Timely and accurate estimates of crop parameters are crucial for agriculture management. Unmanned aerial vehicles (UAVs) carrying sophisticated cameras are very pertinent for this work because they can obtain remote-sensing images with higher temporal, spatial, and ground resolution than satellites. In this study, we evaluated (i) the performance of crop parameters estimates using a near-surface spectroscopy (350~2500 nm, 3 nm at 700 nm, 8.5 nm at 1400 nm, 6.5 nm at 2100 nm), a UAV-mounted snapshot hyperspectral sensor (450~950 nm, 8 nm at 532 nm) and a high-definition digital camera (Visible, R, G, B); (ii) the crop surface models (CSMs), RGB-based vegetation indices (VIs), hyperspectral-based VIs, and methods combined therefrom to make multi-temporal estimates of crop parameters and to map the parameters. The estimated leaf area index (LAI) and above-ground biomass (AGB) are obtained by using linear and exponential equations, random forest (RF) regression, and partial least squares regression (PLSR) to combine the UAV based spectral VIs and crop heights (from the CSMs). The results show that: (i) spectral VIs correlate strongly with LAI and AGB over single growing stages when crop height correlates positively with AGB over multiple growth stages; (ii) the correlation between the VIs multiplying crop height and AGB is greater than that between a single VI and crop height; (iii) the AGB estimate from the UAV-mounted snapshot hyperspectral sensor and high-definition digital camera is similar to the results from the ground spectrometer when using the combined methods (i.e., using VIs multiplying crop height, RF and PLSR to combine VIs and crop heights); and (iv) the spectral performance of the sensors is crucial in LAI estimates (the wheat LAI cannot be accurately estimated over multiple growing stages when using only crop height). The LAI estimates ranked from best to worst are ground spectrometer, UAV snapshot hyperspectral sensor, and UAV high-definition digital camera.

The number of wheat ears in the field is very important data for predicting crop growth and estimating crop yield and as such is receiving ever-increasing research attention. To obtain such data, we propose a novel algorithm that uses computer vision to accurately recognize wheat ears in a digital image. First, red-green-blue images acquired by a manned ground vehicle are selected based on light intensity to ensure that this method is robust with respect to light intensity. Next, the selected images are cut to ensure that the target can be identified in the remaining parts. The simple linear iterative clustering method, which is based on superpixel theory, is then used to generate a patch from the selected images. After manually labeling each patch, they are divided into two categories: wheat ears and background. The color feature “Color Coherence Vectors,” the texture feature “Gray Level Co-Occurrence Matrix,” and a special image feature “Edge Histogram Descriptor” are then exacted from these patches to generate a high-dimensional matrix called the “feature matrix.” Because each feature plays a different role in the classification process, a feature-weighting fusion based on kernel principal component analysis is used to redistribute the feature weights. Finally, a twin-support-vector-machine segmentation (TWSVM-Seg) model is trained to understand the differences between the two types of patches through the features, and the TWSVM-Seg model finally achieves the correct classification of each pixel from the testing sample and outputs the results in the form of binary image. This process thus segments the image. Next, we use a statistical function in Matlab to get the exact a precise number of ears. To verify these statistical numerical results, we compare them with field measurements of the wheat plots. The result of applying the proposed algorithm to ground-shooting image data sets correlates strongly (with a precision of 0.79–0.82) with the data obtained by manual counting. An average running time of 0.1 s is required to successfully extract the correct number of ears from the background, which shows that the proposed algorithm is computationally efficient. These results indicate that the proposed method provides accurate phenotypic data on wheat seedlings.

The number of panicles per unit area is a common indicator of rice yield and is of great significance to yield estimation, breeding, and phenotype analysis. Traditional counting methods have various drawbacks, such as long delay times and high subjectivity, and they are easily perturbed by noise. To improve the accuracy of rice detection and counting in the field, we developed and implemented a panicle detection and counting system that is based on improved region-based fully convolutional networks, and we use the system to automate rice-phenotype measurements. The field experiments were conducted in target areas to train and test the system and used a rotor light unmanned aerial vehicle equipped with a high-definition RGB camera to collect images. The trained model achieved a precision of 0.868 on a held-out test set, which demonstrates the feasibility of this approach. The algorithm can deal with the irregular edge of the rice panicle, the significantly different appearance between the different varieties and growing periods, the interference due to color overlapping between panicle and leaves, and the variations in illumination intensity and shading effects in the field. The result is more accurate and efficient recognition of rice-panicles, which facilitates rice breeding. Overall, the approach of training deep learning models on increasingly large and publicly available image datasets presents a clear path toward smartphone-assisted crop disease diagnosis on a global scale.

