立体图像和平面图像质量评价常用数据库

做图像质量评价总会碰到各种各样的数据库，在此做一个总结，有的是常用的，有的是一些论文中提及的，会持续更新！

立体图像数据库

• LIVE S3D IQA DB Phase I [1] : Twenty reference and 365 distorted S3D image pairs, of which 80 pairs are related to each of JPEG, JPEG2000 (JP2K), additive white Gaussian noise (WN), and Rayleigh fastfading channel distortion (FF), and 45 pairs contain Gaussian blur (BLUR). All distortions are symmetric in nature (i.e., the same degree of distortion in both left and right images). The corresponding disparity map and differential MOS (DMOS) value for each distorted S3D image pair is provided. The DMOS value is in the range [0,80], where lower values represent higher visual quality.

• LIVE S3D IQA DB Phase II [2] : Phase II consists of 8 reference and 360 distorted S3D image pairs with co-registered disparity maps and DMOS values. There are nine levels of each distortion type (JPEG, JP2K, WN, BLUR, and FF), three of which are asymmetrically distorted S3D image pairs (i.e., different degree of distortion in the left and right images).

• IEEE-SA DB [3] : Twenty-six references and 650 distorted S3D image pairs, 130 pairs for each distortion type (JPEG, JP2K, WN, BLUR, and FF). The corresponding disparity maps and differential MOS (DMOS) values for each distorted S3D image pair are provided. Lower DMOS values represent higher visual quality.

• IVC stereoscopic image dataset [6]: This dataset contains 96 stereoscopic images and their associated subjective scores. The resolution of these images is 512*512, which were displayed on a 1280*1024 monitor with an uniform gray around the
image to keep the native resolution of the image. 6 different stereoscopic images are used in this dataset, which is composed
of the 6 reference images (undistorted) and 16 distorted versions of each sources generated from 3 different distortion types (JPEG, JP2K, BLUR) symmetrically to the stereopair images

• NBU 3D IQA Database [7]：NBU-3D 测试库的原始立体图像是由 Mobile3DTV 和 MPEG 提供 , NBU-3D 立体图像测试库包
含 12 对原始立体图像和 312 对失真立体图像 , 包含高斯模糊 (Gaussian blur, Gblur), 白噪声 (whitenoise, WN), JPEG 压缩 , JPEG2000(JP2K) 以及H.264 编码 5 种失真类型 , 左右图像均为失真程度相同的对称失真 , 并给出每组失真立体图像的DMOS 值。

•The MCL_3D image quality database [8]: from the University of Southern California, USA, contains 684 symmetrically distorted 3D images, where each distorted 3D stimuli has been scored by human observers. One third of the database has a resolution of 1024 ×728, and the remaining images have a resolution of 1920 ×1080. Nine image-plus-depth sources were first selected, and a depth-image-based rendering technique was used to render 3D images. Four levels of distortion were applied to either the texture 3D image or depth image prior to 3D image rendering. The distortion types are GBLUR, WN, down sampling blur (SBLUR), JPEG, JP2K, and transmission error (TERROR).

• Waterloo-IVC 3D Image Database Phase I [9]：The Waterloo IVC SIQA Database Phase 1 (WIVC-1) has 6 reference stereopairs and 330 symmetrically or asymmetrically degraded ones. The size of left and right view images is 1390 × 1080. The types of distortions are WN, gaussian blur and JPEG. Each distortion type has 4 distortion levels that ensures a good perceptual separation. Mean opinion score (MOS) values are presented for subjective quality scores. Note that stereopairs in WIVC-1 has larger horizontal disparities than the other databases.

• Waterloo-IVC 3D Image Database and Phase II [10]：The Waterloo IVC SIQA Database Phase 2 (WIVC-2) has 10 reference stereopairs and 460 symmetrically or asymmetrically degraded ones. The size of left and right view images is 1920 × 1080. Note that the resolution of images in WIVC-1 and WIVC-2 is better than the LIVE-1, LIVE-2 and IVC. The types of distortions are the same as them of WIVC-1. MOS values are provided for subjective quality scores.

平面图像数据库

• LIVE [4]: A total of 779 distorted images with five different distortions - JP2k compression (JP2K), JPEG compression (JPEG), White Noise (WN), Gaussian blur (BLUR) and Fast Fading (FF) at 7-8 degradation levels derived from 29 reference images. Each image has a differential Mean Opinion Scores (DMOS), which are in the range [0,100]. Higher DMOS indicates lower quality.

• TID2008[5]: 1700 distorted images with 17 different distortions derived from 25 reference images at 4 degradation levels. In our experiments, we only consider the four common distortions that are shared by the LIVE dataset, i.e. JP2k, JPEG, WN and BLUR. Each image is associated with a Mean Opinion Score (MOS) in the range [0, 9]. Contrary to DMOS, higher MOS indicates higher quality.

Preferences

[1] A. K. Moorthy, C.-C. Su, A. Mittal, and A. C. Bovik, “Subjective evaluation of stereoscopic image quality,” Signal Process., Image Commun., vol. 28, no. 8, pp. 870–883, Dec. 2013.

[2] M.-J. Chen, L. K. Cormak, and A. C. Bovik, “No-reference quality assessment of natural stereopairs,” IEEE Trans. Image Process., vol. 22, no. 9, pp. 3379–3391, Sep. 2013.

[3] (2012). IEEE-SA Stereo Image Database. [Online]. Available: http://grouper.ieee.org/groups/3dhf/

[4] Hamid R Sheikh, Zhou Wang, Lawrence Cormack, and Alan C Bovik, “Live image quality assessment database release 2. online, http://live.ece.utexas.edu/research/quality,” 2005.

[5] Nikolay Ponomarenko, Vladimir Lukin, Alexander Zelensky, Karen Egiazarian, M Carli, and F Battisti, “Tid2008-a database for evaluation of full-reference visual quality assessment metrics,” Advances of Modern Radioelectronics, vol. 10, no. 4, pp. 30–45, 2009.

[6] A. Benoit, P. Le Callet, P. Campisi, R. Cousseau, Quality assessment of stereoscopic images, EURASIP J. Image Video Process. 2008 (2009) 1–13.

[7] F. Shao, W. Lin, S. Gu, G. Jiang, and T. Srikanthan, “Perceptual fullreference quality assessment of stereoscopic images by considering binocular visual characteristics,” IEEE Trans. Image Process., vol. 22, no. 5, pp. 1940–1953, May 2013.

[8] R. Song , H. Ko , C.C.J. Kuo , MCL-3D: a database for stereoscopic image quality assessment using 2D-image-plus-depth source, J. Inf. Sci. Eng. 31 (2015) 1593–1611 .

[9] J. Wang and Z. Wang, Perceptual quality of asymmetrically distorted stereoscopic images: the role of image distortion types, in Proc. Int. Workshop Video Process. Quality Metrics Consum. Electron, Chandler, AZ, USA, Jan. 2014, 1-6.

[10] J. Wang, A. Rehman, K. Zeng, S. Wang, and Z. Wang, Quality prediction of asymmetrically distorted stereoscopic 3D images, IEEE Trans. Image Process., 24, 11, 3400–3414, (2015).