Mosaicking Images Based On Multiply Wide-angle Cameras

Super wide angle images have been widely used in many domains of modern life. They can be obtained by fish-eye lens. However, the image resolution is very poor in the edge of the image and the obvious distortion also exists. The common cameras can not capture the wide angle images due to its small field angle. The image mosaicking has come into being.Prevalent international Image mosaicking Techniques are similar at the present time because the images used to do the mosaicking are captured by common cameras. The core of those methods is searching the same contents of two borders upon images with some arithmetic, and then the relative relationship of the two images is determined. Different arithmetic conduce different image mosaicking effect. Because of the limit of the field angle of view, those mosaicking images have to be gained by scanning. The synchronization of those images cannot be assured. So it will make against image mosaicking.This paper advances an image mosaicking methods based on multi-wide-angle cameras. Several wide-angle cameras are accurately installed on each surface of a cube, then the images can be captured synchronously and the wide field scene can be divided by those cameras. An arithmetic based on the unitary polynomial of image height is proposed in the paper for distortion rectification. Those images with obvious distortion obtained by the wide angle cameras which has a field angle of view of 130 degree will turn to images without distortion. Then a direct mosaicking method based on the relationship between the field angle of view and the image height will be used to do the image mosaicking and the mosaicking range of the images are also confirmed. A math model including the system dimension and the object distance is also proposed in the paper, and will be discussed in several practical conditions. Accordingly, several kinds of mosaicking work are actualized due to different experiment conditions. Aiming at the characteristics of the above mosaicking images, arithmetic of correlation coefficient based on pixel lists which is a fine supplement is also brought forward to accomplish the exactitude mosaicking. Error analysis including the distortion rectification and the system structure is proposed at last and several improvements are also discussed in the paper due to it.