cp2tform structure

Hello,

I want to map 2 images into each other using cp2tform. Main problem is that I've got a lot of point to point correspondencies (more than 10000) and cp2tform will give me an error when i want to use lwm cause it finds to mauch collinear points (its a grid). I want to use picewies linear now (cause its faster and no error) but i get more artefacts especialy at the edges where i've got no points. I want to find these sections using the tform. Can someone give me information about the structure of the tform, especially tform.tdata?