Dissertation > Excellent graduate degree dissertation topics show

Yunyan hyperspectral remote sensing studies on harvest maturity recognition based on 87

Author: LiangYin
Tutor: ZhangYunWei
School: Kunming University of Science and Technology
Course: Agricultural Electrification and Automation
Keywords: hyperspectral remote sensing maturity continuum removal red edge position J-M distance k folds cross validation
CLC: S572
Type: Master's thesis
Year: 2013
Downloads: 1
Quote: 0
Read: Download Dissertation

Abstract


Harvesting is the key cultivating measure determining the quality of the tobacco leaf raw materials. While utilizing the effective technical means and quantitative methods to judge of the tobacco leaf maturity is still unavailable. Hyperspectral remote sensing technology is applied in field management for it is a convenient technical mean to sense the information from crops. However, combining such observation technology and the recognition of maturity of tobacco leaf has not been studied thoroughly and yet can not be applied. In order to do the combination, finding the quantitative methods to judge the maturity and finally fulfilling A wide range of remote sensing monitoring of tobacco maturity process, this paper do the study the problem of tobacco leaf maturity recognition in which choose the species of Yunyan87as the trial material, hyperspectal remote sensing as the technical mean and the ground-based observation as the implementation method.When conducting the continuum removal method and IG. model to process the spectral data, it has been found that within the visible spectrum, there are9spectral features including red edge positions; Absorption peak area in specific regions of spectrum, their ratio and difference, Absorption peak depth in specific regions of spectrum, their ratio and difference; the symmetry of the absorption peak; the half of the absorption peak area.When calculating the JM distance of the adjacent maturity category, it has been found that among the features any high-dimensional feature vectors combined by the chlorophyll absorption peak area or its left half peak area and the rest7spectral features can improve the separability among different classes. However the separability can reach the maximum value1.414, which means the class is separable, but not perfect separable, in the same time, the separability of the high-dimensional vector is no more better than the performance of two-dimensional vector.When modeling the classifier for categorizing the maturity of tobacco, it has been found that support vector machine is a suitable method for this issue. It can produce two mathematical models. The mathematical models are two optimal decision hyperplanes of distinguishing the unripe class and the ripe class, the ripe class and the overripe class on the plane constructed by the red edge position and the chlorophyll absorption peak area wich represented by the formulas of y=-6.8750·x+4854.30and y=-5.2993x+3722.26. However in the process of modeling, the mistake is inevitable for support vector machine is applicable to solve two classification problems, but not the three classification problems, so the algorithm for error elimination is proposed and as it surveyed in the paper, this algorithm can eliminate such mistake.When utilizing holdout cross-validation and10times10folds cross-validation methodology to obtain the generalization error rate when applying the support vector machine to the classification problems of tobacco leaf maturity. It has been found that the error rate of the classifier which categorizes unripe class and the ripe class is around1.04% and1.4%of the classifier which categoraizes the ripe class and the overripe class. hence we consider that it is feasible to use the support vector machine in the identification research of tobacco leaf ripeness degree which is expressed by hyperspectral remote sensing data.

Related Dissertations

  1. Research on Spaceborne Hyperspectral Sensor Simulation,TP391.9
  2. Computing Minimum Distance between Curves/Surfaces Based on PSO Algorithm,O182
  3. Research on Design and Application of Dynamic Textbook,G434
  4. Characterization and Identification of Maturity of Pig Manure Composting,S141.4
  5. Color Parameters Changes of Flue-cured Tobacco during Maturing and Curing and the Influence of Air-curing Before Flue-curing on Its Quality,S572
  6. Modelling Phycocyanin and CDOM Concentration from Hyperspectral Reflectance Data in Lake Taihu,X87
  7. Spectral Characteristics of Rice Damaged by Nilaparvata Lugens (St(?)l) and Cnaphalocrocis Medinalis(Guen(?)e),S435.112
  8. Study on Genetic Diversity on ORFs Regions and Indica-Japonica Consanguinity Differentiation of Weedy Rice,S511
  9. Effects of Fruit Maturity and Fertilizers on the Fruit Quality and Antioxidant Activity of Hilly Pear Jujube,S665.1
  10. Research on Work Risk Perception, and Its Relation with Career Decision-Making Self-Efficacy and Career Marurity of College Students,G647.38
  11. The spatial characteristics of scientific cooperation and Its Influencing Factors,G322.1
  12. Modern Distance Education System personalized recommendation algorithm,TP391.6
  13. Sensorless Control of PMSMs Based on Nonlinear Observer,TM341
  14. Research on Site Selection of Distribution Center of Rural Consumer Goods,F224
  15. Conformal Predictor Based on Transduction,TP181
  16. Path Selection in Intelligent Transportation and Its Information Display System on Mobile Terminal,TP399-C6
  17. Research on Obstacle Detection and Obstacle Avoidance Strategy of Intelligent Vehicle,U495
  18. Study on the Uncertainties of Geotechnical Parameters and Its Application,TU43
  19. Workflow-based distance education and training systems design and implementation,TP311.52
  20. Institutional Environment, the Nature of Ultimate Ownership and Corporate Debt Maturity Structure,F275
  21. The Influence of Psychic Distance on Foreigen Dierct Investement Flows,F224

CLC: > Agricultural Sciences > Crop > Economic crops > Tobacco ( tobacco )
© 2012 www.DissertationTopic.Net  Mobile