Background

The conventional rate-distortion image compression framework has been developed for more than 30 years. Although the peak-signal-to-noise-ratio (PSNR) index has been widely used to measure the quality of compressed image/video, it does not take the Human Visual System (HVS) characteristics into consideration. Several recently proposed IQA/VQA algorithms, such as SSIM, FSIM, EVQA, were built on top of limited quality datasets, the distorted clips have suprathreshold distortion with both the reference and each other. To address this problem, it is essential to explore new perceptual quality indices. In this project, we adopt a statistical approach to study the perceptual quality indices and propose a new methodology to characterize the human visual experience on coded clips based on the notion of just-noticeable difference (JND). We built a H.264/AVC encoded video dataset with a wide variety of video contents and invited a large number of subjects to participate in the subjective test.

Dataset Description

The MCL-JCV dataset consists of 24 source videos with resolution 1920×1080 and 51 H.264/AVC encoded clips for each source sequence. Single-pass constant QP encoding (CQP) was used with the Quantization Parameter (QP) ranging from 1 to 51. More than 120 volunteers participated in the subjective test. Each set of sequences was evaluated by around 50 subjects in a controlled environment. For more detailed information, We refer to [1]. A GMM [2] approach is used to derive the average viewer experience in form of the Stair Quality Function (SQF). The application of the JND notion to compressed image/video quality assessment was first discussed in [3]. All JND data (24 sequences x 3 JND x (~50) samples) were provided as well as all reference and degraded clips. Please note that only 24/30 sequences were released in this dataset due to Intellectual Property (IP) issue.

Download

Download link (FTP): DATASET LINK

Note that the above link was embedded with credentials to download. In case it prompts to login, use mcl_jcv for both the account and password. Let me know if you still encounter difficulties about downloading.
Contact Info: Haiqiang Wang, Email: haiqianw AT usc.edu

If you use this dataset in your work, please cite related papers:

Detailed description of the MCL-JCV
[1] H. Wang et al., “MCL-JCV: A JND-based H.264/AVC video quality assessment dataset,” 2016 IEEE International Conference on Image Processing (ICIP), Phoenix, AZ, USA, 2016, pp. 1509-1513. PDF

JND data processing
[2] Sudeng Hu, Haiqiang Wang and C.-C. Jay Kuo, “A GMM-based stair quality model for human perceived JPEG images,” IEEE International Conference on Acoustic, Speech and Signal Processing, Shanghai, China, March 20-25, 2016. PDF

Overview on JND-based quality measure of coded image/video
[3] Joe Yuchieh Lin, Lina Jin, Sudeng Hu, Ioannis Katsavounidis, Anne Aaron and C.-C. Jay Kuo. “Experimental Design and Analysis of JND Test on Coded Image/Video.” SPIE Optical Engineering+ Applications. International Society for Optics and Photonics, 2015 . PDF

Exemplar Frames from Source Sequences

Copyright Notice

Copyright (c) 2016, University of Southern California

Permission is hereby granted, free of charge, to any person obtaining a copy of the database and associated documentation files (the “MCL-JCV DATASET”), to deal in the database without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, and/or sell copies of the MCL-JCV DATASET, and to permit persons to whom the dataset is furnished to do so, provided that the above copyright notice(s) and this paragraph and the following two paragraphs appear in all copies of the MCL-JCV DATABASE and in supporting documentation.

IN NO EVENT SHALL THE UNIVERSITY OF SOUTHERN CALIFORNIA BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF THE MCL-JCV DATABASE, EVEN IF THE UNIVERSITY OF SOUTHERN CALIFORNIA HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

THE UNIVERSITY OF SOUTHERN CALIFORNIA SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT. THE MCL-JCV DATABASE PROVIDED HEREUNDER IS ON AN “AS IS” BASIS, AND THE UNIVERSITY OF SOUTHERN CALIFORNIA HAS NO OBLIGATIONS TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.