The acquisition, processing, and interpretation of thermal images from unmanned aerial vehicles (UAVs) is becoming a useful source of information for agronomic applications because of the higher temporal and spatial resolution of these products compared with those obtained from satellites. However, due to the low load capacity of the UAV they need to mount light, uncooled thermal cameras, where the microbolometer is not stabilized to a constant temperature. This makes the camera precision low for many applications. Additionally, the low contrast of the thermal images makes the photogrammetry process inaccurate, which result in large errors in the generation of orthoimages. In this research, we propose the use of new calibration algorithms, based on neural networks, which consider the sensor temperature and the digital response of the microbolometer as input data. In addition, we evaluate the use of the Wallis filter for improving the quality of the photogrammetry process using structure from motion software. With the proposed calibration algorithm, the measurement accuracy increased from 3.55 °C with the original camera configuration to 1.37 °C. The implementation of the Wallis filter increases the number of tie-point from 58,000 to 110,000 and decreases the total positing error from 7.1 m to 1.3 m.

A field experiment was carried out to implement a remote sensing energy balance (RSEB) algorithm for estimating the incoming solar radiation (Rsi), net radiation (Rn), sensible heat flux (H), soil heat flux (G) and latent heat flux (LE) over a drip-irrigated olive (cv. Arbequina) orchard located in the Pencahue Valley, Maule Region, Chile (35°25′S; 71°44′W; 90 m above sea level). For this study, a helicopter-based unmanned aerial vehicle (UAV) was equipped with multispectral and infrared thermal cameras to obtain simultaneously the normalized difference vegetation index (NDVI) and surface temperature (T_surface) at very high resolution (6 cm × 6 cm). Meteorological variables and surface energy balance components were measured at the time of the UAV overpass (near solar noon). The performance of the RSEB algorithm was evaluated using measurements of H and LE obtained from an eddy correlation system. In addition, estimated values of Rsi and Rn were compared with ground-truth measurements from a four-way net radiometer while those of G were compared with soil heat flux based on flux plates. Results indicated that RSEB algorithm estimated LE and H with errors of 7% and 5%, respectively. Values of the root mean squared error (RMSE) and mean absolute error (MAE) for LE were 50 and 43 W m⁻² while those for H were 56 and 46 W m⁻², respectively. Finally, the RSEB algorithm computed Rsi, Rn and G with error less than 5% and with values of RMSE and MAE less than 38 W m⁻². Results demonstrated that multispectral and thermal cameras placed on an UAV could provide an excellent tool to evaluate the intra-orchard spatial variability of Rn, G, H, LE, NDVI and T_surface over the tree canopy and soil surface between rows.

Water stress, which affects yield and wine quality, is often evaluated using the midday stem water potential (Ψ_stem). However, this measurement is acquired on a per plant basis and does not account for the assessment of vine water status spatial variability. The use of multispectral cameras mounted on unmanned aerial vehicle (UAV) is capable to capture the variability of vine water stress in a whole field scenario. It has been reported that conventional multispectral indices (CMI) that use information between 500–800 nm, do not accurately predict plant water status since they are not sensitive to water content. The objective of this study was to develop artificial neural network (ANN) models derived from multispectral images to predict the Ψ_stem spatial variability of a drip-irrigated Carménère vineyard in Talca, Maule Region, Chile. The coefficient of determination (R²) obtained between ANN outputs and ground-truth measurements of Ψ_stem were between 0.56–0.87, with the best performance observed for the model that included the bands 550, 570, 670, 700 and 800 nm. Validation analysis indicated that the ANN model could estimate Ψ_stem with a mean absolute error (MAE) of 0.1 MPa, root mean square error (RMSE) of 0.12 MPa, and relative error (RE) of −9.1%. For the validation of the CMI, the MAE, RMSE and RE values were between 0.26–0.27 MPa, 0.32–0.34 MPa and −24.2–25.6%, respectively.

Precision livestock farming has emerged with the aim of providing detailed information to detect and reduce problems related to animal management. This study aimed to develop and validate computer vision techniques to track required features of cattle face and to remotely assess eye temperature, ear-base temperature, respiration rate, and heart rate in cattle. Ten dairy cows were recorded during six handling procedures across two consecutive days using thermal infrared cameras and RGB (red, green, blue) video cameras. Simultaneously, core body temperature, respiration rate and heart rate were measured using more conventional ‘invasive’ methods to be compared with the data obtained with the proposed algorithms. The feature tracking algorithm, developed to improve image processing, showed an accuracy between 92% and 95% when tracking different areas of the face of cows. The results of this study also show correlation coefficients up to 0.99 between temperature measures obtained invasively and those obtained remotely, with the highest values achieved when the analysis was performed within individual cows. In the case of respiration rate, a positive correlation (r = 0.87) was found between visual observations and the analysis of non-radiometric infrared videos. Low to high correlation coefficients were found between the heart rates (0.09–0.99) obtained from attached monitors and from the proposed method. Furthermore, camera location and the area analysed appear to have a relevant impact on the performance of the proposed techniques. This study shows positive outcomes from the proposed computer vision techniques when measuring physiological parameters. Further research is needed to automate and improve these techniques to measure physiological changes in farm animals considering their individual characteristics.

Water stress caused by water scarcity has a negative impact on the wine industry. Several strategies have been implemented for optimizing water application in vineyards. In this regard, midday stem water potential (SWP) and thermal infrared (TIR) imaging for crop water stress index (CWSI) have been used to assess plant water stress on a vine-by-vine basis without considering the spatial variability. Unmanned Aerial Vehicle (UAV)-borne TIR images are used to assess the canopy temperature variability within vineyards that can be related to the vine water status. Nevertheless, when aerial TIR images are captured over canopy, internal shadow canopy pixels cannot be detected, leading to mixed information that negatively impacts the relationship between CWSI and SWP. This study proposes a methodology for automatic coregistration of thermal and multispectral images (ranging between 490 and 900 nm) obtained from a UAV to remove shadow canopy pixels using a modified scale invariant feature transformation (SIFT) computer vision algorithm and Kmeans++ clustering. Our results indicate that our proposed methodology improves the relationship between CWSI and SWP when shadow canopy pixels are removed from a drip-irrigated Cabernet Sauvignon vineyard. In particular, the coefficient of determination (R²) increased from 0.64 to 0.77. In addition, values of the root mean square error (RMSE) and standard error (SE) decreased from 0.2 to 0.1 MPa and 0.24 to 0.16 MPa, respectively. Finally, this study shows that the negative effect of shadow canopy pixels was higher in those vines with water stress compared with well-watered vines.