| 无芒隐子草叶片卷曲度和厚度测量方法 |
| |
| 引用本文： | 张文霞,王春光,王海超,殷晓飞,宗哲英.无芒隐子草叶片卷曲度和厚度测量方法[J].农业机械学报,2021,52(5):184-191. |
| |
| 作者姓名： | 张文霞 王春光 王海超 殷晓飞 宗哲英 |
| |
| 作者单位： | 鄂尔多斯应用技术学院;内蒙古农业大学;呼和浩特职业学院 |
| |
| 基金项目： | 教育部“云数融合科教创新”基金项目（2017A10019）、内蒙古自治区博士研究生科研创新项目（B20151012902Z）、内蒙古自治区高等学校研究项目(NJZY070、NJZY19288)和鄂尔多斯应用技术学院一般项目（KYYB2017004） |
| |
| 摘 要： | 针对叶片卷曲度和厚度交互式测量方式费时、费力、误差大，传统图像处理算法普适性不高等问题，以无芒隐子草叶片为研究对象，采用基于Graham 算法的最小外接矩形法实现叶片卷曲度的测量，采用矢量积法和角点检测相结合的凹凸点检测算法实现叶片厚度的测量。首先，通过石蜡制片获取无芒隐子草叶切片，利用显微镜连接计算机获取切片图像；然后，采用红色灰度化方法结合阈值分割将切片图像的目标和背景分离；最后，根据叶片卷曲度和厚度的实际测量方式，采用Graham算法通过求取目标区域的最小外接矩形实现叶片卷曲度的测量，将矢量积法和角点检测相结合检测目标区域的凹凸点，通过凹点与凹点、凸点与凸点匹配实现叶片厚度的测量。选取30幅无芒隐子草叶切片图像为样本进行了试验，结果显示，采用本文提出的红色灰度化方法和分量法、最大值法、平均法、加权平均法对图像进行灰度化处理后，图像信息熵分别为6.4280、6.3612、5.6679、5.9348、6.0526，图像平均梯度分别为0.0785、0.0242、0.0158、0.0093、0.0104，图像对比度分别为0.2641、0.1130、0.0574、0.0703、0.0784，说明本文方法能更好地保持图像的边缘、细节等信息，图像清晰度更高。进行自动阈值分割后，分割的平均误检率为0.75%，平均漏检率为3.49%，平均整体分割精度达到98.14%。在有效分割目标和背景的基础上，对叶片卷曲度和厚度进行测量，并与交互式测量结果进行相比，结果表明，采用本文方法对叶片卷曲度和厚度的测量值与交互式测量值的平均相对误差分别为0.96%和3.69%，测量速度分别提高了约10倍和37倍。
|
| 关 键 词： | 无芒隐子草 切片图像 灰度化方法 叶片卷曲度 叶片厚度 凹凸点检测 |
| 收稿时间： | 2020/7/20 0:00:00 |
Measurement Method of Leaf Rolling Index and Thickness of Cleistogenes songorica |
| |
| ZHANG Wenxi,WANG Chunguang,WANG Haichao,YIN Xiaofei,ZONG Zheying.Measurement Method of Leaf Rolling Index and Thickness of Cleistogenes songorica[J].Transactions of the Chinese Society of Agricultural Machinery,2021,52(5):184-191. |
| |
| Authors: | ZHANG Wenxi WANG Chunguang WANG Haichao YIN Xiaofei ZONG Zheying |
| |
| Institution: | Ordos Institute of Technology;Inner Mongolia Agricultural University;Huhhot Vocational College |
| |
| Abstract: | The leaf rolling index and thickness are important indexes of plant drought resistance. However, existing measuring methods of these two indicators are wasteful, inefficient and weak universality. To solve this problem,the minimum external rectangle method based on Graham algorithm was proposed to extract the value of leaf rolling index and the concave and convex point detection algorithm combined the corner detection was proposed to measure the leaf thickness. The algorithm used in the article had the following steps: firstly,the permanent sides were obtained by paraffin sectioning technique, and then connected the microscope and computer to obtain slice images. Secondly, a red grayscale method was proposed for the Cleistogenes songorica leaf anatomical structure image to enhance contrast and details, according to the color difference of red and blue between the foreground and the background. Then through subjective judgment (observing the change of processed image, and comparing the changes to find out which method was the best one), it was found out that images processed by grayscale methods had the shortcoming of details blurred and proposed method showed better performance in improving the quality of image and the details of the images than the other methods. Also by evaluation functions: average gradient(AG),contrast(C) and information entropy(E) were used for objectively evaluated the method used and the traditional ones. The average value of AG,C and E was got by processing 30 test images, and it turned out that the values of the method proposed was better than the other methods. On this basis, eliminated noise using morphological operation and using linear filtering to eliminate serrated boundaries and eventually segmented background and objective by the maximum inter class variance (Otsu). Through subjective judgment, it was found out that the segmented target area basically coincided with the original boundary of the target. Also by evaluation functions: false positive rate(FPR), false negative rate(FNR),and global segmentation accuracy(GSA) were used for objectively evaluating the method. The average value of FPR,FNR and GSA was got by processing 30 test images were 0.75%, 3.49% and 98.14%, respectively. Finally,according to the actual measurement mode of leaf rolling index, the longest distance between the two points on the Cleistogenes songorica leaf anatomical structure image was measured, the minimum external rectangle method based on Graham algorithm was used to extract the value of leaf rolling index of the objective. The measurement was compared with its mean value of interactive measurement by ToupTek Toupview software,the average relative error of 30 test images was 0.96% and the average time consumed was 4.87s by the measurement proposed, the speed was improved by 11 times. According to the actual measurement of leaf thickness was to measure the distance between the concave points and the concave points on the left and right boundary of the Cleistogenes songorica leaf anatomical structure image, it was also the distance between the convex points and the convex points,the proposed concave and convex point detection algorithm combined the corner detection algorithm and vector product method to eliminate the useless apexes to obtain the concave and convex points in accordance with the actual measurement. Then, the leaf thickness value was obtained by concave points matching and convex points matching.The measurement was compared with its mean value of interactive measurement by ToupTek Toupview software,the average relative error of 30 test images was 3.69% and the average time consumed was 4.92s by the measurement proposed,the speed was improved by 38 times. In conclusion, the algorithm was more suitable for background segmentation and object measurement of Cleistogenes songorica leaf slices and can also provide a reference for other plant leaf slice images. |
| |
| Keywords: | Cleistogenes songorica slices images grayscale method leaf rolling index leaf thickness concave and convex point detection |
|
| 点击此处可从《农业机械学报》浏览原始摘要信息 |
| 点击此处可从《农业机械学报》下载免费的PDF全文 |
|
