Straw mulching has been one of the most key practices in conservation tillage. The soil erosion can be reduced for the high soil fertility in the sustainable agriculture. Different types of straw mulching can often dominated by the different farmland strategies and agronomic importance. Although remote sensing has been widely used to monitor straw mulching, it is still lacking on the type classification of the straw mulching. Moreover, there is the great variation in the imaging processing of optical and microwave remote sensing. This study aims to systematically monitor and identify the straw mulching types in farmland using optical and microwave remote sensing. The experimental area was selected as Lishu County in Siping City of Jilin Province, Northeast China, due to its representative conservation tillage practices. Sentinel-1 microwave images and Sentinel-2 optical images were utilized to design 3 classification systems. Classification scenarios were constructed to systematically evaluate the potential of the optical and microwave data in identifying different types of corn straw mulching. Specifically, three classification systems were established to represent combined objectives: 1) stubble mulch + stacked root stubble mulch + root stubble without mulch, 2) straw mulch + no straw mulch, and 3) stubble mulch + no stubble mulch. A total of 39 classification scenarios were developed to systematically combine the three classification systems with 13 feature combinations that derived from optical bands, spectral indices, radar backscatter coefficients, and polarization features. A comprehensive comparison was conducted to evaluate the effectiveness of each classification system from three aspects: 1) the histogram distribution of samples under each classification, 2) the Jeffries-Matusita (JM) distance to quantify class separability, and 3) the accuracy of classification. The results showed that all JM distances with Sentinel-2 optical images exceeded 1.0, indicating the high-class separability. The highest overall accuracies were achieved to identify the 3 systems using Sentinel-2 images, which were 76.00%, 81.00%, and 95.00%, respectively, with the kappa coefficients of 0.64, 0.62, and 0.90, while F1 scores of 75.94%, 80.95%, and 95.00%, respectively. As such, the Sentinel-2 optical images were well-suited to identify the different types of straw mulching. According to the ranking of both JM distance and classification accuracy, the highest performance was found in the classification system of stubble mulch + no stubble mulch, followed by straw mulch + no straw mulch, and finally stubble mulch + stacked root stubble mulch + root stubble without mulch. In contrast, Sentinel-1 microwave images shared the relatively low separability, with all JM distances below 1.0. The highest overall accuracies were obtained to identify the 3 systems using Sentinel-1 images, which were only 42.00%, 64.00%, and 63.00%, respectively, with the kappa coefficients of 0.13, 0.28, and 0.26, while the F1 scores of 41.25%, 63.87%, and 62.82%, respectively. Overall, the identification accuracy of straw mulching was relatively low using Sentinel-1 images. Moreover, the integration of Sentinel-1 images as supplementary input to Sentinel-2 cannot enhance the separability or accuracy of any classification scenario. In conclusion, Sentinel-2 optical remote sensing imagery also demonstrated the superior potential to identify the farmland with straw mulching, especially in the unavailable tillage practices. The findings can also provide the valuable technical support and scientific basis for the decision-making on the potential conservation tillage farmland.

Recent approaches to the internal quality inspection of apples with the application of hyperspectral imaging technology are highly cost-intensive because of labor involvement for the data collection on a fixed posture and manual selection of the region of interest (RoI). In addition, several studies have repeated the data acquisition for the same apple. Current methods cannot meet the automation requirements of the sorting line. Therefore, this study proposed a novel method for automatically selecting RoI in hyperspectral images of apples with random poses. Firstly, the preliminary RoI selection of apple hyperspectral image was carried out, followed by the performance of histogram statistics of each pixel with spectral intensity at 700 nm wavelength. The top 40% area of the spectral intensity was reserved to obtain the magnitude relationship of the spectral intensity of each pixel point and a morphological erosion operation. Original apple RoI was acquired and overexposed pixels were removed with spectral intensity greater than 3900 (maximum 4095) in the reserved area at 700 nm. Secondly, the relationship between apple size and prediction accuracy was measured for the in-depth RoI analysis. A partial least square regression (PLSR) model was established between the average spectrum and apple sugar content of RoI with different sizes. Finally, the established model with the top 70% of the spectral intensity achieved the best prediction accuracy. Non-destructive estimation of apple sugar content was performed through hyperspectral imaging technology with reference to the proposed RoI selection method. A competitive adaptive reweighted sampling algorithm along the PLSR (CARS-PLSR) model was established after black-and-white correction and standard normal transformation (SNV) preprocessing and obtained the highest prediction accuracy. The determination coefficient of cross-validation (R_cv) and root mean square error of cross-validation (RMSECV) were 0.9595 and 0.3203°Brix, respectively. The determination coefficient of prediction (R_p) was 0.9308, and the root mean square error of prediction (RMSEP) was 0.4681°Brix. Results proved that the auto-selection of RoI is an efficient and accurate method, which can provide a foundation in practical application for online apple grading systems with hyperspectral imaging technology.