What is calibration target?
Machine vision applications are often expected to deliver high accuracy in alignment or measurement. Some applications in the 3D field or robot guidance expand the 2D sensor image into 3D world or robot coordinates. However lense distortion or measurements in physical units such as millimetre may require some sort of transformation of the 2D image coordinates of a distorted object into a physical unit with high accuracy. Examples of this are size or volume measurements, photogrammetry, 3D reconstruction or robot guidance.
However such an application requirement involves some sort of calibration of the image seen by the camera and the physical world. What we need to do is to calculate a transformation by some sort of geometrical model and this is where a calibration target is being used.
We can use the camera to take a picture of a well known scene which shows a pattern array at locations that are well known with a high degree of accuracy. Based on the known locations of the patterns and the locations of the pattern measured in an acquired image of the scene we can calculate the geometric model of the camera which can be used to transform a pixel position in the image to the e.g. mm position in the scene.
A physical object showing the pattern arrays which are fixed spaces is called a calibration target. This of course has to be produced with a very high accuracy to ensure the generation of a transformation with a certain degree of accuracy.