CNN-Based Macropixel-Level Up-Sampling for Plenoptic Image Coding

• Main Authors: Kaiming Zhang, Xiaohuan Liu, Jing Zhang, Jingyi He, Yanan Shi, Zongqian Zhang
• Format: Article
• Language: English
• Published: IEEE 2019-01-01
• Series: IEEE Access
• Subjects: Plenoptic image compression; convolutional neural network (CNN); down-sampling; up-sampling
• Online Access: https://ieeexplore.ieee.org/document/8736278/

Plenoptic imaging has emerged as a representative approach for recording richer visual information from the real world. With the insertion of a microlens array, plenoptic cameras can record both angular and spatial information of a scene on a plenoptic image. However, the large amount of data calls for efficient coding techniques for both transmission and storage. In this paper, we propose a convolutional neural network (CNN)-based macropixel-level up-sampling method for plenoptic image coding. First, a macropixel-based down-sampling method, which performs the down-sampling in the units of macropixels, is developed for reducing the block resolution. Then, an up-sampling CNN is carefully designed to achieve resolution recovery and quality enhancement for down-sampled blocks. The experimental results show that the proposed method achieves considerable bitrate reduction compared with the HEVC/H.265 format SCC extension profile.