Three-Dimensional Super Resolution Reconstruction by Integral Imaging

Abstract

In this paper, we present a novel three-dimensional (3D) super resolution reconstruction technique by integral imaging to reconstruct a high-resolution 3D image directly from a series of low-resolution perspective projections of a 3D scene. In conventional integral imaging, every depth of the 3D scene is imaged by multiple lenses on the sensor plane with an accurate disparity of multiple perspective viewpoints. However, this disparity is pixilated by the digital structure of the capturing sensor, which leads to a loss of the sub-pixel disparity information of the multiple perspective viewpoints. In this paper, this sub-pixel information is restored by the proposed 3D super resolution method in order to improve the resolution of the reconstructed 3D images. Simulations and experiments are implemented to demonstrate the enhancement of the 3D image quality in the reconstruction.