Identify Background Pixel Colour

(This option is not available in the Freeware version)

A number of WinTopo's image processing functions have to make a determination as to which pixels are considered part of the image detail and which pixels are part of the background.

By default WinTopo determines the background pixels automatically. For the majority of images the default operation works very well, but some types of image may benefit from a different setting. For instance, images which have a coloured background, a darkended background, or where the background is darker than the image detail (for example, a blueprint).

The Background Pixel Identification feature is for three types of raster image...

  1. Raster images which do not have a white (or mainly light coloured) background. These can vectorise poorly, because WinTopo Pro has difficulty determining which pixels are part of lines and which ones are just part of the background. For instance, you may have an image with a dark blue background, and black lines. With the default processing you may end up with no vectors at all because WinTopo Pro couldn't pick out the black lines form the dark blue surroundings. The Identify Background Pixel Colour facility allows you to tell WinTopo Pro that the background is the dark blue colour - then it will process it properly.

  2. Raster images with white backgrounds, but very light coloured lines. The default processing may fail to extract many of the lines because they are so faint that they are assumed to be unimportant. You can use the Identify Background Pixel Colour facilty to tell WinTopo Pro that the background is the White colour and therefore everything darker is to be used in the vectorising process.

  3. Rasters with a background which is darker than the lines. The default processing will try to make vectors out of the dark background rather than the light lines! You can use the Identify Background Pixel Colour facility to indicate that the darker colours are the background and the lighter colours are the lines.

All of the above problems could be overcome by use of the Brightness, Contrast and Negative adjustments instead, but not without changing the colours of the lines on the drawing.

Accordingly, there is the Identify Background Pixels... option on the Image menu. Selecting this option brings up this window...

Identify Background Pixel Intensity Dialog

This window allows you to identify the range of pixel colours that are to be considered background (not part of the image details).

The range of colours extends from the shown colour (in the sample box and the numeric Red-Green-Blue (RGB) intensity values) through to white (RGB 255, 255, 255).

By selecting the darker setting instead of the lighter setting, the range can be inverted to cover the shown colour through to black (RGB 0, 0, 0). With this setting you can process an image which has a dark background with light coloured details (for example, a negative, blueprint, or screen grab).

Pick from Screen

You can select the cut-off colour for the range by clicking the Pick from screen button, and then click on a background pixel in your raster image. The RGB value of that pixel will be shown in the RGB boxes and a sample of the colour is shown in the sample box.

Tolerance

Often the background of a raster is not a unifom colour. This is especially true for JPG images. In general, if you pick the colour of one pixel from the background of the image, there will be other pixels on the background which are just outside of that colour range. Therefore, there is a Tolerance value to determine how accurate the picked colour is - i.e. the background is the specified colour including other pixels which are nearly the same colour.

A Tolerance value of 0 (zero) indicates that the specified colour is the absolute edge of the colour range. The default Tolerance is 20, which should be sufficient for most images where the background is fairly uniform. If you have an image with a significantly varying background colour range, you can increase the Tolerance value to catch a wider range of colours.

Why is it useful to identify the background colour?

Here is an example. If you have a greyscale or colour image where the image detail varies in intensity significantly, the default setting may not vectorise all the lines - the lighter intensity lines may get left out.

This is a sample of an image with significantly varying detail intensity. And next to it is the vectorisation result with the default Background Pixel setting.

Image with intensity variation    Default vector results

You can see that only the darker details from the raster image get converted to the vector drawing. This is because some of the line details are too light for the default operation..

Here is the vector result after selecting the background pixel colour with the Pick from screen option...

Better vector result

Which operations are affected by the background colour setting?

This setting affects nearly all of WinTopo Pro's image processing features...


Related Topics: