Ziel der vorliegenden Arbeit ist die Entwicklung einer rechnergestützten Beschreibung, Analyse und Interpretation von Steigbildern. Die Methode macht zwei Untersuchungen möglich. Sie erlaubt es erstens, die Relevanz von morphologischen Steigbildmerkmalen zu ermitteln, um Proben unterschiedlicher Herkunft zu gruppieren. Zweitens zeigt sie, in welchem Umfang diese Merkmale zu einer Charakterisierung von Kulturpflanzen unter verschiedenen Anbaubedingungen und mit unterscheidbaren physikalisch-chemischen und biologischen Eigenschaften beitragen. Fernziel ist die Erstellung einer mehrdimensionalen Datenbank, in welcher neben der Auswertung von Steigbildern auch solche anderer bildschaffender Methoden erfasst werden können. Die Datenbank soll so angelegt sein, dass sie auch Grundlage für unabhängige weitere Forschung sein kann. Ein weiteres Ziel ist die Definition von universellen, quantifizierbaren Kriterien unabhängig davon, ob es sich um Steigbilder, Kupferchloridkristallisationen oder Rundfilterchromatographie handelt. Für die Entwicklung der Methode wurden Steigbilder von 60 Proben von Weizenkörnern aus biologischem Anbau gemacht. Sie stammten von zehn verschiedenen Sorten (Sommerweizen) und wurden mit zwei verschiedenen Düngungsstufen an drei Orten angebaut. Zehn quantifizierbare Bildmerkmale wurden ermittelt und für die Beschreibung der Bilder verwendet. Einige Merkmale wurden mit einfachen PC-Programmen errechnet, andere direkt auf den originalen Steigbildern gezählt oder gemessen. Die Messdaten wurden mit HCA (hierarchical cluster analysis) und PCA (principal component analysis) und mit dem Mann-Whitney-Test ausgewertet. Der Vergleich der Ergebnisse von HCA und PCA zeigte eine Übereinstimmung der Gruppierung von morphologischen Merkmalen. Mit dem Mann-Whitney-Test konnte gezeigt werden, dass diese Merkmale einen signifikanten Einfluss auf die Aufteilung in Clusters hatten (gemessen mit der «Ward-Distanz»). Die Clusters, und demnach ausgewählte morphologische Merkmale, korrelierten auch mit ver- schiedenen Anbau Parametern: Sortenherkunft, Düngungsstufe und Ort, ebenso wie chemischen und biologischen Eigenschaften wie Eiweissgehalt, Unkrautdruck, Ährenschieben, Erntezeitpunkt und Ertrag. Weiterführende Studien werden zeigen, ob die vorgestellte Methode auch bei anderen Kulturpflanzen Clusterbildung aufgrund morphologischer Steigbildmerkmale ergibt und ob vergleichbare Korrelationen zwischen Steigbildmerkmalen und Anbaubedingungen auftreten.
N1 -The aim of the present work is to develop a computer-based steigbild characterization, analysis and interpretation method. The method should make it possible to examine the extent to which morphological steigbild criteria are relevant for grouping of samples, and the extent to which morphological steigbild criteria can contribute to the characterization of differences in crop cultivation parameters and in physical/ chemical and biological parameters. A proper quality assessment (ranking) of the samples included in the current study is outside the article’s scope. The long-term goal of this work is to build up a multi-dimensional data library, which also contains data from analyses of results from other imaging methods, and in which one is able to store large amounts of data. Such a data library must also be able to provide additional interpretation possibilities for individual investigations carried out by individuals or groups. A long-term goal is also based upon data in the library to find universal quantifiable criteria that are independent of the type of samples, whether from steigbild, biocrystallization or circular chromatography. The material for the development of the method consisted of images from 60 grain samples from organic cultivation comprising 10 varieties of spring wheat which were grown with two levels of manuring at three different localities. Ten quantifiable steigbild criteria were selected in the images of grain samples and used to characterize the images. Some of the criteria were calculated using simple computer- based software, others were counted or measured directly on the original images. Results of the characterization were analysed using hierarchical cluster analysis (HCA), principal component analysis (PCA) and the Mann-Whitney test. A comparative analysis of the data set with HCA and PCA showed consistent grouping of morphological criteria. The Mann-Whitney test revealed that the morphological criteria had significant influence on the division into clusters at various branch levels calculated using Ward distance. The result of the analysis showed that clusters, and thus the selected morphological criteria, were correlated with some of the reported cultivation conditions parameters: e.g. subspecies, manure level and locality, as well as chemical and biological parameters such as protein, weed pressure at earing and harvest, and yield. It will be valuable to clarify, whether the use of the two-sided cluster analysis with morphological steigbild criteria of other crops will be able to demonstrate the formation of clusters, as has been shown in the current study. It will also be valuable to clarify, whether there is a similar correlation between steigbild criteria and cultivation conditions as in the current study.
AB -The aim of the present work is to develop a computer-based steigbild characterization, analysis and interpretation method. The method should make it possible to examine the extent to which morphological steigbild criteria are relevant for grouping of samples, and the extent to which morphological steigbild criteria can contribute to the characterization of differences in crop cultivation parameters and in physical/ chemical and biological parameters. A proper quality assessment (ranking) of the samples included in the current study is outside the article’s scope. The long-term goal of this work is to build up a multi-dimensional data library, which also contains data from analyses of results from other imaging methods, and in which one is able to store large amounts of data. Such a data library must also be able to provide additional interpretation possibilities for individual investigations carried out by individuals or groups. A long-term goal is also based upon data in the library to find universal quantifiable criteria that are independent of the type of samples, whether from steigbild, biocrystallization or circular chromatography. The material for the development of the method consisted of images from 60 grain samples from organic cultivation comprising 10 varieties of spring wheat which were grown with two levels of manuring at three different localities. Ten quantifiable steigbild criteria were selected in the images of grain samples and used to characterize the images. Some of the criteria were calculated using simple computer- based software, others were counted or measured directly on the original images. Results of the characterization were analysed using hierarchical cluster analysis (HCA), principal component analysis (PCA) and the Mann-Whitney test. A comparative analysis of the data set with HCA and PCA showed consistent grouping of morphological criteria. The Mann-Whitney test revealed that the morphological criteria had significant influence on the division into clusters at various branch levels calculated using Ward distance. The result of the analysis showed that clusters, and thus the selected morphological criteria, were correlated with some of the reported cultivation conditions parameters: e.g. subspecies, manure level and locality, as well as chemical and biological parameters such as protein, weed pressure at earing and harvest, and yield. It will be valuable to clarify, whether the use of the two-sided cluster analysis with morphological steigbild criteria of other crops will be able to demonstrate the formation of clusters, as has been shown in the current study. It will also be valuable to clarify, whether there is a similar correlation between steigbild criteria and cultivation conditions as in the current study.
ST - Development of a Comprehensive Method for Steigbild Characterization, Analysis and Interpretation UR - https://dx.doi.org/10.18756/edn.99.51 Y2 - 2024-12-22 09:48:05 ER -