Precision distortion correction technique based on FOV model for wide-angle cameras in automotive sector - Emerging Trends in Image Processing, Computer Vision, and Pattern Recognition - page 458

2D plane is due to a mismatch of the center point with regard to the optical axis occurred dur-
ing the camera manufacturing process in the case of cameras with a number of layered lenses.
Therefore, while applying the FOV model, which does not estimate the center of distortion
separately, there is a problem of deterioration of distortion correction accuracy as the error of
the distortion center of the lens and the center point of the image becomes larger.
FIGURE 7 Result of the distortion center error ( X axis: − 30 pixels, Y axis: + 30 pixels).
The experiment to determine the error of the measurement angle using the Zhang algorithm
[ 11 ] , which is regarded as a representative method of distortion correction, produced four
measurement results as shown in Table 2 . In addition, it was verified that deflection of the lens
center was discovered in the X direction, to the right of the Y axis, and in upper portion of the
image.
Table 2
Estimation Result of Distortion Center Using the Zhang Algorithm
Cod x Cod y
First measurement 12.1 − 1.13
Second measurement 12.03 − 1.01
Third measurement 6.96 − 1.87
Fourth measurement 10.43 − 0.78
Accordingly, this chapter solved the problems that straight lines were expressed as curves
due to no estimation of the distortion center in the FOV model and deflected representation
by means of precise correction of the distortion center. Figures 8 and 9 show the correction
result after the distortion center in the FOV model was estimated in the horizontal and vertical
directions.
FIGURE 8 Estimation of the distortion center in the (a) vertical and (b) horizontal directions
using the FOV model.