Macro Focus
Focal Length: fc = [ 1178.66797 1181.15151 ] ± [ 1.86813 1.79633 ]
Principal point: cc = [ 618.46199 459.29765 ] ± [ 1.76481 1.98827 ]
Skew: alpha_c = [ 0.00000 ] ± [ 0.00000 ] => angle of pixel axes = 90.00000 ± 0.00000 degrees
Distortion: kc = [ -0.02074 0.15073 -0.00083 -0.00249 -0.00000 ] ± [ 0.00569 0.02501 0.00055 0.00057 0.00000 ]
Pixel error: err = [ 0.29254 0.27918 ]
Telephoto Focus
Focal Length: fc = [ 1042.37150 1057.40693 ] ± [ 6.24922 5.87439 ]
Principal point: cc = [ 606.96519 414.09653 ] ± [ 1.87835 3.74890 ]
Skew: alpha_c = [ 0.00000 ] ± [ 0.00000 ] => angle of pixel axes = 90.00000 ± 0.00000 degrees
Distortion: kc = [ 0.02206 -0.06737 -0.00293 0.00225 0.00000 ] ± [ 0.00456 0.01213 0.00072 0.00072 0.00000 ]
Pixel error: err = [ 0.19551 0.20738 ]
For computer vision tasks it is often important to have an accurate camera model to describe the focal length, camera distortion and other parameters. The camera calibration models shown above were created for a Quickcam 9000 Pro using the MATLAB Camera Calibration Toolbox. These camera models are fairly rough estimates based on 16 images each. The focal length is in pixels for a 1280x1024 image.
Looking at these rough calibration models for each end of the range of focus, it is clear that the auto-focus effects the camera model significantly. Unfortunately based on these results it looks like a look up table will need to be used to map between the current state of the camera's auto-focus and the appropriate camera calibration, or you can just turn off auto-focus.
No comments:
Post a Comment