I have a map image, similar to this:
At 100%, the ocean is noisy and therefore not effectively compressed to png.
Please point me in a direction of an intelligent tool for the job. (I tried GIMP but failed.)
In GIMP, my tool of choice for this purpose is Filters → Blur → Selective Gaussian Blur:
With some tweaking of the parameters, it can give results very similar to what Lipis produced with Vector Magic. The basic rules are:
Here's the actual result I got with the settings shown above (radius = 15, max. delta = 25). It's fat from perfect, but still much better than a straight conversion to PNG. Most of the remaining noise is on the edges; simply scaling down the result to the same size as Lipis's version should hide most of it.
Ps. Converting the image to a reduced palette (with Image → Mode → Indexed, 16 colors, no dithering) helped shrink it a lot further, probably mainly by getting rid of subtle gradients remaining in the background. It's now down to 140 kB at full size (127 kB after PNGOUT) and looks like this. I did have to manually recolor a few spots that got desaturated in the process. Alternatively, just selecting the sea background with the color select tool and bucket filling it with a uniform color would probably accomplish much the same thing.
ImageAnalyzer has lossy PNG compressor that blurs the image as you lower the quality, so in this case it compresses the image pretty well.
However, you'll have better result if you de-noise the image or save noisy version as JPEG.
The only problem with vector magic is that takes your image and changes it all to vectors so you are begining to lose some of the detailing on the text as it converting it all to shapes.
If you have the original map file in either photoshop, illustrator ect first i would remove the noise on the sea. Which will save an awful lot of space as it will just become one clear blue shape. Then i would export it as a PNG as they tend to be excelent at line image, strong graphics ect, if doing that it should come out about the 40kb (of the top of my head) you can also further reduce that further using a png compressor, i use "png compressor" http://itunes.apple.com/gb/app/png-compressor/id434886325?mt=12