基于图像分割的甘薯抗褐变种质资源的快速鉴定  

Rapid identification of browning resistant sweetpotato germplasm resources based on image segmentation

作　　者：高燕萍 何培文 吕尊富 崔鹏[1,2] 徐锡明 庞林江[2] 陆国权[1,2] GAO Yanping;HE Peiwen;LV Zunfu;CUI Peng;XU Ximing;PANG Linjiang;LU Guoquan(College of Advanced Agricultural Sciences/The Key Laboratory for Quality Improvement of Agricultural Products ofZhejiang Province,Zhejiang A&F University,Hangzhou 311300,China;Institute of Root&Tuber Crops,Zhejiang A&F University,Hangzhou 311300,China)

机构地区：[1]浙江农林大学现代农学院/浙江省农产品品质改良重点实验室,杭州311300 [2]浙江农林大学薯类作物研究所,杭州311300

出　　处：《中国农业大学学报》2023年第7期46-56,共11页Journal of China Agricultural University

基　　金：国家现代农业产业技术体系(CARS-10-GW21);浙江省重点研发计划(2021C02057);浙江省三农九方项目(2022SNJF008)。

摘　　要：为探究基于图像识别技术的甘薯种质抗褐变能力快速鉴定方法,以98份甘薯品种为材料,基于MATLAB R2018b,利用二值化分割图像背景,检测甘薯切片图像像素值,并利用RGB颜色通过聚类分析分类,应用欧氏距离计算不同类型的甘薯褐变面积,并计算褐变面积在整个甘薯切面的比重。结果表明:分割图像背景后,通过聚类分析将98份材料按薯肉色分为40个白色系品种,35个黄色系品种,23个桔红色系品种。利用不同品种褐变面积比重及方差分析筛选出的优质抗褐变甘薯品种有:白肉色系品种‘紫云薯’、‘川薯231’、‘川薯218’;黄肉色系品种‘龙薯21’、‘齐宁31号’、‘冀菜薯7号’;桔红肉色系品种‘金薯69’。因此,基于颜色空间和欧氏距离的图像分割方法可用于甘薯种质的抗褐变能力快速、准确地鉴定。In order to explore a method for rapid identification of the browning resistance of sweetpotato germplasm based on image recognition technology,98 varieties(lines)were selected as study materials.Based on MATLAB R2018b,using binarization to segment the image background and then pixel values of sweetpotato slice images were detected.The pictures were categorized by RGB color and cluster analysis.For different groups,Euclidean distance was adopted to calculate the browning area of different types of sweetpotatoes and the specific weight ratio of the browning area in the whole image was calculated.The results showed that:After segmenting the background,the 98 materials was divided into three groups by color through cluster analysis.There were 40 varieties of white tone flesh,35 varieties of yellow tone flesh and 23 varieties of orange tone flesh.The highly anti-browning material identified included white tone flesh variety‘Ziyunshu',‘Chuanshu 231',‘Chuanshu 218',yellow tone flesh varieties‘Longshu 21',‘Qining 31',and‘Jicaishu 7',and orange tone fleshvariety‘Jinshu 69'.Therefore,image segmentation means based on color space and Euclidean distance are suitable for rapid and accurate identification of sweetpotato germplasm for browning resistance.

关 键 词：甘薯 褐变 图像分割 MATLAB 

分 类 号：S531[农业科学—作物学]