We propose a method for the accurate recovery of the parameters of a PIV camera using a single image of a plane target as well as the camera focal length and the sensor horizontal and vertical pixel pitches. After the target marks have been located, an optimal projective transform is searched for by a least square method between the target plane coordinates and the image plane coordinates. Using only the camera focal length and sensor pixel pitches as additional information, a complete pinhole camera model is recovered via a canonical decomposition of the linear part of the projective transform and an iterative search for the location where the optical axis intersects the image plane.
The method has been evaluated with real images: for 25-mm focal length cameras located at 377 mm of the target origin and with an angle of 12 degrees with it, the locations of the optical center were found with a 2 mm accuracy and the mark location reconstruction RMS error was below 0.5 pixels.
The method has also been evaluated with the synthetic images available from the Visualization Society of Japan and it was observed that there was no significant decrease in the quality of the flow field evaluation when the recovered camera model was substituted to the exact one even though the optical pathway includes an air to water refraction index change that theoretically makes the pinhole model inappropriate.
The main limitation of this method is that it is based on a pinhole camera model which is not appropriate if the camera has significant non linear optical distortions and if there are changes of refraction index along the optical pathway between the sensor plane and the target plane. But, even in these case, if the non-linearities are not too strong, the method may be useful.