Maybe I am too much of a skeptic, but when you see so many image search/recognition/match etc programs that don’t work you should ask yourself why? The reason is: the 3D to 2D conversion - information is lost when you represent 3D objects as 2D images. It is what’s called an “ill-posed problem”: a question that has many different answers. The same 2D image can come from two different 3D objects. This is what so many optical illusions are based on (consider Necker cube). Is it a ball or is it a concavity? If the light comes from the right it is convex, if from the left concave. Here’s a good example of that.