Determination of the Image Distance in a Plenoptic Camera

Abstract

Two methods for determining the distance between the lens of a plenoptic camera and the focal plane, known as the image distance, are presented and tested. The plenoptic camera is a form of light field imaging, which seeks to record both the spatial and angular information of the light within a volume. The plenoptic camera requires one calibration image to determine the location of the focal points of each microlens within the camera. The focal point method of calibration uses the location of these focal points to geometrically determine the image distance. The magnification method of calibration instead requires an additional calibration image of a ruler and uses the standard, well-studied imaging equations to back out this distance. This thesis seeks to analyze both methods of calibration for accuracy and robustness across a wide range of possible image distances. The magnification method is proven to provide a sound estimation for the image distance for all of the tested cases. Synthetic results show that the focal point method of calibration is mathematically sound, but experimental results show that the real application of the method is limited by several deficiencies, namely errors caused by assuming that a complex lens exhibits the same properties as a thin lens. Results show that the focal point method can provide comparable accuracy to the magnification method by applying a correction equation specific to each lens settings, but overall the magnification is more robust and much simpler for general experiments.