摘要 |
The magnification M<SUB>ref </SUB>for a line scan camera can be found by exploiting a difference in the way M<SUB>ref </SUB>affects the notion of 'focus' in the x and y directions. M<SUB>ref </SUB>enters into the calculations for selecting z while focusing in the y direction, but not in x. A thin opaque calibration target is provided at a convenient height in z called the reference plane, and has straight edges aligned parallel to the x and y directions. To find M<SUB>ref </SUB>the line scan camera forms images of the calibration target over a range of trial z values known to include the height of the calibration target. An arbitrary and possibly incorrect trial value M<SUB>i </SUB>of M<SUB>ref </SUB>is then assumed and many slices of the calibration target are calculated. Within these slices the edge parallel to the x direction will be sharply defined for some z<SUB>x</SUB>, while at some other z<SUB>y </SUB>the edge parallel to the y direction will be sharply defined. Make a note of e<SUB>i</SUB>=z<SUB>y</SUB>. Repeat these steps for some number different trial M<SUB>i </SUB>that cover the plausible range of M<SUB>ref</SUB>. Now fit a curve (e.g., a quadratic) to the data set {(e<SUB>i</SUB>, M<SUB>i</SUB>)}, and find the y-intercept (where e=0). The associated value of M is the magnification M<SUB>ref </SUB>in the reference plane containing the calibration target, and we found it without knowing the actual length of any part of the calibration target.
|