A Comprehensive Review of Remote Sensing Technology for Mass-Flowering Crops Extraction

• Published in: IEEE journal of selected topics in applied earth observations and remote sensing Vol. 18; pp. 14382 - 14405
• Main Authors: Meng, Qingji, Zang, Shuying, Zhu, Bingxue, Song, Kaishan, Li, Miao, Sun, Li
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Asia; Classification; Crop identification; Crop planting; Crops; Data integration; Data mining; Data sources; Deep learning; Earth; Extraction; Flowering; Flowering plants; Information retrieval; Landsat; Laser radar; Machine learning; mass-flowering crops; Monitoring; Normalized difference vegetative index; Radar data; Rapeseed; Remote sensing; remote sensing (RS); Soft sensors; Spatiotemporal data; Statistical analysis; Statistical methods; sunflower; Vegetation
• DOI: 10.1109/JSTARS.2025.3569094
• ISSN: 1939-1404

Precise and reliable crop evaluations hold significant value in ensuring agricultural security and fostering agricultural progress. Using the flowering characteristics of crops during their growth period to accurately identify crops is a hot research direction in the field of agricultural remote sensing. This article presents a statistical analysis of 46 articles on flowering crops published between 2004 and 2023. Based on the findings, it is evident that China, the United States, and Ukraine are the primary focus of research in this particular field. The main subjects of study are rapeseed, accounting for 50% of the research, and sunflower, which makes up 19.57% of the study. In the extraction of mass-flowering crops and the observation of their flowering periods, commonly used remote sensing data sources include optical data (Sentinel-2, Landsat 8, Landsat 5, MODIS, etc.) and radar data (Sentinel-1, TerraSAR-X, etc.), and the fusion of multisource data is an effective means to improve the research accuracy in this field. Features such as vegetation indices (notably normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI)), band features, polarization features, and phenological features are essential for analyzing mass-flowering crops. Machine learning and deep learning have proven to be valuable tools for conducting classification research in areas with intricate crop planting structures. Spatiotemporal data fusion is an important way to supplement missing images in crop flowering period identification. Sampling points can be obtained through methods such as combining flowering period characteristics with cloud platforms, sample migration, and crowdsourcing activities. This work explores an efficient approach to quickly generate comprehensive crop classification datasets on a global scale. It also presents an overview of the future development of mass-flowering crop extraction, focusing on data sources, information extraction techniques, training samples, and classification methods.