Методологічні підходи щодо радіочастотної корекції результатів дистанційного зондування посівів, отриманих за допомогою БПЛА

Abstract

Methodological approaches to correction resaults of scaning in remote sensing plants using UAV

V. Lysenko, O. Opryshko, D. Komarchuk, A. Martsyfey, N. Pasechnik

One of the most promising uses of UAV remote sensing is vegetable plantations for the purpose of rational use of fertilizers. Compared to satellite and unmanned air monitoring have fundamental advantages not only for accuracy and cost monitoring, but also their possible use in low cloud cover, which is extremely important given the need for operational monitoring.

The analysis of the literature made the following conclusions:

1. To produce optical standards should be used self-colored gray panel or panels, the surface of which consists of areas where the colors meet the threshold in the additive model color formation.

2. In the RF correction is advisable to use a series of images with different exposure duration

3. It is advisable to consider the relationship between the values of the parameters of the camera setup file exiff and results on the intensity of color components.

For experiments we used samples of spring wheat sort Elehiya, In experiments the samples were in the stage of the growing season - out of the tube. For the photography samples were cut to 5.10 mm above the root. Reflective panel (template) images in variable lighting made of paper density 80g / m2. For experiments we used a standard digital camera for UAV PHANTOM VISION FC200 camera and smartphones from Lenovo S660 and A1000.

For intensity values of color component samples used software Land Damage Expert (LDE).

The resulting relationship between the respective color component prototypes (wheat) and white optical pattern of FC200 and S660 for all three color components showed good reproducibility. And approximated exponential dependence.

Calibration was based on the use of official data on the camera setup parameters that are obtained from official exif file.

The results were approximated stupnevym simple equation. The analysis of materials of this study can be argued that a calibration compared to the previous (template) gives no worse reproducibility and a wide margin for the application.

Calibration largest Light Value makes it easy to evaluate the data obtained with less exposure duration (maximum LV). This feature is important, based on the needs of a large number of images in the fields of monitoring UAV platform, given the limited duration of its flight.

Conclusions

1. When using optical monitoring templates for vegetable plantations on the state of mineral nutrients most suitable R- and G- components. B component is characterized by small measuring range and the worst convergence of results (R2- 0,41).

2. Calibration based on the reduced duration exposure gives good results for the convergence of all components (R2≥0,94), but with insufficient exposure possible substantial error (EV ˂ -1).

3. The highest accuracy for the full range of exposure compensation received in the calibration provides the largest Light Value, with a maximum convergence results (R2≥ 0,99), but depending on the gauge are created for a specific brand of equipment (sensor).