Expanding a fish-eye panoramic image through perspective transformation
Ling Zhu, Yu Liu, Shiming Lai, Jing Li · 2016
Although image mosaic and panorama photography are developing rapidly these years, obtaining a panorama is still expensive. To address this issue, this paper presents a method to extract information from a top-view panoramic fish-eye image and expand it into a visually acceptable panorama. The fish-eye image is deformed to obtain additional environmental information. Obtaining panoramas from fish-eye images is a challenge. What we need to obtain a panoramic picture of is a top ceiling fish-eye image shot with a fish-eye camera. First, we obtain a partial perspective picture of fish-eye images to eliminate local deformation. Second, we select a series of partial perspective picture sequences through field angle covered relations. Lastly, we synthesize these images into the panorama we need to establish correspondence between the fish-eye image and the panoramic image. In this study, we can obtain required panoramas only through a fish-eye camera. Compared with traditional algorithms, the current method is more effective in solving the radial distortion problem in expanded images.