Final Activity Report Summary - LF-IRMAT (Optimization and exploitation of linear features for image registration and matching)
There is great interest in the European Community in generalising and standardise remote sensing data treatment, as is underlined by the activities of the EuroSDR and some of the projects being supported by the ESA. Geometric modelling and multi-sensor image fusion are to the benefit of basic information generation for infrastructure development, cartography, environmental monitoring, precision agriculture, mobile communication, transport and many other sectors on global level with increasing demand of spatial data.
Satellite image registration is one of the highlighted research topics in the last few years especially after the launch of several commercial high-resolution satellites. This is due to the absence of sensor calibration and satellite orbit information for most of the new high-resolution satellites. In dealing with this problem, the use of several point-based mathematical models is investigated and the effects of different factors that might affect the image registration process are studied. Appropriate empirical forms of mathematical models have been also proposed and procedures for the use of these empirical models have been recommended.
Combined with the use of point-based empirical mathematical models, the availability of defining linear features either by scanning/digitising existing hardcopy maps or by a terrestrial mobile mapping systems raises the importance of using such features as control features for satellite image registration, especially when GCPs do not exist or are difficult to be located. Build on that, the line based transformation model (LBTM) which can deal with linear features only, and/or linear features with number of ground control points is developed by the project researcher. The model is tested and consolidated through this project which helps in advancing the technology of information generation from remote sensing in particular where traditional methods of image registration cannot provide adequate support.
Satellite image registration is one of the highlighted research topics in the last few years especially after the launch of several commercial high-resolution satellites. This is due to the absence of sensor calibration and satellite orbit information for most of the new high-resolution satellites. In dealing with this problem, the use of several point-based mathematical models is investigated and the effects of different factors that might affect the image registration process are studied. Appropriate empirical forms of mathematical models have been also proposed and procedures for the use of these empirical models have been recommended.
Combined with the use of point-based empirical mathematical models, the availability of defining linear features either by scanning/digitising existing hardcopy maps or by a terrestrial mobile mapping systems raises the importance of using such features as control features for satellite image registration, especially when GCPs do not exist or are difficult to be located. Build on that, the line based transformation model (LBTM) which can deal with linear features only, and/or linear features with number of ground control points is developed by the project researcher. The model is tested and consolidated through this project which helps in advancing the technology of information generation from remote sensing in particular where traditional methods of image registration cannot provide adequate support.