| The technical complexity of the present remote sensing systems, the different levels of processing involved to obtain the geophysics parameters, the interest in multi-temporal and multi-sensorial studies and the growing request of accuracy and temporal resolution in satellite measurements, requires the establishment of a hierarchy of processes that allow the generation of operational products and the development of processing algorithms characterised by its precision, autonomy and efficiency.; The main work in this thesis has been focussed on the development of a systematic methodology to NOAA AVHRR images, enabling the present users to link to the research in oceanographic physics, as well as future researchers, to make use of a set of automatic and efficient tools in order to guarantee obtaining high precision operational products.; Within the different levels of processing, special emphasis has been placed in the procedures for atmospheric and geometric corrections. In the frame of the atmospheric correction, the day and night cloud detection algorithms been modified and adapted. Also all the aspects related to the regional correction and optimization terms of the coefficients of a non linear split-window algorithm have been studied.; In the context of the goal of geometric correction, an orbital Keplerian model for the NOAA satellites has been proposed. To compensate the residual geometric errors, two new techniques based on contours optimization have been investigated. Both techniques model the corrections directly in the image domain without an explicit identification of the distortion sources. Results of such techniques, applied to images having big areas with partial and/or total occlusions, have been presented, assessing the precision from a new method relying on a field of error contours.; Finally, towards the goal of comparing and integrating, in the space-time frame, of multi-sensorial data, images received from the SeaWiFS sensor have been processed, evaluated and corrected. In a comparative context, different examples of oceanographic structures have confirmed the exportability and accuracy of the proposed techniques.; In summary, the work performed in this thesis provides image processing tools that improve the precision of the data analysis and interpretation, enabling its application to future Earth observation sensors or other applications of the remote sensing technology, as can be the multi-temporal classification, the multi-sensorial data fusion, the movement detection and the structures recognition. |
