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 459

FIGURE 9 Correction result using the distortion center estimation method.

In the case of incorrect distortion correction due to the displaced distortion center, the image
was corrected only in either the vertical or horizontal direction as shown in Figure 8(a) and
(b) , respectively. This phenomenon occurred because of the incorrect designation of the dis-
tortion center axis for the distortion correction. To minimize this error, the precise center axis
can be found using the distortion center estimation method by means of 2D patterns such as a
chessboard patern. This resulted in minimizing the distortion correction error that might oc-
cur owing to instrument error while configuring the experiment or camera mounting ( Table
3 ).

Table 3
Result of the Proposed Distortion Estimation Method

Cod x Cod y

First measurement 10 − 11

Figure 10 shows the distortion correction results as image data were received in real time
from the 190° wide-angle camera. As shown in Figure 11(a) , distorted correction can be found
in the distorted monitor image in the horizontal and vertical directions. Figure 11(b) shows
the distorted result image when the original image was corrected while (c) and (d) show the
full unwrapped images of (b). If images collected from a wide-angle lens are corrected in a
distorted manner, they are transformed into images of infinite size so that cropping at an ap-
propriate scale should be used to produce images for monitoring and composition.

FIGURE 10 Result of the distortion center correction using 2D planar patterns.