Light Field Reconstruction Using Shearlet Transform in TensorFlow

Shearlet Transform (ST) is one of the most effective approaches for light field reconstruction from Sparsely-Sampled Light Fields (SSLFs). This demo paper presents a comprehensive implementation of ST for light field reconstruction using one of the most popular machine learning libraries, i.e. Tensor Flow. The flexible architecture of TensorFlow allows for the easy deployment of ST across different platforms (CPUs, GPUs, TPUs) running varying operating systems with high efficiency and accuracy.

[1]  Robert Bregovic,et al.  Light Field Reconstruction Using Shearlet Transform , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Robert Bregovic,et al.  Accelerated Shearlet-Domain Light Field Reconstruction , 2017, IEEE Journal of Selected Topics in Signal Processing.

[3]  Bastian Goldlücke,et al.  A Dataset and Evaluation Methodology for Depth Estimation on 4D Light Fields , 2016, ACCV.