[gdal-dev] Calibration/georeferencing and warping of map with unknown projection

[contact at sharemap.org]

Hello,
Project ShareMap[1]  uses gdal_warp on the server side to allow users 
to work with calibrated rasters maps as overlay on web mercator maps. 
In case of maps in large scale it performas relatively well (country 
maps - almost ok, provinces maps - perfect)[2]. The problem is if we 
try to calibrate small scale maps (like this historical europe map) 
using only GCP.

This map[3] is a good example - if we split it to separate countries 
and calibrate every country alone, result is ok, but I don't know how 
to calibrate entire map with only GCP's (map does not contains 
informations about projection)

ShareMap currently execute such commands to warp map to web mercator:

gdal_translate -of GTiff -gcp 281.0 1290.0 -606604.2360752566 
4290256.933470774 -gcp 832.0 1512.0 1213208.4721505132 
3982062.8778166603 -gcp 1722.0 1668.0 3896453.822854374 
3441500.2881376203 -gcp 1436.0 1021.0 3226253.9813841325 
5618426.55426589 -gcp 653.0 701.0 168772.95277900095 6670199.918799774 
map.jpg map.referenced.tif

Values of GCP are in in meters for WebMercator - but in fact this map 
does not look like mercator projection.

Next step is warping

gdalwarp -t_srs +proj=merc +a=6378137 +b=6378137 +lat_ts=0.0 
+lon_0=0.0 +x_0=0.0 +y_0=0 +k=1.0 +units=m +nadgrids=@null +no_defs 
map.referenced.tif map.referenced_warped.tif

My question is very short - what is the best way of georeferencing and 
transforming to web mercator map with unknown projection, where we have 
several (4 or more) pairs of LAT/LNG and X/Y of pixel.

LAT;LNG;X;Y
35.92464;-5.44921;281;1290
33.65120;10.89843;832;1512
29.51610;35.00244;1722;1668
44.98033;28.98193;1436;1021


[1] ShareMap - ShareMap.org is new web mapping platform that allows 
author to create maps on free sources using data from OpenStreetMaps 
and trace from calibrated raster maps
[2] See this movie for example - www.youtu.be/gPJ7AD15wRI
[3] 
http://www.romanscotland.org.uk/pages/infrastructure/maps/RE395adL.jpg