Research On Distortion Correction Algorithm For Fisheye Image

The fisheye lens is a camera which has super wide-angle. The image captured by the fisheye lens is called the fisheye image. The images has more angle and information than the general images. Because the bigger angle the fisheye lens, A set of common lens could be replaced by a fisheye lens, thus simplifying the process of image acquisition, saving the hardware resources.Because of the characteristics above, the fisheye lens has a broad market prospect in video conference, security surveillance, intelligent transportation and so on. The distortion of the images captured by the fisheye lens makes us difficult to understand the image information and affects the result of image recognition, so that the utility of fisheye lens is reduced, therefore, The images captured by the fisheye lens must be corrected to let us realize the image information except the field of photographic art.This paper analyzes the basic principle and characteristics of all kinds of fisheye image correction model firstly, then compared the characteristics of various models. This paper presents a improved distortion correction algorithm which is based on the checkerboard template, and optimized the key parameters of the correction model. The main work is as follows:(l)The paper analyzed and summarized some literatures related to the fisheye lens, elaborated the background and actuality of the research, discussed the relevant theoretical knowledge of image correction, such as,all kinds of coordinates, the pinhole image, perspective projection, all kinds of distortion and image preprocessing. And shows the fisheye image model based on the imaging characteristics.(2)The paper studied the correction methods based on the grid template, concentric circle template and equivalent spherical model is in detail. Both of the grid template and the concentric circle template are using the coordinate mapping of the feature points to achieve correction of the image distortion. The equivalent spherical model is assumed to be a distorted image is a surface, then the surface is expanded into a plane, and then calculates the plane image, so as to complete the correction of the distortion of the image.(3)The paper proposed a improved fisheye image correction algorithm based on the template method. A Taylor polynomials is used to approximate the general model of the fisheye lens firstly, after the process of extracted the coordinates of the distortion image and the coordinates of the ideal image.Acounted into the rationality and accuracy of the operation, this paper lets the distance from the center of a image to a corner as the real distance, and the distance from the center of the checkerboard to the corner as the dream distance. And then use the least squares method to solve the distortion coefficient. Then, it used the bilinear interpolation to get the corrected fisheye image.(4)Although the feature extraction has been accurate to sub-pixel level, but taking into account the template, and feature point extraction and calculation accuracy has some deviation inevitable, the paper did some operations such as four polynomial fitting, inverse function and multiple fitting calculation. And finally the paper realized the optimization of distortion parameters.The image which used the optimized parameters achieved satisfactory results.