You'd have to somehow project your 3-D point cloud onto a 2-D plane so it can be compared to your 3-D image. But then some of the points are not visible to your camera photo so you'd have to figure out how to remove those from the projection. You could maybe do it by trial and error, varying the cone angle of projections, to see when the projected outline matches that of your image. Inspecting points and lines interior to the outline could be tough so I'd see if you can get away with just comparing the outline.
I think the best approach is, if it's really important to you, is to get a microCT X-ray 3-D image of your part and then compare that to your theoretically perfect point cloud model. There are 3-D point matching routines in programs like Avizo.
Next best approach would be to just snap images of a perfect version of your part at different orientations and then do the same for your worn parts. Use 2-D image analysis to do some basic gauging/caliper dimensions.
