Converting from state plane to geographic projections

[Guest JohnC]

attn Richard Hill....here is the conversion/derivation...Assume you have a state plane projection Tif, with units in feet, say 8,000 x 8,000 pixels.LizardTech.com's utility (mrsidgeodecode_win.exe) extracts blocks of data from a large MrSid state plane projected database.Sample code in a command window:mrsidgeodecode_win.exe -i Franklin_county.sid -o Pic0101.tif -ulxy 2373 7800 -wh 8000 8000 -of tifgIt will look like a Geographic projection, until you assemble a large number of them together and find there is serious drift over many miles so that coasts and airports no longer line up with photoreal.There must be minor warping and rotation differences.Consider the attached drawing below:Inside the rectangular state plane image (light blue), there exists a geographic image (gray) defined by constant loci of latitude (black) and longitude( red) drawn to deliberately pass through each of the state plane image corners. They are likely not perfectly parallel or orthoganal lines.If we can calculate the pixel locations of all four corners of this gray geographic subset, it is a simple task in a good paint program to drag (green arrows) each of those pixel locations to the four original corners of the state plane image, eliminating a very small part of the pixels in a triangular wedge along each boundary. (The PSP 5 distortion tool does this easily). The result would be a geographic projection from 99% of the original image because it is now bounded by constant lat and constant lon lines.The four corners of the geographic subset x,y coordinates can be calculated directly from the known state plane projection coordinates.For one example, the geo corner labelled A has (1) Latitude = the NW state plane latitude, and (2) Longitude = the SW state plane longitude. So we get the inner coordinates directly, and converting the coordinates to pixels with the known feet/pixel value then produces exact pixel locations to be dragged (green arrows) to the orthoganal state plane corners. Converting them to lat and lon in degrees with Corpscon 6 provides the information for your inf file.And there you have it. An image with constant lat and lon along its edges, with each pixel having been moved proportionately with the distortion tool in PSP..A real world example from my images would have an 8000 x 8000 pixel state plane image have the four corners typically adjusted by 60 pixels or less, essentially preserving approx 99% of the original image.The entire calculation set can be done in a few cells in Excel, and this preserves the calculation record.Doing this the first time placed a photoreal airport runway image directly under the high resolution default runway there, with the runways at exactly the same headings and angle.48000 meters East, the North South roads I had in photoreal fell within a simple road width of the same Microsoft default roads, a minor tweak at best.The pixel adjustment varies as you move East and South so you are continuously adjusting for any second tier non-linear effects. For example, the NW corners of my pixel adjustment to geographic projection changed from 74(0.92%) to 54(0.68%) to 34(0.425%) pixels every 16000 meters Eastward.Try it and see if it fits your needs. Globalmapper has many happy customers and can do all this and more with with one or two mouse clicks, but it isn't a free utility.http://forums.avsim.net/user_files/174674.gif