[gdal-dev] GDAL Warp API

[Jim Klassen]

I'm trying to do some aerial photo processing specifically converting RAW imagery to DOQs. Right now I have working code that uses GDAL to read/write the rasters and uses my own code to do the actual image processing. I would like to use the Warp API so I can take advantage of the ChunkAndWarpImage method, in particular, it's ability to break the problem down into chunks that will fit into memory.

The problem I am facing is I can't write the part of the transformer function that converts from destination to source coordinates because the destination coordinate could occur nearly anywhere in the source image and likely will occur more than once. (The source data is from a push broom sensor so each scan line in the image has a different transformation depending on where the plane was and how it was orientated when the scan line was recorded.) From what I can gather, the Warp API is doing this to determine which part(s) of the source image it needs to grab to fill the destination image. In my case, I know that I will want to use the entire source image and have already calculated from the source image the extents destination image, but I don't see a way to tell this to the Warp API.

So my question: Is there a way to get the Warp API to skip trying to discover the required source image extents or is there a way to write the transformer function to trick the Warp API into just using the entire source image without the transformation having to be mathematically rigorous?

Jim Klassen