Re-calibration of camera space manipulation techniques accounting for fisheye lens radial distortion

摘要

We explore the possibility to use Linear Camera Space manipulation (LCSM) algorithm [1] to map a large workspace using cameras equipped with fisheye lenses. We tested how the distortion of the image caused by the lens influences the precision of the LCSM algorithm when used to estimate the positions of visual cues in 3-D space. We used two Hero3 GoPro? (H3GP) cameras as a good compromise between acquisition rate and number of pixel in the image. We used a previously published correction algorithm to eliminate the fisheye distortion [2] and compare it with standard calibration. The model estimate the radial distortion with a bi-quadratic equation which coefficients K and K₂To estimate such coefficients a non-linear calibration called Stereo Calibration Method was also used [2]. Stereo calibration uses a pair of images, also known as a stereo pair, to estimate the relative depth of points in the scene. These estimates are represented in a stereo disparity map, which is constructed by matching corresponding points within the stereo pair. These coordinates are used to calculate the distortion factors of each camera lens.