Imagery transformation and synthesis

[anonymous]

We know that there are errors in sattelite maps due to physical principles of imaging and uncontrolable environment conditions.
1 The idea is to find correspondence of osm traced contours to contours detected on sattelite images of multiple providers, average the coordinates got from each images and get the reference coordinates of features of contours which will have reduced error if we assume that errors of different maps are ranfom. Then for each map (including osm) create a smooth function describing displacements its coordinates from reference coordinates and put it on public server.

2 Since we have a set of such functions we can align all the images with each other and OSM and try to synthsize the image which is better than any image from all the maps alone.

[Klumbumbus]

Some aerial imageries (e.g. from governmental survey offices) have a pretty exact alignment and the mean value of gps sources and other aerial imageries would result in a worse alignment.

[Don-vip]

You're asking for orthorectification. This is way too complex for JOSM, it requires a lot of data, CPU and time and could not be made in real time.

[Hb---]

Replying to anonym:

We know that there are errors in sattelite maps due to physical principles of imaging and uncontrolable environment conditions.
1 ... Then for each map (including osm) create a smooth function describing displacements its coordinates from reference coordinates and put it on public server.

Mind the message of this step. It is like OSM would publicly say: We do know that our map data is displaced by around 8 or somewhat meters – and were not going to correct that! But we are gracious enough to give you the data how to correct it by yourself on your system.

Sounds crazy to me.