|Ramalingam, N - OSU|
|Ling, P - OSU|
Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: December 14, 2004
Publication Date: March 15, 2005
Repository URL: http://hdl.handle.net/10113/5483
Citation: Ramalingam, N., Ling, P.P., Derksen, R.C. 2005. Background Reflectance Compensation and its Effect on Multispectral Leaf Surface Moisture Assessment. Transactions of the ASAE. 48(1):375-383. Interpretive Summary: Many chemical applications are made using carrier volumes that wet plant surfaces. Leaf surface water can also be a predictor of disease infection periods. The ability to quickly and accurately assess leaf surface wetness could aid pest management advisors in assessing application techniques as well as predict the need to make fungicide treatments. The goal of this research is to establish means for accurately distinguishing between a wetted and a dry leaf and to determine the minimum thickness of water it may be possible to detect on a leaf surface. A relatively low cost multispectral imaging system using a camera and filters was used to take images of target leaves and to measure reflectance properties. Compared to a spectroradiometer, the reflectances calculated from the multispectral images were representative of true reflectances of the leaves. Spectral and spatial techniques were assessed as means for compensating for background information such as produced soil reflectance. The accuracy of the assessments was improved when background image information was accounted for. This imaging system was found to be able to detect surface water thicknesses of at least 0.006 cm which could be produced by typical high volume spray applications. This research shows it is possible to implement rapid and relatively inexpensive means for assessing sprayer performance in the field and for predicting disease infection periods related to leaf wetness periods.
Technical Abstract: Pest management advisers and pesticide applicators lack rapid and accurate methods for evaluating spray coverage or any other moisture on plant surfaces. The volume of water capable of being retained by plant material can be significant and may permit non-invasive strategies for detection. The objectives of this research were to develop means for non-contact leaf wetness sensing and to develop techniques to compensate for background information. The multispectral imaging system consisted of a monochrome tube camera and filters spanning the VIS, NIR, and MIR wavebands. Using these techniques, spatial and spectral information were simultaneously acquired. The performance of the multispectral imaging system was evaluated by comparing the calculated reflectances with those measured using a spectroradiometer. There was a high correlation between the calculated and measured reflectance values. There were no significant performance differences between the spectral and spatial approaches to leaf surface water assessment. Sensitivity analysis showed that the multispectral imaging system can be used to detect a layer of water as thin as 0.006 cm representing a delivery rate of 500 L/ha for a leaf area index of unity. Background reflectance did affect results. Assessment of leaf surface water was significantly improved when background contributions to the multispectral information was compensated. Spatial techniques were preferred over spectral approaches for background compensation because they were easier to implement.