期刊文献+

基于饱和度分割的叶面积图像测量方法 被引量:10

Image measurement method of leaf area based on saturation segmentation
下载PDF
导出
摘要 针对复杂光照环境下的图像叶面积测量问题,提出一种基于饱和度分割的图像叶面积测量方法。首先,利用矩形标定框顶点坐标对拍摄图像进行透视畸变校正和裁剪,以去除倾斜拍摄引起的图像变形;其次,将图像像素由RGB颜色空间变换到HSI颜色空间,提出一种混合阈值法对饱和度分量进行阈值分割;然后,对分割后的二值图像进行连通域分析、孔洞填充和小区域删除,提取形状特征区分叶片区域和标准块区域;最后,统计叶片像素个数与标准块像素个数的比值,求取叶片面积。结果表明:本研究方法能够适应低光照、不均匀光照等复杂光照拍摄环境,适宜的拍摄距离为20~40 cm,适宜的拍摄倾角为0°~50°。该方法具有较好的重复性,在多种摆放方向下对女贞、九里香、豆瓣绿、藜、何首乌、韭菜、三叶草、香椿、樟等9种植物叶片进行重复测量,最大叶面积标准差为12.9 mm^(2),最大叶面积相对标准差为1.33%。 Leaf is the main part of organic nutrients and photosynthesis in crop production.Therefore,rapid and accurate measurement of plant leaf area is of great significance in modern agricultural production.With the development of computer technology,the method of leaf area measurement based on image processing has been developed rapidly.In recent years,image measurement methods of leaf area are combined with smart phones,which is flexible,accurate and fast.Leaf segmentation is a key step in image measurement of leaf area.The existing methods mainly adopt the following two steps:extract the leaf area by detecting green pixels in RGB,YCbCr and other color spaces;extract the leaf area by thresholding,watershed,and other image segmentation algorithms after image graying.These methods have high requirements on the lighting environment and cannot be applied to complex lighting environment such as low light and uneven light.Aiming at solving the problems of image measurement of leaf area in the complex illumination environment,this study proposed an image measurement method of leaf area based on saturation segmentation.Firstly,the vertex coordinates of the calibration rectangle were used to correct the perspective distortion of image to diminish the image deformation caused by oblique shooting.Secondly,the pixels were transformed from RGB color space to HSI color space,and a hybrid threshold method was proposed to segment the saturation image.Thirdly,the connected region analysis,hole filling,and small area deletion were carried out and shape feature was extracted to distinguish leaf and standard block.Finally,the ratio of pixel numbers of leaf and standard block was counted to calculate the leaf area.The experimental results showed that the proposed method can adapt to the complex illumination environment,such as low light and uneven light.The suitable shooting distance was 20-40 cm,and the appropriate shooting angle was 0°-50°.Nine kinds of leaves,including Ligustrum lucidum,Murraya exotica L.,Peperomia tetraphylla,Chenopodium album,Polygonum multiflorum L.,A.tuberosum Rottl.ex Spreng.,shamrock,Toona sinensis and Cinnamomum camphora(L.)presl,were measured repeatedly in various directions.The maximum standard deviation of leaf area was 12.9 mm^(2) and the maximum relative standard deviation of leaf area was 1.33%.
作者 李秋洁 杨远明 袁鹏成 薛玉玺 LI Qiujie;YANG Yuanming;YUAN Pengcheng;XUE Yuxi(College of Mechanical and Electronic Engineering,Nanjing Forestry University,Nanjing 210037,China)
出处 《林业工程学报》 CSCD 北大核心 2021年第4期147-152,共6页 Journal of Forestry Engineering
基金 国家自然科学基金(31901239) 江苏省农业科技自主创新资金项目[CX(18)1007]。
关键词 叶面积测量 图像处理 饱和度分割 透视畸变校正 leaf area measurement image processing saturation segmentation perspective distortion correction
  • 相关文献

参考文献16

二级参考文献202

共引文献179

同被引文献137

引证文献10

二级引证文献10

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部