AutoTrace - Step 1: Raster preprocessing

AutoTrace a Background image - Step 1: Raster preprocessing Pro version only

AutoTracing a raster image is always a challenging task. There isn't here a perfect mathematical single solution for this problem.
The first step, preparing the raster image, is the most important task.
According with the type of background raster image we can apply different preprocessing strategy.
The goal is to get a cleaned raster image with the raster lines we want to trace thinned to just 1 pixel thickness.

Thresold: as the first internal stage of preprocessing is converting the raster image to a 2 color image (Black/White), we must specify the value that divide what we will be considered as White from what will be Black. This is the Thresold value. Its range goes from 0 to 255. The original pixel values are 0 for Full Black and 255 for Full White, so generally a good Thresold value could be 128 (the middle value). The application when this dialog starts compute a 'probably good' value for the Thresold, so generally you haven't to take care about it!

Erode: the second (optional) inner stage of raster preprocessing is the Erode/Dilate operation. You can specify if you want to apply it, and the value in pixel. A positive value will Erode the Black pixels, a negative value will dilate them. This is an optional operation, and most of times you haven't to take care of it, leaving it as disabled. You can use it if you are rather skilled about raster processing.

Enhancing: this is the most important operation in raster preprocessing. The possible options are at the third position because they are at the third internal stage in preprocessing. We will take an indeep look at the Enhancing operation below, with some examples

Apply Preprocessing: pressing this button you will start the preprocessing job. If you want to work just in a zone, enable the 'Ask for Zone' checkbox and you will be prompted to click the 2 vertices surrounding the zone you like. You can press more times this button, and every time the process will start using the original raster as source.

Show the original image combo box: after you execute the raster preprocessing, you can select one of these views:

Show the original image - you see the original image in Black/Gray and the Traced elements in Red
Show the preprocessed image - you see the preprocessed image in Black/Gray and the Traced elements in Red
Show the preproc. image (no traced pixels) - you see the preprocessed image in Black/Gray, but the already traced pixels are hidden. And the Traced elements in Red

You can adjust the transparency of the raster image scrolling the included cursor.

Now we want to explain better the Enhancing operation

As advised the goal is to get a cleaned raster image with the raster lines we want to trace thinned to just 1 pixel thickness.
In this way will be easier for the Autotracing operation to follow the right line.

This option will enhance the Border lines between Black and White pixels. The border will be traced on the outer Black pixels. This is the right option if you want to cut filled shapes, see the example below with the original and the Border enanched images:

This option will enhance the Border lines between Black and White pixel, but the enanched border is external to the Black shape. The border will be traced on the white pixels surroinding the Black ones. This may be the right option if you want to cut traced shapes, but they are poor drawn and/or include internal lines.
For example we want to trace the external shape of this former:

If we apply the first option, Enhance Borders, you will get this:

As you can see the lines are still not cleaned, as the internal and external enhanced border touches each other (the original line was just 1 pixel thick, and dirty). Applying the External Borders Enhance we will give this:

As you can see the result is now much better, as we can select among an external pure shape and an intenal contour with spar slots included.

This option will enhance the mean lines of Thick Lines. This is the right option if you want to trace a drawing and be sure to have thin lines. Here is an example with the original and the preprocessed images:

The lines in the first (original) image are 2 or 3 pixel wide, and in the second (preprocessed) image are thinned to 1 pixel.
Here below a Zoom of the same images:

This option will enhance the mean lines of CLOSED Thick Lines. It's the same of the previous Enhance ALL MEAN LINES , but here all the not closed lines are deleted. This is the right option if you want to trace a drawing and be sure to have thin closed lines. Here is an example with the original and the preprocessed images:

This option will not apply any enhancing. You can use this if you are sure about the 1 pixel tichiness of every line in the raster image. If some thick lines are included in the raster image, the Autotracing operation will give a very dirty result.