图像增强匹配在MISR多角度影像云迹风反演中的研究

遥感影像云层和雪层的立体匹配是云高解算、云迹风反演和云雪分离的重要前提。密集且匹配精确的点集可为后续研究提供更高的精度保障,然而由于云雪层灰度值偏高、细节信息不突出的特点,高密度和高精度的匹配效果难以获得。采用Terra-MISR多角度AN、AF、AA影像,对所选实验区分别进行分段线性和高斯低通滤波两种方法的增强,并对三影像共同匹配,最终两组增强影像获得的匹配密度和精度较原始影像均大幅上升。在此基础上,还提出了增强合并法,并通过云迹风反演的实例证明了该方法可将风产品分辨率提高至825 m左右的初步可行性,实现了对小云团及大云团内部水平风信息的精细化监测。

Image Enhancement Matching in the Research of MISR Multi-Angle Image Cloud Motion Wind Inversion

Stereo matching of cloud and snow layer in remote sensing images is an important prerequisite for cloud height solver, cloud motion wind inversio and cloud and snow separation. Dense and accurate matching point set can provide higher accuracy guarantee for the follow-up study. However, due to the cloud and snow layers have the characteristics of high gray value and inconspicuous detailed information, the high density and precision matching effect is difficult to obtain. This article used multi-angle images, AN, AF and AA, of the Terra-MISR, and applied two methods, piecewise linear and Gaussian low-pass filtering, respectively to enhance the selected experimental areas, and matched the three images together. Then their final matching density and accuracy both raised significantly. On this basis, the article also proposed the enhancement consolidation method, and through the cloud motion wind inversion, the preliminary feasibility of the approach, which can increase the resolution of the wind products to about 825m, was demonstrated, which realized meticulous monitoring for horizontal wind information in small and large clouds.