As far as we now, there is no program (not yet) that can download Google Earth’s streets in vector format. Although this task can be done using Open Street Maps; what a pity! it doesn’t exist all the cities in this software.
But if someone is interested in Google Earth’s streets; the solution would be first, downloading them like images, and then, working hard and tirelessly (Spanish idiom: “a lo bestia”, synonym: to work like a beast) in converting them to vector format. Here I leave some tips to decrease the beast levels:
1. Putting a blackbackground image
We do so, in order to get that the satellite image does not obstruct streets visibility. This is done creating a black image in bmp format using Mspaint, then this image is upload to Google Earth, stretching it over the area of interest.
2. Downloading the image with Stitchmaps
Now, using Stitchmaps, we choose a mosaic allowing us to see texts with a thickness lower than street vector’s.
Note that, even if not all the streets can be seen at image height in Google Earth; in Stitchmaps, we can view all of them, because we choose a low height, in this case the value is 384 meters.
Once the mosaic is defined, a download command is given, waiting till this object is ready. Finally, we save it for TIFF format including calibration for OziExplorer (.map). The image appears like the figure shown below: the one in the right is a zoom view:
As an additional note, if we want to convert it to .ecw format, first we need to upload the file to Global Mapper, second, a projection is assigned and then needs to be rectify from a .map file. After this, you can export it to an .ecw format for better handling in another program.
3. Digitizing with a tracing program
Drawing a line could be annoying, so if you want to move fast, you can use an automatic layout program, such as, for example, Microstation Descartes.
It is understood that the image, which is in .ecw format, is georeferenced (although you could do this using Descartes); what comes next, is to convert the image to vector using the same procedure shown in a previous post.
A mask is done for the yellow tones, and another one is created for the gray ones. After this, it must be converted to vector with topological cleaning. In the segment where the text is placed, the vector won’t be created; we will have to make the union manually; though if you want to take advantage of Descartes, it’s possible to convert all of these text tones to the gray tone used in the streets, that’s why we did it smaller. If the text is going to be vectorized, it is used the oriented text command.
4. In case of not having MicroStation Descartes
The same work can be done with these softwares: AutoDesk Raster Design, ArcScan, Manifold GIS, and even Corel Trace.