I'm planning a concept for calibrating a robot model. For this purpose I want to use computer vision approaches. Since I'm not that experienced in that field I have some basic questions on the accuracy of NON STEREO approaches like the famous Aruco markers aruco marker description.

Using a simple pin hole camera model where $x_{img} = \frac{f_x}{z_C} x_C $ where $C$ denotes the cameras coordinate frame.

I know the size of a pixel of my camera ( $\Delta x_{img}\approx10\mathrm{\mu m}$) and the focal length ($f_x\approx664\mathrm{m}$). Somehow I'm having problems to estimate the accuracy I can expect. Especially in the depth ($z_C$) direction.

My fist approach was to use the distance between two pixels in the image frame to calculate a distance in z-direction but i) this depends on the $x_C$ coordinate, and ii) this leads to unrealistic values $\Delta z_C \approx 6.64e7 x_C$.

Now I'm seeking for help.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.