A Crucial Approach for Image Rectification Based on a Self-Calibrated Camera

In this article we will present a robust technique of image rectification, the images to be rectified are taken from two different views of an unknown 3D scene with a CCD self-calibrated camera. The image rectification task consists to apply a pair of 2D transformations or homographies to the original images to render the epipolar lines horizontal and parallel to the line formed by the two camera centers. The proposed method of image rectification discussed in this paper is divided in two steps: The first consists to determine an initial pair of rectification homographies and the second one is a robust estimation of rectification homographies by using the initial pair witch the aim is to reduce the distortion of rectified images. The strong point of our proposed method reside in the use only of the matching points between the two original images. The experimental results on the synthetic and real data show the robustness of this approach.