论文部分内容阅读
针对几何精校正过程中人工选取控制点误差大、未考虑高光谱数据光谱特征一致性等问题,提出了基于SIFT特征的自动几何精校正方法。首先提取图像的SIFT特征,利用高光谱数据的地理坐标定位进行局部特征匹配,然后为了进一步提取高精度、分布均匀的控制点,提出了一种分区域的随机采样一致(Random Sample Consensus,RANSAC)算法。利用航空高光谱成像仪Hymap获取的新疆东天山数据进行算法性能的分析与验证,并采用CE90/CE95以及均方根误差等指标进行定位精度的评价,提出的基于SIFT特征的自动几何精校正方法能够达到0.8像元的定位精度,并且校正前后光谱的光谱角小于0.01 rad。
Aiming at the problems such as the error of artificial selection control points during geometric precision calibration and the inconsistency of spectral features of hyperspectral data, the automatic geometric precision correction method based on SIFT features is proposed. Firstly, the SIFT feature of the image is extracted, and the local feature matching is performed by using the geographical coordinates of the hyperspectral data. In order to further extract high-precision and evenly distributed control points, a sub-region Random Sample Consensus (RANSAC) algorithm. The performance of the algorithm was analyzed and verified by using the data from the Hymap of Xinjiang Hymap, and the CE90 / CE95 and root mean square error (RMSE) were used to evaluate the positioning accuracy. The proposed method was based on SIFT The positioning accuracy of 0.8 pixels can be achieved, and the spectral angle of the spectrum before and after correction is less than 0.01 rad.