Subjective Evaluation of Light Field Image
Compression Methods based on View Synthesis
Abstract
Light field (LF) images provide rich visual information enabling amazing applications, from post-capture image processing to immersive applications. However, this rich information requires significant storage and bandwidth capabilities thus urgently raises the question of their compression. Many studies have investigated the compression of LF images using both spatial and angular redundancies existing in the LF images. Recently, interesting LF compression approaches based on view synthesis technique have been proposed. In these approaches, only sparse samples of LF views are encoded and transmitted, while the other views are synthesized at decoder side. Different techniques have been proposed to synthesize the dropped views. In this paper, we describe subjective quality evaluation of two recent compression methods based on view synthesis and comparing them to two pseudo-video sequence based coding approaches. Results show that view synthesis based approaches provide higher visual quality than the naive LF coding approaches. In addition, the database as well as subjective scores are publicly available to help designing new objective metrics or can be used as a benchmark for future development of LF coding methods.
CONTENT:
1- 192 Pseudo Video Sequences:
- 160 pseudo video sequences (Testing set): 10 selected images × 4 QP × 4 selective compression schemes
- 32 pseudo video sequences (Training set): 2 selected images × 4 QP × 4 selective compression schemes
2- Four Coding Configurations:
- HM: All the 8x8 views are encoded by HM in Random Access.
- JEM: All the 8x8 views are encoded by JEM in Random Access.
- LA: Half of views are encoded by JEM and the other half views are linearly estimated.
- DL16: 16 views are encoded by JEM and the others views are synthesized with the trained CNN.
3- File-Naming For Video Sequences:
- Each video sequence is named by the following rule: [name]_[Config]_[QP].avi
- The range of "QP" is {22, 27, 32, 37}.
- The range of "name" is {Bikes, Fountain_&_Vincent_2, Friends-1, Overexposed-Sky, Rusty-Fence, University, Bee1, Cactus, KidsHouse and Kniphofia}.
4- Database Description
- Microsoft Excel spreadsheet, giving all subjective results is included in the above shared folder in Google Drive.
- Pseudo video sequences of the four coding configurations (used in this paper) are available here for download.
- More Light Field images (28 light fields) and the calibration files are available here for download.
- Thumbnails :
Kniphofia
Kids_House
Jasmine
Toys
Lake_&_foliage
Road_Garden
Cartoons_Fighters
Teeter_Totter
Tree_Trunk
Fountain
Spring_Flowers
Bird
Red_Door
Mixed_Roses
Office
Rose_Alabaster
Bulldozer
Cakes
Waterfall
Knapweed
Public_Seating
Cars
Lake_&_foliage2
Tea_Box
Textured_Window
Garden
Garden_Stairs
Outdoor_Gym
Citation : if you use this database for your research, please cite our paper.
N. Bakir, S. Fezza, W. Hamidouche, K. Samrouth and O. Deforges, "Subjective Evaluation Of Light Field Image Compression Methods Based On View Synthesis", 27th European Signal Processing Conference, Coruña, Spain, Sep 2019.
@INPROCEEDINGS{bakir2019,
author = {N. Bakir, S. Fezza, W. Hamidouche, K. Samrouth, O. Deforges},
title = {Subjective Evaluation Of Light Field Image Compression Methods Based On View Synthesis},
booktitle = {27th European Signal Processing Conference},
month = {Sep},
year = {2019},
}
Contacts
- Nader Bakir, nader.bakir@insa-rennes.fr
- Wassim Hamidouche, wassim.hamidouche@insa-rennes.fr
- Sid Ahmad Fezza, sidahmed.fezza@gmail.com