Mathematical Modeling On Fisheye Images Correction And Its Applications

Fisheye lens has a large view of 180 to 270 degree, for the particularity of its optical properties, which is widely used in different kinds of monitoring field, like bank, transportation, school, etc. It also can be applied to varieties of panorama system, with the superiority of simple fusion, full covering and no blind area. Despite the various of merits, they can not avoid the distortions when taking photos. Fisheye images always have large distortions. To enhance the visual effect of fisheye images, the distorted fisheye images need to be recovered to the undistorted normal visual images. Many correction algorithms have been proposed in recent years. However, all these correction methods can not ignore one point, that is none can avoid all the distortions caused by the fisheye lens. And typical users appear accustomed to and tolerant of these distortions. In addition, millions of circular fisheye images have been shared on the Internet like Flickr. Therefore, instead of correcting the distorted fisheye images, we attempt to find another way to handle such images.In this paper, we mainly study three aspects including the algorithms of fisheye images correction, fisheye images beautification, fisheye images forensic. After investigating the current applications of perspective images, such as image beautification, image forensics and so on, we find such applications can totally be extended to fisheye images. For example,we can also find some guidelines to beautify fisheye images and develop some forensics detecting methods by using the geometric invariant properties of fisheye lens.The main contributions of this paper can be summarized as follows:(1) Analyzing and discussing the imaging principle of fisheye images, elaborating the research background and the significance, summarizing the current research on fisheye images, including correction algorithm and its application.(2) Summarizing the relevant literature on fisheye images correction algorithm, and extending the corresponding correction algorithm.(3) Putting forward the concept of fisheye images beautification, and first identifying the direction and shape as the guidelines of the beautification and facilitating a fisheye image editing tool, extending the methods of editing fisheye images in a new perspective.(4) Using some invariant properties of geometric constraints, according to previous literatures, to detect the tampered objects in fisheye images, we then propose a simple and effective method, based on the vanishing points.