|Paper title||Suitability of thermal UAV data to detect stones and artificial objects in agriculture|
|Form of presentation||Poster|
Stones on agricultural land can cause serious damage to agricultural machinery, when they are getting inside the machinery. This phenoma is especially pronounced in regions with high frequency of stones occurring on agricultural lands, e.g. in glacial morainic landscapes, as they occur in northern Germany. Therefore, stones must be removed from farmland several times a year. A worfklow for drone-based detection of stones is currently under development at the Geoecology department of MLU Halle to assist solving this problem.
With our workflow, we demonstrate the particular suitability of UAS-based thermal data to differentiate between stones and soil surface on agricultural lands. Thermal inertia effects can be used to make significant temperature differences between stone and soil detectable. Which enables precise stone detection through UAS based thermal imaging. We have conducted extensive laboratory testing to investigate the suitability of thermal imaging to detect stones and to find the optimal pre-requisite for thermal UAV flights. We selected the most important variables that have an high impact on the thermal detectability and thus analyzed the influence of soil moisture, air temperature, wind and radiant heatto evaluate thermal detectability by DJI Zenmuse H20T camera.
Within our laboratory experiment we used two identical plastic boxes insulated to the side and bottom with styrofoam and filled with about 40 cm of soil. A total of 4 stones of different sizes were placed on top of the soil. In the center a black aluminum plate was placed for the calibration of the thermal data(see figure 1). The temperatures were simultaneously monitored with a 4 channel logger (PerfectPrime TC0520) attached with 4 RS PRO thermocouples type T (temperature range from -75 °C to +250 °C, IEC 584-3, tolerance class 1).
The following experiments were performed in a clima chamber experiment to account for the different influcening factors:
Scenario No. Temperature Soil moisture Duration Note
1 10°C to 17°C
3,2 % Vol.
7 hours (1°C / hour increased)
2 10°C to 17°C
26,2 % Vol.
7 hours (1°C / hour decreased)
3 constant 17°C 2,6 % Vol. 4 hours 2x 350 W radiators for 2 hours direct irradiation of the examination objects
4 constant 17°C 28,4 % Vol. 4 hours 2x 350 W radiators for 2 hours direct irradiation of the examination objects
By means of radiometric thermal imaging camera attached to a DJI M300 RTK platform imagery data was acquired. The spectral range of the camera is 8-14 μm, the focal length is 13.5-mm, and the sensor resolution is 640 × 512 pixels.. The thermal imaging camera captured an image of the test objects (cf. Fig.1) every minute. The thermal images are stored in a proprietary format and subsequently converted into 8-bit unsigned TIFF files using the DJI Thermal SDK software. The output files were processed to receive text files in the format of X- & Y-image coordinate and temperature. Statistical analysis of the laboratory data were conducted using programming language R and packages raster, rgdal and pracma.
The results show that there are significant temperature differences between stones and soil in the time course of an average temperature scenario during typical stone harvest periods between October and February. The experiment revealed that the factor of soil moisture significantly influences detectability. Likewise, the factor of radiant heat has a significant influence on the detectability of temperature differences between stones and soil.
Based on these insights from standardized laboratory conditions the next steps will focus on the investigation of these approaches under real conditions in the field. The results from the experiment show a great theoretical potential to detect stone by means of thermal UAV imagery and thus this will be evaluated under field conditions in the following month. At the ESA LPS we would like to show up the results of the laboratory experiment and hope to substantiate these with latest information from successfully conducted field experiments during the winter months.