Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


osu1104392582.pdf (4.22 MB)

ETD Abstract Container

Abstract Header

Non-contact multispectral and thermal sensing techniques for detecting leaf surface wetness

Ramalingam, Nagarajan
http://rave.ohiolink.edu/etdc/view?acc_num=osu1104392582

Abstract Details

2005, Doctor of Philosophy, Ohio State University, Food, Agricultural, and Biological Engineering.
Leaf surface wetness detection is important in plant production for pesticide application evaluation, disease management, and misting control. Efficient application of pesticides may be possible using the feedback from the leaf wetness detection system reducing both the overall input cost and environmental contamination. The goal of this study was to develop non-contact sensing techniques for leaf surface wetness detection. Several non-contacting techniques using spectral, thermal, and imaging sensors were evaluated for the development of an automated feedback controlled spraying system. The study was divided into several sub studies focusing on leaf level and canopy level experiments inside a laboratory under artificial illumination, and canopy level experiments in a greenhouse under natural solar illumination. Multispectral reflectance of the leaves and canopies was measured using a spectroradiometer and a custom built low cost multispectral imaging system. The changes of surface temperature and spectral reflectance in visible (400-700 nm), very-near-infrared (700-1300 nm) and near-infrared ranges (1300-2500 nm) caused by leaf surface moisture were investigated. The spectral information collected using the non-contact sensors was a mixture of reflectance of the objects of interest and also the background in the sensor’s field of view. For accurate leaf surface water analysis, background compensation techniques were evaluated to obtain compensated reflectance spectra. Two approaches were investigated for background reflectance compensation, a spectral approach, which aimed at compensating the measured reflectance for background-contamination using a linear unmixing technique, and a spatial approach, which aimed at extracting only the vegetation pixels from the multispectral images using a vegetation index. Visible and near-infrared regions were found less affected by background whereas very-near-infrared regions had large background effects. Background-reflectance compensation significantly improved the accuracy of leaf surface water assessment. Leaf wetness was assessed using the relative differences in the spectral and thermal measurements recorded before and after spraying. Leaf surface wetness was quantified as the difference between the average equivalent water thickness (EWT) values of sprayed and non-sprayed canopy. The EWT values were calculated using model inversion techniques from the measured multispectral reflectance. Leaf and canopy temperatures were measured using infrared thermometry. The studies on both the leaf and canopy levels indicated that the multispectral reflectance and infrared thermometry techniques were able to differentiate plants with and without surface wetness. It was found that the multispectral sensors could be used to detect leaf surface wetness resulting from a high volume pesticide application. The temperatures of the canopies without surface water were found to be 4.4-5.5 0C higher than that of the canopies with surface water. In the canopy level studies under solar illumination, the feasibility of using the developed sensing methodology to detect leaf surface wetness was evaluated in a greenhouse. A non-contact sensor array consisting of spectral, temperature, and imaging sensors was constructed and mounted on a commercial irrigation boom in the greenhouse. Algorithms were developed to compensate for outdoor lighting variation and background interference on the reflectance measurements of the vegetation. Spectral ratioing techniques were used to differentiate canopies with different surface moisture conditions. The irrigation boom had capabilities to be controlled locally using a handheld controller and also remotely from a personal computer. An onboard computer was used to collect data from the sensor array, process the information, and make spraying decisions. This dissertation explains the results of the experiments that were conducted to validate the concept of non-contact leaf wetness sensing techniques. The multispectral technique had sufficient sensitivity to detect leaf surface water thickness of 0.006 cm or more. A quantitative spectral index has been established to quantify surface water thickness. The infrared temperature sensing technique was able to differentiate wet and non-wet canopies rapidly.
Peter Ling (Advisor)
271 p.
Engineering, Agricultural
Remote sensing; plant monitoring; machine vision; image processing; feedback control; instrumentation; data acquisition; equvalent water thickness; spray coverage assessment; dynamic thresholding; segmentation; background compensation; sensors

Recommended Citations

Citations

  • Ramalingam, N. (2005). Non-contact multispectral and thermal sensing techniques for detecting leaf surface wetness [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1104392582

    APA Style (7th edition)

  • Ramalingam, Nagarajan. Non-contact multispectral and thermal sensing techniques for detecting leaf surface wetness. 2005. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1104392582.

    MLA Style (8th edition)

  • Ramalingam, Nagarajan. "Non-contact multispectral and thermal sensing techniques for detecting leaf surface wetness." Doctoral dissertation, Ohio State University, 2005. http://rave.ohiolink.edu/etdc/view?acc_num=osu1104392582

    Chicago Manual of Style (17th edition)

osu1104392582
2,755
© 2004, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.