Research Of The Fisheye Lens Distortion Correction Method Based On Multiple Images

With application in the computer vision, fisheye camera with field of view 180°attracts more and more attention in recent years. However, they need be corrected because fisheye images often have severe distortions. This paper studies the correction methods of the severely distorted fisheye images.Main contents are as follow:1. This paper presents a fisheye lens distortion correction method from multiple fisheye images. A measurement matrix is first constructed from part feature points, which are less distorted than other feature points, in the center areas of multiple fisheye images, and estimate the projective matrix of multiple fisheye images by using the factorization method. Then, suppose skew of fisheye camera is zero, constraints on the intrinsic parameters of camera are derived from the projective matrix and the initial value of camera intrinsic parameters is computed. Last, the distorted parameter of fisheye camera is refined on minimizing the objective function, which is obtained from the property that a conic in the fisheye image corresponding the same space line is projected a great circle onto the unit sphere. Experimental results with simulated and real image show that the correction results appear satisfactory using this method.2. A fisheye camera distortion correction method based on the bilinear function model is presented in this paper. Fisheye image points is lifted from a 2D point to a four dimensional quadratic space at first. And the bilinear function model can be built by using a linear combination of the distortion parameters, these parameters are computed by using the property that 3D lines in the space are projected to curves in the fisheye image under the bilinear function model. Thus, fisheye image is corrected from these model parameters. Experimental results with synthetic data and real fisheye images show this correction method is feasible and effective.