hello, i am wondering about the Structure from Motion example that is provided on MathWorks. It says that scale factor cannot be determined. Does this mean that i cannot use this to make a 3-D object with a known real word size? What is scale factor?

hello, i am wondering about the Structure from Motion example that is provided on MathWorks. It says that scale factor cannot be determined. Does this mean that i cannot use this to make a 3-D object with a known real word size? What is scale factor?

My application is that i wish creat a 3-d object of a machine of mine, from which i can determine the exact locations of certain points on the machine. The points on the machine are marked with retroflective dots, with hopes that this will help thecomputer recognize the points

Answer 1

[Steven Lord]

If I were to show you a picture of a car (nothing else in frame; no person, no buildings, nothing) could you tell me for certain if it was a real automobile that a human being could get into and drive or if it was a toy? Without some additional information to tell you whether the car was full size or a 1/64 scale model as shown in the first image on that Wikipedia page, ideally you couldn't tell. [In reality the toy is unlikely to be as detailed as the original, but it could be.]

Now if there was a ruler next to the car, or something of a known size, you could use that information to distinguish the toy car and the real car. You need to relate one pixel to a real-world measurement. That's the scaling factor.

[Ben Anstrom]

Thank you Steven,

So if I know that the focal length is 16mm, and the sensor width is is 23.5mm, as well as a 4000x6000 pixel count, might this be enough information to make a scale?

Or alternatively, how might I go about having the computer vision toolbox implement a scale by communicating to it that a particular retroflective dot is 8mm in diameter?