基于NDVI和深度学习的多源影像冬小麦提取研究

Research on Winter Wheat Extraction from Multi-Source Images Based on NDVI and Deep Learning

冬小麦是我国重要的粮食作物之一,准确提取冬小麦种植区域范围对保证粮食安全具有重要意义。文章以山东省威海乳山市为研究区域,使用GF-1C、GF-6、ZY-3多源影像数据,提取乳山市冬小麦种植范围。根据乳山市冬小麦种植及生长情况,选取4月中旬至5月上旬最佳时期的卫星影像;用计算NDVI作为新波段替代红光波段与绿、蓝波段进行合成,更加突出植被信息;利用深度学习训练冬小麦提取模型,实现冬小麦种植范围的自动提取,提取精度为94.39%,效果较好。

Winter wheat is one of the important food crops in China.It is great significance to extract the range of planting area of winter wheat accurately for ensuring food security.In this paper,we take Rushan City,Weihai City,Shandong Province as the research area,and use GF-1C,GF-6,ZY-3 multi-source image data to extract the planting range of winter wheat in Rushan City.According to the planting and growth of winter wheat in Rushan city,satellite images of the best period from mid-April to early May are selected;calculate NDVI as a new band to replace the red band and the green and blue band to synthesize,more prominent vegetation information;deep learning is used to train the winter wheat extraction model to realize automatic extraction of the planting range of winter wheat with an extraction accuracy of 94.39%,it has a good effect.