Research On Observation Data Reconstruction By Temporal Spatial And Spectral Complementary Information Fusion

With the developments in space observation technology, we can take advantage of a variety of different ways to obtain earth surface spatial information. But in the observation process, due to the limitations of observation mode, the environment of observation, the failure of observation platform and other factors, the obtained data is often spatial discontinuity. This "space gap" seriously affected the subsequent applications. Therefore, it is rather meaningful to exploring how to eliminate the invalid information in observed data and obtain spatial seamless observation data.The main contents are as follows:(1) Reconstruction remote sensing data using spectral complementary information fusionIn the observation process using hyperspectral or multispectral sensors, the obtained image may be in some bands due to failure or noise in several detectors. In this research, using the redundancy in spectral image data; the relationship between bands is explored. Based on the similarities between multi-band images, the spatially discontinuous images are reconstructed.(2) Reconstruction remote sensing data using temporal complementary information fusionIn some cases the observations are affected by the failure of sensor plantform or the observation environment, the obtained images are spatially discontinuous. Based on the analysis of difference between multi-temporal data, a similar information extraction method is proposed. It can overcome the negative effects caused by complex scenes and so on. Then the temporal complementary informations are fused in the spatially discontinuous region.(3) Reconstruction remote sensing and in-situ data using spatial complementary information fusionSome data are measured in ground stations. Although the measurements are well calibrated, they are limited in space, and inadequate to support applications on large spatial scales. But the remote sensed information is also limited to the retrieval accuracy. To solve this problem, this paper analyzes the interaction of different factors in imaging process, build the site-satellite correlation model. Then a geostatistics based fusion method is studied and used on atmospheric pollutants monitoring.