Employing a Fish-Eye for Scene Tunnel Scanning

Abstract

This work employs a fish-eye to scan cityscapes along a street and register scenes in a compact scene tunnel image. A fish-eye has complete field of view along a route. We mount the fish-eye camera on a vehicle and estimate its pose initially with respect to the vehicle by referring to 3D structure lines of such as roads and buildings on a street. Sampling curves are then allocated in the image frame for dynamic scanning route scenes as the vehicle moves forward. The accurate alignment of the curves ensures less distortion of shapes in the scene tunnel image. We also analyze the scanned results and evaluate alignments of the sampling curves to improve the scanning. The resulting scene tunnel is a continuous archive of the entire route in a city, which can be used for environment visualization and assessment, Internet based virtual navigation, city information indexing, etc.

Equidistant (fθ) fish-eye perspective with application in distortion centre estimation

“…Zheng and Li [19] describe a method by which the roll, tilt and yaw of the camera are determined by extracting the vanishing points of parallel lines in a fish-eye camera. However, they do not determine any of the intrinsic parameters of the camera, and fail to fully describe the perspective of the fish-eye camera.…”

Scanning Depth of Route Panorama Based on Stationary Blur

“…Vertical lines on building rims in the image converge to a vanishing point. We estimate the vanishing point in the image plane Zheng and Li 2006) and the slit is located to pass through it for a vertical PoS. From the vanishing point, we further calculate the camera tilt and then the projection of horizon, denoted by h, in the image.…”

Digesting omni-video along routes for navigation

“…An omni-video can be taken when the camera moves along a certain path with several types of omni-cameras such as fisheye camera [14,16], ladybug camera with spherical view [4] and mirror reflection camera with a cutoff field of view at top [2,15]. Usually a fisheye lens only needs one camera.…”