Image quality enhancement by calibration for autostereoscopic display

Abstract

In this dissertation, we address a calibration method for autostereoscopic 3D display. We establish a mathematical model of the autostereoscopic 3D display. We propose three calibration methods using visual pattern analysis, which is based on the physical model. First, we propose a calibration method of global parameters for autostereoscopic 3D display. For highly accuracy, we analysis captured patterns on frequency domain. Second, we present a method for local deformation calibration of the autostereoscopic display. In this case, we employ light field concept to estimate a deformed shape of the device and to render an image which compensate the deformation. Finally, a novel calibration method is presented to calibrate all parameters for eye-tracked autostereoscopic display. To make our method both simultaneous and automated, we design not only an efficient algorithm but also a calibration module. Through the proposed calibration method, the quality of the 3D image is dramatically improved.