I am often asked why isn't it possible just to take a scanned image of a map and upload it directly into a GPS. The reason is that a scanned image is just a collection of coloured dots. Even if you were able to upload it to your GPS, if you then zoomed in, all you would see is dots.
A GPS needs a different sort of image to tell it things like "the A30 runs from this coordinate then to this, then to this" etc. That's why you have to trace over the features you want in your map.
Couldn't a program do all that for you? It's often been discussed but personally I don't think it's feasible. Is that diagonal line a crease in the map or the M4 motorway? Is that brown splodge a coffee stain or Greater Manchester? Does the label "Newbury" refer to the dot below, above or left of the writing? There would be so many things an automatic conversion program got wrong that it would take more time and effort to tidy up the map after the event than it would have taken to trace it out manually in the first place.
On the other hand it should be possible for software to do something to reduce the work involved in tracing out a map. This software assistance has been long overdue in the DIY mapping arena but it's here at last. Ladies and gentlemen I am proud to present to you...
Automapit works if you have a background image in which there are lines (roads, rivers etc.) and/or regions (lakes, woods etc.) which are clearly distinguishable from their surroundings by colour.
With Automapit you still start by clicking on the object, which allows you to give it a name and say what sort of thing it is (road, river etc.) but you only have to click once for a region or twice for a line (to specify the initial direction). Then hit Ctrl&A and watch while MapMan automatically follows the line, or traces the boundary of the region, based on the colours in your background image.