Research On Fisheye Image Matching And Correction Algorithm

Fish-eye lens has been widely used in different fields because of its short focal length and large field of view(about 180 to 270 degrees),such as military,monitoring,meteorological science,aerospace,medical,ball screen projection and panoramic simulation,etc.However,the fisheye image captured by fisheye lens has serious geometric distortion,which greatly affects the practical value of fisheye image.The main contents of this paper are as follows: firstly,a new matching algorithm is proposed,which combines the scale invariant feature transformation with the feature description of the local intensity order model,and carries out the experiment on the uncorrected fish eye image registration.The experiment shows that compared with the original SIFT algorithm,this algorithm not only completes the task of matching points,but also obtains more logarithm of matching points,Moreover,the matching time is shortened and the calculation efficiency is improved.Secondly,a scan line approximation algorithm is proposed to determine the effective region of fish eye image,which solves the problem that the classical scan line approximation algorithm is prone to bias when there are a large number of black pixels on the boundary of the effective region of fish eye image.Finally,an algorithm is proposed to correct the fish eye image both horizontally and vertically.The experimental results show that the horizontal correction effect of the new algorithm is similar to that of the fish eye image based on longitude correction,and the vertical correction effect is obviously better than that of the fish eye image based on longitude correction,and the correction speed is also improved,And the image definition after correction is much better than that of fish eye image algorithm based on longitude correction.