For short focal length and large field of view, fish-eye lens has been widely applied in such fields as virtual reality, video surveillance, intelligent transportation and robot navigation, but it is necessary to correct the severely distorted images taken by camera with fish-eye lens into the perspective projection images that are visually perceived by human. The 3D fish-eye image correction projects the pixel points of fish-eye image into the space and then maps them into the corrected image based on the correction model. Considering the shortcomings of existing algorithms for extracting the effective region, this paper proposes an improved algorithm for such purpose to ensure both accuracy and efficiency, and presents the fish-eye image correction from the approach of 3D space.
展开▼