Computer Science · Computer Vision and Pattern Recognition
Using planar point correspondence to calibrate camera arrays for light field acquisition
Ashley W. Stewart, Donald G. Dansereau
See on the mapAbstract
Light field cameras are an emerging technology with unique post-processing capabilities. For certain light field applications such as synthetic aperture photography, existing calibration procedures are either metric calibrations that are difficult to execute with low-cost hardware, or non-metric procedures that are inflexible to arbitrary sub-camera pose or require cameras with tightly controlled poses. We present a novel and comparatively simple non-metric procedure for light field acquisition that estimates the geometric transforms between camera images with respect to a calibration plane. The procedure is suitable for mobile camera arrays, and flexible to unknown or varied sub-camera pose. It only requires one image from each camera of a calibration pattern positioned to span the camera array's full field of view. We also provide a quantitative measure of calibration quality, and use it to demonstrate the procedure's efficacy with our Raspberry Pi camera array. The results highlight the procedure's robustness to variable camera orientation in contrast to existing state of the art techniques. Finally, we present qualitative results by rendering light fields at varying levels of focus and occlusion, and demonstrate success in capturing and rendering light field video.