RIT will permanently shut down the RIT Wiki (Confluence wiki) on February 2, 2024. Review the Wiki Shut Down Overview for information on alternative tools, exporting content from the wiki, and other FAQs.
http://graphics.stanford.edu/~wilburn/wilburn_thesis.pdf
High-Speed Videography Using a Dense Camera Array.pdf
-Bennett Wilburn, et al.
Wilburn, et al., discuss their implementation of synthetic aperture imaging, as well as geometric calibration, in the dissertation and the paper. The implementation included the idea of "High-X" imaging, where "X" refers to imaging system characteristics such as dynamic range, resolution, signal-to-noise ratio, etc. There is not much information regarding the alignment of the cameras. The only mention is that the camera array does not rely heavily on precise alignment, and that the cameras are made flexible so they can be arranged for any application. A contact with the author could possibly aid in determining their compensation for alignment errors.
Geometrical Compensation for Multi-view Video in a Multiple Camera Array.pdf
An Efficient Rectification Algorithm for Multiview Images in a Parallel Camera Array.pdf
-Kang, et al.
The authors (same for both papers) discuss the geometric errors in camera arrays. Since cameras are aligned with human hands, there are going to be alignment issues in the horizontal and vertical directions, as well as rotational alignment errors. With error in alignment comes irregularity in vertical and horizontal positions, and some images where rotation is off a certain degree. This can be corrected in post-processing. A common optical center is found between the cameras by finding midpoints between adjacent cameras, connecting those midpoints, and doing the same for the new midpoints of the connector lines until there is one line left which is used as the baseline for the optical center. Then, the optical axis, focal length, and principal point are found for the system by averaging each of these, respectively, for each camera. The compensated camera parameters are used to calculate a compensating transformation transformation, which is then applied to the images, and they are now aligned properly. There is a lack of discussion as to how the process is carried out, but a contact with the author would certainly be useful.
http://graphics.stanford.edu/~vaibhav/pubs/thesis.pdf
Using Plane + Parallax for Calibrating Dense Camera Arrays.pdf
-Vaibhav Vaish, et al.