基于图像处理技术的小麦条锈病严重度自动分级系统构建

Development of an automatic grading system of wheat stripe rust severity based on image processing

条锈病是小麦上最重要的病害之一,严重影响小麦安全生产。该病害的严重度准确评估对病害预测和病害管理措施制定至关重要。为了实现小麦条锈病严重度的准确评估,本研究基于图像处理技术建立了小麦条锈病严重度自动评估方法,构建了小麦条锈病严重度自动分级系统。借助图像处理软件对获取的小麦条锈病各个严重度级别的发病单叶图像进行了手动分割,获取叶片区域图像和病斑区域图像,并统计获得每一发病单叶的叶片区域和病斑区域的像素数量,计算获得了每一发病单叶的病斑面积占发病叶片总面积的实际百分率。基于图像处理技术,利用4种图像分割方法对发病单叶图像进行叶片区域和病斑区域自动分割,与利用图像处理软件对病害图像进行手动分割所获取结果进行比较,获得了最优的病害图像自动分割方法。利用最优病害图像自动分割方法情况下获得的每一发病单叶的病斑面积占发病叶片总面积的百分率,分别依据每一个严重度级别的基于平均值中间值的实际百分率参考范围和病斑面积实际百分率的99%参考值范围,对每一发病单叶进行严重度评估,评估结果表明,基于病斑面积实际百分率的99%参考值范围的严重度评估方法最优,其严重度评估平均准确率为88.19%。利用最优的病斑图像自动分割方法和最优的严重度评估方法,结合PyQt5库、Qt Designer和PyUIC5设计工具,使用Python语言编程构建了小麦条锈病严重度自动分级系统。本研究为基于图像处理技术的小麦条锈病严重度自动评估提供了基础,并为其他植物病害的严重度评估提供了方法和思路参考。

Stripe(yellow)rust caused by Puccinia striiformis f.sp.tritici is a devastating disease on wheat,which seriously affects the security production of wheat.Correct severity assessment is essential for disease forecasting and adopting effective disease management measures to reduce wheat yield losses.To realize accurately assess the severity of wheat stripe rust,in this study,the methods for the severity assessment of wheat stripe rust were investigated based on image processing and an automatic grading system of wheat stripe rust severity was developed.Based on the acquired disease images of single leaves of wheat stripe rust,manual disease image segmentation operations and pixel statistics operations were performed successively with an image processing software,and the segmented leaf region and lesion region images and the pixel numbers of the corresponding whole leaf regions and lesion regions were obtained.According to the obtained pixel numbers,the actual percentages of lesion areas in the areas of the corresponding whole diseased leaves were calculated.Based on image processing technology,four image segmentation methods were utilized to implement automatic segmentation to obtain leaf region images and lesion region images.Then,the results obtained by using the four automatic segmentation methods were compared with those obtained by using the manual segmentation method via the image processing software,and the optimal automatic segmentation method was achieved.Subsequently,based on the percentages of lesion areas in the areas of the corresponding whole diseased leaves obtained by using the optimal automatic segmentation method,the severity of each diseased leaf was assessed according to the midpoint-of-two-adjacent-means-based actual percentage reference range and the 99%reference range of the actual percentages for each severity class of wheat stripe rust,respectively.The results showed that the assessment method based on the 99%reference range of the actual percentages for each severity class of wheat stripe rust was the optimal,with the average accuracy of 88.19%.Finally,by using the optimal automatic image segmentation method and the optimal severity assessment method,in combination with the PyQt5 library,Qt Designer,and PyUIC5 design tools,an automatic grading system of wheat stripe rust severity was developed with the Python language.This study provided a basis for the automatic assessment of wheat stripe rust severity based on image processing technology,and provided methods and ideas for the severity assessments of other plant diseases.