LNC (leaf nitrogen content) in crops is significant for diagnosing the crop growth status and guiding fertilization decisions. Currently, UAV (unmanned aerial vehicles) remote sensing has played an important role in estimating the nitrogen nutrition of crops at the field scale. However, many existing methods of evaluating crop nitrogen based on UAV imaging techniques usually have used a single type of imagery such as RGB or multispectral images, seldom considering the usage of information fusion from different types of UAV imagery for assessing the crop nitrogen status. In this study, GS (Gram–Schmidt Pan Sharpening) was utilized to fuse images from two sensors of digital RGB and multispectral cameras mounted on UAV, and the specific bands of the multispectral cameras are blue, green, red, rededge and NIR. The color space transformation method, HSV (Hue-Saturation-Value), was used to separate soil background noise from crops due to the high spatial resolution of UAV images. Two methods of optimizing feature variables, the Successive Projection Algorithm (SPA) and the Competitive Adaptive Reweighted Sampling method (CARS), combined with two regularization regression algorithms, LASSO and RIDGE, were adopted to estimate the LNC, compared to the commonly used Random Forest algorithm. The results showed that: (1) the accuracy of LNC estimation using the fusion image is improved distinctly by a comparison to the original multispectral image; (2) the denoised images performed better than the original multispectral images in evaluating LNC in rice; (3) the RIDGE-SPA combined method, using SPA to select the MCARI, SAVI and OSAVI, had the best performance for LNC in rice, with an R² of 0.76 and an RMSE of 10.33%. It can be demonstrated that the information fusion of multiple-sensor imagery from UAV coupling with the methods of optimizing feature variables can estimate the rice LNC more effectively, which can also provide a reference for guiding the decision making of fertilization in rice fields.

The resulting maps of land use classification obtained by pixel-based methods often have salt-and-pepper noise, which usually shows a certain degree of cluttered distribution of classification image elements within the region. This paper carries out a study on crop classification and identification based on time series Sentinel images and object-oriented methods and takes the crop recognition and classification of the National Modern Agricultural Industrial Park in Jalaid Banner, Inner Mongolia, as the research object. It uses the Google Earth Engine (GEE) cloud platform to extract time series Sentinel satellite radar and optical remote sensing images combined with simple noniterative clustering (SNIC) multiscale segmentation with random forest (RF) and support vector machine (SVM) classification algorithms to classify and identify major regional crops based on radar and spectral features. Compared with the pixel-based method, the combination of SNIC multiscale segmentation and random forest classification based on time series radar and optical remote sensing images can effectively reduce the salt-and-pepper phenomenon in classification and improve crop classification accuracy with the highest accuracy of 98.66 and a kappa coefficient of 0.9823. This study provides a reference for large-scale crop identification and classification work.

Precise fertilization of rice depends on the timely and effective acquisition of fertilizer application recommended by prescription maps in large-scale cropland, which can provide fertilization spatial information reference. In this paper, the prescription map was discussed based on the improved nitrogen fertilizer optimization algorithm (NFOA), using satellite and unmanned aerial vehicle (UAV) imagery, and supplemented by meteorological data. Based on the principles of NFOA, firstly, remote sensing data and meteorological data were collected from 2019 to 2021 to construct a prediction model for the potential yield of rice based on the in-season estimated yield index (INSEY). Secondly, based on remote sensing vegetation indices (VIs) and spectral features of bands, the grain nitrogen content (GNC) prediction model constructed using the Random Forest (RF) algorithm was used to improve the values of GNC taken in the NFOA. The nitrogen demand for rice was calculated according to the improved NFOA. Finally, the nitrogen fertilizer application recommended prescription map of rice in large-scale cropland was generated based on UAV multispectral images, and the economic cost-effectiveness of the prescription map was analyzed. The analysis results showed that the potential yield prediction model of rice based on the improved INSEY had a high fitting accuracy (R² = 0.62). The accuracy of GNC estimated with the RF algorithm reached 96.3% (RMSE = 0.07). The study shows that, compared with the non-directional and non-quantitative conventional tracking of N fertilizer, the recommended prescription map based on the improved NFOA algorithm in large-scale cropland can provide accurate information for crop N fertilizer variable tracking and provide effective positive references for the economic benefits of rice and ecological benefits of the field environment.

Chlorophyll, as a key component of crop leaves for photosynthesis, is one significant indicator for evaluating the photosynthetic efficiency and developmental status of crops. Fractional-order differentiation (FOD) enhances the feature spectral information and reduces the background noise. In this study, we analyzed hyperspectral data from grape leaves of different varieties and fertility periods with FOD to monitor the leaves’ chlorophyll content (LCC). Firstly, through sensitive analysis, the fractional-order differential character bands were identified, which was used to construct the typical vegetation index (VI). Then, the grape LCC prediction model was built based on the random forest regression algorithm (RFR). The results showed the following: (1) FOD differential spectra had a higher sensitivity to LCC compared with the original spectra, and the constructed VIs had the best estimation performance at the 1.2th-order differential. (2) The accuracy of the FOD-RFR model was better than that of the conventional integer-order model at different fertility periods, but there were differences in the number of optimal orders. (3) The LCC prediction model for whole fertility periods achieved good prediction at order 1.3, $R^2$ = 0.778, RMSE = 2.1, and NRMSE = 4.7%. As compared to the original reflectance spectra, R2 improved by 0.173; RMSE and NRMSE decreased, respectively, by 0.699 and 1.5%. This indicates that the combination of FOD and RFR based on hyperspectral data has great potential for the efficient monitoring of grape LCC. It can provide technical support for the rapid quantitative estimation of grape LCC and methodological reference for other physiological and biochemical indicators in hyperspectral monitoring.

Nitrogen is the main nutrient element in the growth process of white radish, and accurate monitoring of radish leaf nitrogen content (LNC) is an important guide for precise fertilization decisions for radish in the field. Using white radish LNC monitoring as an object, research on radish nitrogen hyperspectral estimation methods was carried out based on leaf hyperspectral and field sample nitrogen data at multiple growth stages using feature selection and integrated learning algorithm models. First, the Vegetation Index (VI) was constructed from hyperspectral data. We extracted sensitive features of hyperspectral data and VI response to radish LNC based on Pearson’s feature-selection approach. Second, a stacking-integrated learning approach is proposed using machine learning algorithms such as Support Vector Machine (SVM), Random Forest (RF), and Ridge and K-Nearest Neighbor (KNN) as the base model in the first layer of the architecture, and the Lasso algorithm as the meta-model in the second layer of the architecture, to realize the hyperspectral estimation of radish LNC. The analysis results show the following: (1) The sensitive bands of the radish LNC are mainly centered around 600–700 nm and 1950 nm, and the constructed sensitive VIs are also concentrated in this band range. (2) The Stacking model with spectral features as inputs achieved good prediction accuracy at the radish spectral leaf, with $R^2$ = 0.7, MAE = 0.16, MSE = 0.05 estimated over the whole growth stage of radish. (3) The Lasso algorithm with variable filtering function was chosen as the meta-model, which has a redundant model-selection effect on the base model and helps to improve the quality of the integrated learning framework. This study demonstrates the potential of the stacking-integrated learning method based on hyperspectral data for spectral estimation of nitrogen content in radish at multiple growth stages.