Accurate crop planting area information is of significance for understanding regional food security and agricultural development planning. While increasing numbers of medium resolution satellite imagery and improved classification algorithms have been used for crop mapping, limited efforts have been made in feature selection, despite its vital impacts on crop classification. Furthermore, different crop types have their unique spectral and phenology characteristics; however, the different features of individual crop types have not been well understood and considered in previous studies of crop mapping. Here, we examined an optimized strategy to integrate specific features of individual crop types for mapping an improved crop type layer in the Sanjiang Plain, a new food bowl in China, by using all Sentinel-2 time series images in 2018. First, an automatic spectro-temporal feature selection (ASTFS) method was used to obtain optimal features for individual crops (rice, corn, and soybean), including sorting all features by the global separability indices for each crop and removing redundant features by accuracy changes when adding new features. Second, the ASTFS-based optimized feature sets for individual crops were used to produce three crop probability maps with the Random Forest classifier. Third, the probability maps were then composited into the final crop layer by considering the probability of each crop at every pixel. The resultant crop layer showed an improved accuracy (overall accuracy = 93.94%, Kappa coefficient = 0.92) than the other classifications without such a feature optimizing process. Our results indicate the potential of the ASTFS method for improving regional crop mapping.
Using satellite remote sensing has become a mainstream approach for extracting crop spatial distribution. Making edges finer is a challenge, while simultaneously extracting crop spatial distribution information from high-resolution remote sensing images using a convolutional neural network (CNN). Based on the characteristics of the crop area in the Gaofen 2 (GF-2) images, this paper proposes an improved CNN to extract fine crop areas. The CNN comprises a feature extractor and a classifier. The feature extractor employs a spectral feature extraction unit to generate spectral features, and five coding-decoding-pair units to generate five level features. A linear model is used to fuse features of different levels, and the fusion results are up-sampled to obtain a feature map consistent with the structure of the input image. This feature map is used by the classifier to perform pixel-by-pixel classification. In this study, the SegNet and RefineNet models and 21 GF-2 images of Feicheng County, Shandong Province, China, were chosen for comparison experiment. Our approach had an accuracy of 93.26%, which is higher than those of the existing SegNet (78.12%) and RefineNet (86.54%) models. This demonstrates the superiority of the proposed method in extracting crop spatial distribution information from GF-2 remote sensing images.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.