Fisheye lens calibration using shifted lattice pattern and correction of barrel distortion using spatially adaptive interpolation

In this paper, we present a novel fisheye lens calibration method and a spatially adaptive interpolation algorithm for correcting the radial distortion of the lens. The proposed calibration method uses a shifted lattice pattern (SLP) image that contains sufficiently many feature points with different distances from the image center. The calibration is performed by matching all the points in the simulated image and the corresponding points in the input original image. After calibration, we correct the image using the proposed directionally adaptive interpolation to remove the peripheral degradation caused by the radial distortion. The proposed method can accurately calibrate the fisheye lens and restore the radial distortion without using complicated three-dimensional modeling. As a result, it can be applied to the various imaging systems using fisheye lens such as digital zooming, high resolution video surveillance, and vehicle rear view cameras.