作物遥感分类与识别是农业遥感监测中重要研究内容之一，是准确获取农作物种植面积、长势、产量、作物品质以及主要农业灾害评价等遥感空间分布信息的基础。混合像元现象普遍存在于遥感影像中，亚像元定位技术作为解决混合像元问题的后续有效手段对作物遥感分类与识别信息准确提取具有重要意义。针对现有模型及算法难以兼顾两种尺寸地物进行亚像元定位的不足，以及进一步提高亚像元定位精度和制图效率的需求，本研究首次引入像元聚集度概念模拟小尺寸地物空间分布；结合空间引力亚像元定位技术，开展基于像元聚集度的亚像元定位模型研究，使新模型在理论上兼顾两种尺寸地物，并进一步提高模型定位精度。将所提模型应用于我国重要农业区中高空间分辨率多光谱遥感影像进行模型验证，以期更好地突破遥感影像空间分辨率的限制。项目成果将有助于提高作物遥感分类与识别精度和水平，并为亚像元定位以及遥感影像分类提供模型理论依据和方法技术借鉴。

Crop remote sensing classification and identification is not only one of important research topics in agricultural remote sensing monitoring, but also is the basis of remote sensing spatial distribution information for accurately obtaining area, growth, yield, quality of crops and main agricultural hazard assessment. Mixed pixels generally exist in remote sensing images and sub-pixel mapping technology as a follow-up effective mean has great importance to solve the problem of mixed pixels and accurately extraction of crop remote sensing information. Aiming at the shortage that the existing models and algorithms are difficult to considerate two kinds of land cover with different size, and to further improve the sub-pixel mapping accuracy and model efficiency，this study firstly introduce pixel aggregation degree to simulate spatial distribution of small-size land cover; Combining with the spatial attraction sub-pixel mapping technology to carry out the study of sub-pixel mapping model based on pixel aggregation degree which can consider two kinds of land cover with different size and further improve the mapping accuracy of model. The proposed model can be applied to fine spatial resolution multispectral remote sensing images in important agricultural areas of China for model verification, so as to better break through the limitation of spatial resolution of remote sensing images. The results of this project will help to improve the accuracy and level of crop remote sensing classification and identification, and provide theoretical basis and technical reference for sub-pixel mapping and remote sensing image classification.