Analysis of the characteristic and orientation of linear array stereo imagery from satellite sensors

The study of the thesis is the orientation and analysis of linear array imagery from satellite platforms. Several types of space borne sensors have been considered for the production of new map products. SPOT data is at present the most commonly used type of imagery, mainly due to its high resolution CCD sensor, high stability, world coverage and availability. However, other types of linear array sensors have been under study for future space programs, and these are studied in this thesis. An overview of different methods of orientation of SPOT stereopairs is given and the advantages and disadvantages of these models are discussed. The weaknessess of across track stereo imagery obtained with SPOT data are pointed out and the use of along track stereo imagery is discussed. The applicability of the different orientation methods to other types of linear array imagery is studied. Existing methods being used for the orientation of along track linear stereo imagery are assessed and a critical view is presented. Based on this study, an orbital approach is developed for the orientation of along track stereo imagery. This model makes use of an additional parameter describing the relative positions of different images taken during the same orbit, reducing the number of control points needed by half. Data was simulated for the tests because no real data is yet available, and the simulation method is described. SPOT data was used as the raw data to simulate imagery of future along track linear array cameras. Tests were carried out to evaluate the preicision expected to be obtained with along track stereo imagery from future sensors such as ASTER, OMI and OPS. Tests on blocks of linear stereo imagery considering simultaneously across and along track stereo imagery were also carried out and the performance of the algorithm evaluated. The effect of the characteristics of the images forming a stereopair, and of the conditions of orientation of that stereopair, on the accuracy of the model after orientation was tested for several cases. High correlations between the orientation parameters result in problems in stability in the modelling algorithm. A study of the model variations due to small variations of the orientation parameters was made and weighting matrices were adopted to reduce the effect arising from these correlations. However, the use of weights did not improve the orientation in the case studies, as theoretically expected. Research is advised in this subject using data from future linear array sensors. Early tests carried out using conjugate points to reduce the number of control points required during the orientation process of SPOT imagery are presented. Similar tests were carried out with simulated along track stereo data. Problems of convergence were found when using conjugate points and a reduced number of control points during model orientation.