A Non-linear Model of Nondestructive Estimation of Anthocyanin Content in Grapevine Leaves with Visible/Red-Infrared Hyperspectral
The anthocyanin(Anth) content in leaves provides valuable informa-tion about the physiologocal status of plant. Thus, there is a need for accurate, efficient, practical methodologies to estimate this biochemical parameter. Hyper-spectral measurement is a means of quickly and nondestructively assessing leaf Anth in situ. Wet chemical methods has traditionally been used for this purpose. Recently, NIR(near-infrared)/green, red/green, anthocyanin reflectance in-dex(ARI), and a modified anthocyanin refelctance index(MARI) was been used to estimate the anthocyanin content. In this paper, a an artificialintelligence technique model was introduced to establish the relationship between the anthocyanin content and reflectance of 400-750nm spectum, variation of species and growth stages. The objective of this study was to test the overall performance and accuracy of this new nondestructive techniques for estimating Anth content in grapevine leaves. Although Anth in validation data set was widely variable, the new methods were capable of accurate predicting Anth content in grapevine leaves with a root mean square error below 1.65 mg/m2, which is lower than that of MARI or ARI20.It documents the facts that such an approach is more suitable for developing simple hand-held field instrumentation for accurate nonde-structive Anth estimation and for analyzing digital aribome or satellite imagery to assist in making informed decisins vineyard management.
Grapes Hyperspectral SVM (Support Vector Machine) Anthocyanin.
JiangLin Qin Donald Rundquist Anatoly Gitelson Zongkun Tan Mark Steele
Guangxi Institute of Meteorological Disaster-Reducing Research No.81, Minzu Avenue,Nanning, Guangxi, Center for Advanced Land Management Information Technologies (CALMIT),University of Nebraska-Lincoln Guangxi Institute of Meteorological Disaster-Reducing Research No.81, Minzu Avenue,Nanning, Guangxi,
国际会议
南昌
英文
47-62
2010-10-22（万方平台首次上网日期，不代表论文的发表时间）