Lens distortion is a commonly observed aberration in images. It appears as a warping of image data towards the image periphery, resulting in an unfaithful representation of the scene geometry in the image. More precisely, lens distortion is a variation in magnification with distance from the image (distortion) centre. It is one of the five Seidel aberrations. Models for lens distortion are typically radially dependent, although for wide-angle and fisheye lenses alternative lens models are generally used that are a function of the entrance angle of incoming rays.

Example image exhibiting severe barrel distortion, and corrected image with distortion removed.
Mouse over the image to see correction.


Lens geometric distortion can be broadly classified as either barrel or pincushion, depending on how the distortion effects the resulting image. Barrel distortion causes image compression towards the periphery, whilst pincushion distortion results in image expansion towards the periphery. The distortion function of a prime lens is fixed since geometric distortion is primarily a function of lens focal length. For zoom lenses, the extent and type of distortion can vary with focal length, and so a single zoom lens often exhibits pincushion and barrel distortion effects at different points in the zoom range.


Correction Solution

We have developed a distortion calibration technique that requires only simple input images of a structured target, which can be acquired without any specialized equipment or lab conditions. Therefore we can provide the end user with simple fool-proof technology to precisely correct lens distortion in any camera or lens, with no restrictions on developer support. We have also developed a technique to estimate lens distortion for a single image based on a view of a particular target.

These new techniques are enabled by two important advancements. The first is extending the lens distortion model to cover variable lenses, i.e. lens distortion can be removed for any image the camera takes at any zoom and aperture setting. The second innovation relaxes the prerequisite that the calibration target must have a specific number of features, detected and arranged in a predefined manner. Instead a fully automatic feature registration process is applied that allows a user to use any number of features regardless of the position or orientation of the features in the image.


Key Advantages

  • Distortion characteristics are identified in calibration phase
  • Fully automatic process developed to extract the necessary information from images of a structured planar target


Examples of our distortion correction solution applied to images from varies cameras/lenses are shown below.

Olympus point-and-shoot: Original image exhibiting barrel distortion, and corrected image with distortion removed.
Mouse over the image to see correction.

Nikon D200, 55-200mm lens: Original image exhibiting pincushion distortion, and corrected image with distortion removed.
Mouse over the image to see correction.



copyright jaliko     |     Jaliko Ltd., Invent DCU, Glasnevin, Dublin 9, Ireland.     |     Company Reg. No. 492193     |     Site Terms & Conditions