Background Image Concepts

(This functionality is available with the Background Image module)

LISCAD supports raster image files as a background to the vector data objects. Background images, such as maps, plans and orthophotos can be used to digitise information from, or they can be used as a location guide in the case or aerial photographs.

Most image formats are supported, including georeferenced files in GeoTIFF, MrSid and ER Mapper ECW format.

Before an image can be used in the background it must be registered with the grid system used by the SEE file. Registration requires control pixels in the raster image to be assigned co-ordinates that accord with the map datum being used. The original raster image is then transformed to fit the map datum and the transformed image can then be displayed in the background that registers with the overlying vector data. This process may also be referred to as georeferencing the image.

If an image is being imported that contains georeference information this will be read. If the georeferenced image is on a grid, 2 control points will appear at the top left and bottom right corners of the image. Otherwise the image will have 9 control points across the image. Although an image is georeferenced it still needs to go through the registration process to be imported into LISCAD.

When an image is registered and transformed to match the LISCAD data base it is stored internally in ECW format to keep the image size to a minimum.

Four image transformation techniques are available for selection by the user, however to achieve the best results the correct transformation type should be used for the type of raster image selected.

Conformal Transformation:

This transformation method requires a minimum of 2 control points.

The conformal transformation should only be used where the original raster image is based upon an orthogonal map grid with minimal distortion.

Affine Transformation:

This transformation method requires a minimum of 3 control points.

The affine transformation is best used where the original raster image is based upon an orthogonal map grid which may have systematic scale and axis rotation distortion. It can also be used effectively where the original raster image is based upon a latitude / longitude grid or on orthophoto map images.

2nd Order Polynomial Transformation:

This transformation method requires a minimum of 6 control points.

Polynomial transformations should be used where the original raster image has distortions other than just uniform scale distortions in the x and y directions and a rotation distortion between the x and y axes. It can be used to remove distortions such as aircraft tilt in an aerial photograph as such distortions are systematic and apply over the entire image. Polynomial transformations cannot be used to remove random errors over an image due to terrain relief distortion as these are not mathematical distortions and vary at any point in the image. Polynomial transformations are best used on aerial photographs with minimal elevation differences across the image.

3rd Order Polynomial Transformation:

This transformation method requires a minimum of 11 control points.

The third order polynomial transformation can be used where more control points are available to achieve a better determination of the distortions. The same conditions apply to 3rd order polynomials as to 2nd order polynomials. Although the additional control points of the 3rd order polynomial will generally give lower residuals at the control points than the 2nd order polynomial it may be at the expense of accuracy between the control points, depending upon the image.

Note: Polynomial transformations should not generally be used on images where the distortions are essentially just scale and rotation. Results can be very unpredictable if a polynomial transformation is used on a conformal or map projection image. As there are minimal errors in these types of images the mathematics of the polynomial can become unstable and either fail or grossly distort the image. Remember polynomial transformations like distortions!

Using a conformal or affine transformation on an image that has considerable distortion will not fail but will result in large residuals at the control points.

It is the users responsibility to ensure that the resultant transformed image registers with data base within acceptable tolerances at any point within the image.

LISCAD does not rectify aerial photograph images.

When an image is georeferenced with the grid system the transformed image is copied into the SEE data file. Information about the original image file is maintained by the system but the original image is not copied into the SEE file. If at any time you wish to check the image registration the original image file must be available. By default LISCAD will look for the file where it found it last time. If in the mean time you have moved, renamed or deleted the original image you will be asked to browse to where the original image is that you want to use.

When an original image is selected as a background image in LISCAD it may be that not all of the image is required to be seen. E.g. the border around a map showing the grid co-ordinate values. At the time of registration it is possible to select four points to define an area so that only that part of the image within the four points is used.

LISCAD can use as many background images as you want at any one time. Performance will vary considerably on different computers depending upon the operating system, type of processor and graphics card installed, hard disk speed, amount of memory and the number and size of the images being used.

When multiple images are being displayed concurrently, some images may overlap others. Display Features will allow you to control which images are displayed at any given time as well as the order the images are displayed in.

Images can be obtained from many commercial outlets or alternatively it is possible to scan your own images. Even if you only have an A4 scanner it is possible to scan large maps in A4 sections and provided there is overlap between all the sections they can be merged together accurately into a single image using commercially available image stitching software. This software is freeware or very reasonably priced and can be sourced by searching for image stitching software on the internet.