.;Agricultural and Rural Office,Guanhaiwei Town in Cixi, Zhejiang Province;
关键词：
droplet coverage;;droplet deposition density;;image segmentation of adhesive droplet;;contour solidity;;watershed algorithm
摘  要：
In order to accurately identify the leaf adhesive droplets and improve the accuracy of detection of deposition and distribution of leaf droplets, a smart phone was used to collect droplet images of field plant protection operation. Nine groups of sprayed leaf samples were selected, and the algorithm based on gray level morphology was applied to extract the features of leaf area and droplet area. A method of droplet type discrimination based on contour solidity was proposed and compared with the traditional methods based on circularity and shape factor, the traditional distance transform watershed image segmentation algorithm was improved by using the result of iterative opening operation of adhesive droplets as a marker. The results showed that the improved watershed detection algorithm had 31.93% and 33.43% higher droplet coverage than OTSU method and block threshold method, respectively. The minimum accuracy of droplet extraction was 91.19% and the average accuracy was 97.23%; the accuracy of droplet type discrimination method based on solidity was 97.5%, which was 21% and 5% higher than that based on circularity and shape factor, respectively; the average relative error between deposition density and manual statistics was 6.61%, the lowest accuracy of adhesive droplet segmentation was 91.03%, and the average accuracy was 94.74%. The identification method of adhesive droplets based on contour fixation and the improved watershed algorithm of image segmentation can provide technical support for the image detection of field operation quality of plant protection equipment.

相似文章

PDF下载 PDF预览
导出引用
计量
文章访问数: 17
HTML全文浏览量: 0
PDF下载量: 0

所属期刊

推荐期刊

友情链接Friendly link
农业信息研究所Agricultural information institute 农业智能知识服务平台Agricultural intelligent knowledge service platform 农业学术服务平台Agricultural Knowledge Service System 中国知网CNKI 维普网VIP 万方WANFANG DATA
联系我们Contact us

电话Tell：010-82109652-808

邮箱E-mail：agrijournal@caas.cn

服务Service
关于我们About us 意见建议Opinions and suggest
版权说明Copyright
单位：中国农业科学院农业信息研究所 中国农学会 © 2017中国农业科学院农业信息研究所 ICP备案号：京ICP备09089781号-46