Wednesday 11 May 2022

Orthorectification in a Nutshell

Raw satellite imagery has distortions caused by sensor orientation, variations in topography, and the earth’s curvature. After collecting satellite imagery, it has to be processed to filter out inaccuracies through a process known as orthorectification of aerial photographs. It involves the removal of image perspective (tilt) and relief (terrain) effects for creating a planimetrically correct image.

The orthorectified image contains a constant scale representing features in their accurate positions. It enables the measurement of distances, angles, and areas accurately.

The inputs required for orthorectification are:

1.     Image with accurate sensor geometry

2.     An elevation model of the ground surface in DEM, DTED, or SRIM format.

Imagery Distortions

The distortions come in the form of feature displacement, creating inconsistencies in scaling via image. The distortion levels change throughout the imagery scene. It means that a simple ground control point adjustment cannot nullify these errors. Therefore, a correction process for topographical variations and sensor orientation is necessary to rearrange the displaced pixels to their correct position. The final orthorectified image can be used for direct measurement, extracting features, and other applications that need measurement-ready data.

Accurate Elevation Models Are Crucial

Feature distortion on raw imagery is greatly affected by terrain variation. An accurate elevation model is needed        to calculate the terrain variation effects on the image pixels. This calculation is therefore used to position the pixels to the correct locations. There are free terrain data sources present, such as SRTM and ASTER.  

It is often that the importance of using accurate elevation models for orthorectification goes unnoticed. The reason is the high costs of procuring accurate elevation data and investing in hardware, software, data and skilled professionals. But these things should not be given higher value than quality elevation models. If you focus on orthorectification and geo rectification of satellite imagery in your practice, it needs to be detailed and reliable.