Measuring object width from 3D image without reference point

I have a set of three 2D images of the same location. At these locations are a number of trees which I want to measure the width of. The 2D images are all from slightly different viewpoints. The trees are at different depth-of-field within the image. The key problem is that there is no reference object within the images. I have topographic data set that descibes the location in x,y,z.

Without a reference object I can't use triangulation. I don't think image segmentation is useful due to the depth-of-field (unsure - plus I'd need a way to relate the pixel count to real-world coordinates). I could use structure-from motion, but again this needs to be related to a real-world coordinate system.

Key questions:

• Is there a way to project my topography data onto the images without using a reference point? The only way I can think of is by roughly aligning the image topography to the data, but again I don't know if there's a function that could do this.
• Can image segmentation account for depth-of-field, and can pixel count be related to a real-world coordinate system?

Thanks!