To obtain an accurate count of wheat spikes, which is crucial for estimating yield, this paper proposes a new algorithm that uses computer vision to achieve this goal from an image. First, a home-built semi-autonomous multi-sensor field-based phenotype platform (FPP) is used to obtain orthographic images of wheat plots at the filling stage. The data acquisition system of the FPP provides high-definition RGB images and multispectral images of the corresponding quadrats. Then, the high-definition panchromatic images are obtained by fusion of three channels of RGB. The Gram–Schmidt fusion algorithm is then used to fuse these multispectral and panchromatic images, thereby improving the color identification degree of the targets. Next, the maximum entropy segmentation method is used to do the coarse-segmentation. The threshold of this method is determined by a firefly algorithm based on chaos theory (FACT), and then a morphological filter is used to de-noise the coarse-segmentation results. Finally, morphological reconstruction theory is applied to segment the adhesive part of the de-noised image and realize the fine-segmentation of the image. The computer-generated counting results for the wheat plots, using independent regional statistical function in Matlab R2017b software, are then compared with field measurements which indicate that the proposed method provides a more accurate count of wheat spikes when compared with other traditional fusion and segmentation methods mentioned in this paper.

Timely and accurately estimates of crop biomass and grain yield estimation are crucial for agricultural management. Optical remote sensing techniques can provide crop parameters (e.g., biomass, fractional vegetation cover (FVC)) at regional and larger scales. However, such techniques saturate at high crop canopy cover and cannot detect biomass stored in reproductive organs. The AquaCrop model can be used to estimate FVC, biomass, and grain yield output based on crop growth environmental parameters (e.g., temperature, rainfall, irrigation). In this work, we developed a method for estimating and mapping crop biomass and grain yield using unmanned aerial vehicle (UAV) remote sensing images and AquaCrop. The combination of low-cost UAV remote sensing data and AquaCrop can be used to map wheat biomass and grain yield before harvest. This work investigated whether biomass and grain yield can be predicted using measurements of biomass and FVC in several winter-wheat fields at an early growing stage using ground-based and remote sensing-based methods. FVC, biomass, and grain yield were measured at several key growing stages (jointing (S1), heading (S2), flowering (S3), grain filling (S4)) and harvest (H) during 2014–2015. We specifically evaluated the performances of using field- and UAV-based FVC and biomass measurements from different growing-stage combinations (e.g., S1–S4, S1–S3, and S1–S2) to calibrate the AquaCrop model and estimate biomass and grain yield. The results indicate that winter-wheat biomass and grain yield can be estimated by calibrating AquaCrop using (i) biomass and (ii) biomass and FVC. The results also reveal that the biomass and grain yield estimated using FVC and AquaCrop have poor accuracy compared with the biomass and grain yield estimated using (i) only biomass and (ii) biomass and FVC. The results suggest that the combined use of UAV remote sensing and AquaCrop can be used to obtain maps of biomass and biomass yield. When estimating winter-wheat biomass and grain yield one month (13 May) before harvest (11 June), the predicted biomass and grain yield agreed with the measured values (biomass: n = 96, coefficient of determination (R²) = 0.61, mean absolute error (MAE) = 1.69 t ha^–1, root-mean-square error (RMSE) = 2.10 t ha^–1, normalized RMSE (nRMSE) = 18.5%; grain yield: n = 48, R² = 0.63, MAE = 0.96 t ha^–1, RMSE = 1.16 t ha^–1, nRMSE = 21.9%). When estimating winter-wheat biomass and grain yield one and a half months (26 April) before harvest (11 June), the predicted biomass and grain yield was lower than when estimating winter-wheat biomass and grain yield one month before harvest (biomass: n = 144, R² = 0.55, MAE = 2.52 t ha^–1, RMSE = 3.34 t ha^–1, nRMSE = 23.6%; grain yield: n = 48, R² = 0.34, MAE = 1.68 t ha^–1, RMSE = 2.17 t ha^–1, nRMSE = 35.7%).