Line-of-sight geometric models can be established for an image and used in orthorectification. The steps involved are:

• Import the image from a standard image format and associate the geometric model with the image.
• Set the output frame parameters. This includes the choice of coordinate system (map projection), bounding rectangle and pixel size for the rectified image. The system can automatically calculate and set the minimum frame that completely contains the rectified image.
• If the image is to be orthorectified then associate a DEM with the image. Otherwise select a reference elevation as the reference surface.
• Compute locations of output pixels in the input image. This inverse line-of-sight transform is not performed for every pixel, as computationally this would be too costly. Instead, it is performed through the use of a regular grid over the image. The density of the grid is adjusted to be small enough that a bilinear interpolation of positions within a grid cell is sufficiently accurate.
• Resample the input image by interpolating the non-integer positions of the output pixels. The interpolation method is selectable. Standard methods, such as cubic convolution, bilinear and nearest neighbour are directly selectable, but any other interpolation/filtering kernel can easily be added to the system in the form of tables of coefficients.
• Save the rectified image to disk in the selected image format.