Image Georectification / Georeferencing

Georeference is the process of defining somethings existence in physical space or in other words establishing its location in terms coordinate systems representing earth defined through projection systems. It is used when establishing the relation between raster or vector images by determining the spatial location of the geographical features. This procedure is mandatory for data modeling in the field of geographic information systems (GIS). When data from different sources or time periods (like time series satellite images) need to be combined and then used in a GIS application, (eg. for change detection, assess damages after a natural disaster etc.), it becomes essential to have a common referencing system. One can georeference a set of points, lines, polygons, images, or 3D structures (eg. shape file georeferencing). This is brought about by using various georeferencing techniques. Most georeferencing tasks are undertaken either because the user wants to produce a new map or because they want to link two or more different datasets to the same geographic locations or mapping new features or updates in an existing data.

For georeferencing (other wise referred as registration), it needs to establish control points, input the known geographic coordinates of these control points, choose the coordinate system and other projection parameters and then minimize residuals. Residuals are the difference between the actual coordinates of the control points and the coordinates predicted by the geographic model created using the control points. The residuals help in determining the level of accuracy of the process. In the case of images (Raster georeferencing), in the established position the image may be resampled to special GIS file formats (like GeoTiff, MrSID) or be accompanied by a world file (contains georeferenced information including rotation) when stored in common graphic formats (like tif, jpg, png).

For aerial images and satellite images if the orthorectification is not possible due to any unforeseen reasons or inadequate supporting datasets, as an alternative, direct georeferencing can be employed to serve the purpose to a certain extent. For example time series satellite imageries needs coregistration and subpixel correlation among the layers to achieve certain analysis tasks, especially when analysing vegetation status using CIR (Colour Infra Red) or NIR (Near Infra Red) imagery.

Colour infrared (CIR) enhances aerial images of vegetation by colouring the green vegetation red in colour. Red colour feature of the vegetation is because of the infrared ray emitted back to the atmosphere by the leaf pigments, after receiving the sunlight. CIR orthophotos are generally used to determine the area of vegetation.