View Synthesis Method with Partitioned Image for Angular Super-Resolution

Abstract

Recently, with the development of deep learning technology, various view synthesis techniques based on deep learning have been proposed. However, as network performance improves and becomes more complex, the memory shortages occur when processing high-resolution images. In this paper, we propose view synthesis method with partitioned image for angular super-resolution. Our method synthesizes high resolution images using limited memory. The proposed algorithm deals with efficient image compositing methods according to various image partition methods, and the experimental results show which method has better performance.