中文摘要：

为探究江苏省冬小麦种植面积遥感提取的田块适宜尺度,选取冬小麦拔节期空间分辨率为2 m×2 m的GF-1全色影像和30 m×30 m的HJ-1多光谱影像进行研究。首先,将GF-1全色影像进行8 m和16 m重采样,形成3种不同尺度全色影像,然后分别与HJ-1多光谱影像进行主成分变换法融合,生成空间分辨率分别为2 m×2 m、8 m×8 m和16 m×16 m的3种不同尺度多光谱影像,并通过融合影像质量评价以及光谱特征值和非监督分类的面积精度进行比较。结果表明,3种尺度融合影像均值差异不大,分别为81.20、79.26和79.44。3种尺度融合影像的平均梯度值和标准差差异明显,其中16 m×16 m融合影像显著高于2 m×2 m、8 m×8 m融合影像,融合影像质量较原始多光谱影像质量发生明显改善,突出了丰富的植被光谱信息。从3种融合影像冬小麦种植面积提取精度可以看出,16 m×16 m融合影像冬小麦种植面积的提取精度较理想,达到96.73%。研究结果说明,16 m×16 m空间分辨率遥感影像较为适合江苏省冬小麦种植的田块分布特征,有利于冬小麦种植面积遥感准确提取。

英文摘要：

To develop a suitable scale of remote sensing extraction plots of winter wheat planting area in Jiangsu province, spatial resolution of 2 m× 2 m panchromatic image of GF?1 and 30 m× 30 m multispectral images of HJ?1 at jointing stage were selected for study. Three different scales of full?color images resampled from the panchromatic image of GF?1 were fused with the multi?spectral image of HJ?1 respectively using principal component transformation method to generate three different scales of multispectral images with the spatial resolutions of 2 m× 2 m, 8 m× 8 m and 16 m× 16 m. The image quality, characteristic value of spectrum and the area accuracy based on non?supervised classification were compared. The average values of three scales of fusion images were 81?20, 79?26 and 79?44 indicative of a close brightness. The fusion image of 16 m×16 m showed higher gradient value and the standard deviation, sugges?tive of improved quality of fusion image and enriched veg? etation spectral information. The accuracy of the fusion image of 16 m× 16 m reached 96?73%. It was concluded that, 16 m×16 m spatial resolution remote sensing image is favorable for the extraction of winter wheat planting area in Jiangsu by remote sensing.