论文部分内容阅读
遥感数据从时间尺度、空间尺度为人类生存的环境要素提取提供了有力的支持。遥感数据资源的重要性日益突出,数据积累周期越长其存在的潜在价值就越大,但是随着时间的推移,遥感数据也会随着传感器的衰退或损坏导致数据质量降低、无效数据等情况。针对该现象提出了一种自适应的多元拟合方法来修复光学遥感数据由于传感器损坏导致的无效观测;该方法通过考虑地物波谱反射特性实现坏像元修复,基于地物波谱在不同通道上的反射特性之间的关系,构建波谱自适应筛选因子来优化波段反射率修复模型,从而精确地修复坏像元的地表反射率。基于该方法,修复了Terra Moderate Resolution Imaging Spectroradiometer(MODIS)5波段数据在北京地区的无效观测(传感器坏像元);验证表明该方法精度统计指标RMSE为0.011,相比其他方法最好的结果提升大约8%,R~2达到0.82。整体上,该方法不但具有较好的精度,且具有更好的适用性。
Remote sensing data provide strong support for the extraction of environmental elements of human existence from the time scale and the spatial scale. The importance of remote sensing data resources has become increasingly prominent. The longer the data accumulation period, the greater the potential value of its existence. However, with the passage of time, remote sensing data will also cause data quality degradation and invalid data as sensors deteriorate or deteriorate . Aiming at this phenomenon, an adaptive multivariate fitting method is proposed to repair invalid observations of optical remote sensing data due to sensor damage. This method achieves bad pixel restoration by considering the spectral reflection characteristics of the object. Based on the feature spectrum of the object, The spectral adaptive filtering factor was constructed to optimize the band reflectivity restoration model, so as to accurately repair the bad pixel reflectivity. Based on this method, the invalid observation (sensor bad pixel) of Terra Moderate Resolution Imaging Spectroradiometer (MODIS) 5-band data in Beijing area was repaired. The verification showed that the RMSE of the method was 0.011, which was the best compared with other methods About 8%, R ~ 2 reaches 0.82. As a whole, this method not only has better accuracy, but also has better applicability.