|DREW, PHILLIP - University Of Missouri|
|Sudduth, Kenneth - Ken|
|THOMPSON, ALLEN - University Of Missouri|
Submitted to: Computers and Electronics in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 4/9/2019
Publication Date: 4/18/2019
Citation: Drew, P.L., Sudduth, K.A., Sadler, E.J., Thompson, A.T. 2019. Development of a multi-band sensor for crop temperature measurement. Computers and Electronics in Agriculture. 162:269-280. https://doi.org/10.1016/j.compag.2019.04.007.
Interpretive Summary: Quantifying spatial and temporal variations in plant stress is important for several precision agriculture applications, including variable rate irrigation and variable rate nutrient application. A common approach to plant stress detection is crop canopy temperature measurement. Canopy temperature measurement can be accomplished using relatively inexpensive infrared thermometers (IRT), but care must be taken that the IRT sees only a representative, sunlit canopy region. Infrared cameras are another option, allowing discrimination of crop from non-crop pixels, but they are generally expensive. Our goal in this research was to develop an infrared imaging system that would allow discriminating crop pixels at a cost level similar to an IRT. This was done by combining an inexpensive, low-resolution infrared camera with a standard color camera and specialized interface hardware and software. The system was successfully calibrated to measure temperature, and was able to discriminate plant from non-plant areas in images. Field tests showed that this system was able to provide temperature data more representative of the crop canopy than IRT measurements. The results of this research will be of use to researchers and instrumentation developers interested in inexpensive ways to improve crop canopy temperature measurements.
Technical Abstract: A system combining a miniature long wavelength infrared (LWIR) camera with a visible, or RGB, camera was developed to capture a field of view and derive a plant-specific temperature measurement. The electronic and software development of the instrument, including calibration, field operation, and post-processing of data, are described. Calibration of the LWIR camera was accurate to 0.65ºC when relating pixel output to a thermal measurement, allowing it to act as a thermal camera. A processing algorithm was developed to identify plants within a visible camera image using a binary mask to identify crop/non-crop components within the field of view. The mask was then used to obtain a crop-only region of interest from the thermal image, over which data were integrated to create a temperature measurement. The instrument was tested against an infrared thermometer on soybean plots during September and October 2016. It was capable of removing shaded areas and soil from thermal images to produce temperature measurements more representative of the crop canopy.