Research On Key Problems Of Quantificational Remote Sensing Based On Radiation Transfer Model

Due to the multi-sensor and multi-band character and the repetitive earth observation ability of remote sensing, data and images acquired by remote sensing are main supports for national resource survey, disaster monitoring, land use, basal mapping and update, et al. Along with the ceaseless enlargement and deepening of the application of remote sensing technology, new requirements of the remote sensing information quality have been put forward. This kind of requirements promotes the research of remote sensing quantification model and the research of exploring the technologies to exclude the uncertainty factors during quantitative information acquiring and extracting. This thesis, centering on the research hot spots mentioned above and mainly taking examples for Japanese ASTER data and CBERS (China-Brazil Erath Resource Satellite) data, selects the fundamental elements affecting the application of quantificational remote sensing, such as land surface condition, atmospheric condition and sensor band, carries out systemic theoretic exploration, model experiment and field measurement. Several notable achievements are listed below. 1, An improved thin cloud removal algorithm is brought forward. The traditional homomorphic filtering method is developed by replacing complicated frequency domain filtering with spatial domain filtering. The algorithm gets a good result in ASTER image process. 2, The improved homomorphic filtering method is combined with mean reflectance matching to improve CBERS-2 image quality. 3, Supported by the National Nature Science Foundation (No.40371086), image pixel adjacency effect measurement and correction algorithm experiment are carried out, and adjacency effect in synchronously acquired CBERS-2 CCD image is effectively corrected. 4, By combining adjacency effect field spectrum measurement and CBERS-2 satellite data, target pixel contribution rate is determined and the technology...