Image Processing Reference
In-Depth Information
FIGURE 1 Distortion correction algorithm of 2D planar pattern.
The center of distortion was found using the estimated distortion coefficient and distorted
curve component. Based on the center of distortion ( C x , C y ), a lookup table (LUT) is produced,
which represents the relationship between distorted location and 2D planar location using the
FOV model. Distortion can be corrected by applying the LUT produced offline to real images.
3.2 FOV Distortion Model
The FOV model [ 3 , 4 ] calculates a location value in the image over the plane through the co-
ordinate in which distortion occurs when radial distortion is detected once a video image is
acquired by a fish-eye lens. Figure 2 shows the FOV model.
FIGURE 2 FOV model.
The distance from the center of a sphere to one point of the planar image is r u , whereas r d is
the distance to the distorted location projected onto the sphere plane. Since the FOV model is
based on the optical model, it is derived by trigonometric functions with regard to the angle
ω . Once a point (A) is moved to a point (B) due to radial distortion projected onto the plane of
the sphere and acquired by a fish-eye lens, it is projected onto the image sensor, thereby pro-
ducing radial distortion. The FOV model can compute r d and r u using the distortion functions
and their inverse functions.
Search WWH ::

Custom Search