Multiscale structural similarity for image quality assessment
The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index; none of the 14 sources its claims cite is checked for support yet.
Abstract
The structural similarity image quality paradigm is based on the assumption that the human visual system is highly adapted for extracting structural information from the scene, and therefore a measure of structural similarity can provide a good approximation to perceived image quality. This paper proposes a multiscale structural similarity method, which supplies more flexibility than previous single-scale methods in incorporating the variations of viewing conditions. We develop an image synthesis method to calibrate the parameters that define the relative importance of different scales. Experimental comparisons demonstrate the effectiveness of the proposed method.
Overview
CoChat Summary
TL;DR
This paper proposes a multi-scale structural similarity (MS-SSIM) image quality method that combines contrast and structure comparisons across scales and luminance comparison at the coarsest scale. On the LIVE JPEG/JPEG2000 database, MS-SSIM outperformed PSNR, the Sarnoff model, and single-scale SSIM variants on all reported evaluation criteria.
Key findings
- MS-SSIM achieved the best reported performance: CC 0.969, ROCC 0.966, MAE 3.86, RMS 4.91, and OR 1.16%, compared with the best single-scale SSIM at M=2 (CC 0.963, ROCC 0.959, MAE 4.21, RMS 5.38, OR 2.62%) and PSNR (CC 0.905, ROCC 0.901, MAE 6.53, RMS 8.45, OR 15.7%).
- Single-scale SSIM performance varied with scale; the best single-scale result was at M=2, and single-scale SSIM tended to give higher scores as scale increased.
- The calibrated cross-scale exponents were β1=γ1=0.0448, β2=γ2=0.2856, β3=γ3=0.3001, β4=γ4=0.2363, and α5=β5=γ5=0.1333.
- The authors conclude that multi-scale SSIM provides a meaningful balance between scales and outperforms the best single-scale SSIM and state-of-the-art image quality metrics.
Methods
- The MS-SSIM system iteratively applies a low-pass filter and downsamples by 2; at each scale it computes contrast and structure comparisons, computes luminance comparison only at Scale M, and combines them using exponents normalized so that ∑γj=1.
- Cross-scale parameters were calibrated by image synthesis: 5 scales and 12 distortion levels (range 2^3 to 2^14) were used with 10 original 64×64 images, producing 600 distorted images; 8 subjects selected images of equal perceived quality across scales at a viewing distance of 32 pixels per degree of visual angle.
- Evaluation used the LIVE database with 344 JPEG and JPEG2000 compressed images (bit rate 0.028 to 3.150 bits/pixel), MOS from 13–25 subjective scores, and compared PSNR, Sarnoff JNDmetrix 8.0, single-scale SSIM for M=1–5, and MS-SSIM using CC, MAE, RMS, OR, and ROCC after logistic mapping.
Limitations
- Stated: the image-synthesis calibration approach is “still rather crude,” and the authors are working on a more systematic approach.
- Stated: CSF cannot be directly used to derive the parameters because it is typically measured at visibility threshold with simplified stimuli, whereas the goal is comparing complex structured images at visible distortion levels.
- Evident from the text: calibration used only 10 original 64×64 grayscale images and 8 subjects, including one author, and evaluation was limited to the LIVE JPEG/JPEG2000 database.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 14 Not checked · 18 Own results and interpretations
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It can also be observed that the single-scale model tends to supply higher scores with the increase of scales.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Finally, for every one of the objective evaluation criteria, multi-scale SSIM model outperforms all the other models, including the best single-scale SSIM model, suggesting a meaningful balance between scales.”
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
“Experiments show that with an appropriate parameter settings, the multi-scale method outperforms the best single-scale SSIM model as well as state-of-the-art image quality metrics.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In general, the subjects agreed with each other on each image more than they agreed with themselves across different images.”
plainstates an associationn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The resulting parameters we obtained are β 1 = γ 1 = 0.0448, β 2 = γ 2 = 0.2856, β 3 = γ 3 = 0.3001, β 4 = γ 4 = 0.2363, and α 5 = β 5 = γ 5 = 0.1333, respectively.”
plainn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“However, the SSIM index algorithm introduced in [5] is a single-scale approach.”
What the sources say
- Not matchedcites [5] [5] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, 'Image quality assessment: From error measurement to structural similarity,' IEEE Trans. Image Processing , vol. 13, Jan. 2004. -
More than one paper fits this citation.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
10 more claims
Create a free account to see all 33 claims and every source verdict.
- It may state two separate claims as one. 8
- Probably not a substantial claim. 13
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Further, as research in quality assessment has demonstrated, incorporating multiscale information when assessing quality produces QA algorithms that perform better in terms of correlation with human perception [13], [29].”
From No-Reference Image Quality Assessment in the Spatial Domain (2012) · cites this paper as [13]
Citations
Cites
Image quality assessment: from error visibility to structural similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · IEEE Transactions on Image Processing · doi:10.1109/tip.2003.819861 · Cited by 57,892
Self-citationA universal image quality index
Wang, Z., & Bovik, A. C. (2002) · IEEE Signal Processing Letters · doi:10.1109/97.995823 · Cited by 5,757
Self-citationImage quality measures and their performance
Eskicioglu, A. M., & Fisher, P. S. (1995) · IEEE Transactions on Communications · doi:10.1109/26.477498 · Cited by 2,117 · cited 1×
“Traditional perceptual image quality assessment methods are based on a bottom-up approach which attempts to simulate the functionality of the relevant early human visual system (HVS) components.”
BackgroundA Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients
Portilla, J., & Simoncelli, E. P. (2000) · International Journal of Computer Vision · doi:10.1023/a:1026553619983 · Cited by 1,806 · cited 2×
“These synthesis methods provide a powerful and efficient means of testing a model, and have the added benefit that the resulting images suggest improvements that might be made to the model [11].”
MethodSelf-citation(1995) · Medical Entomology and Zoology · Cited by 1,212
Pyramid-based texture analysis/synthesis
Heeger, D. J., & Bergen, J. R. (1995) · Proceedings of the 22nd annual conference on Computer graphics and interactive techniques - SIGGRAPH '95 · doi:10.1145/218380.218446 · Cited by 1,095
Video quality assessment based on structural distortion measurement
Wang, Z., Lu, L., & Bovik, A. C. (2004) · Signal Processing: Image Communication · doi:10.1016/s0923-5965(03)00076-6 · Cited by 1,024
Self-citationImage Quality Assessment: From Error Measurement to Structural Similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · Cited by 919
Self-citationTeo, P. C., & Heeger, D. J. (2002) · Proceedings of 1st International Conference on Image Processing · doi:10.1109/icip.1994.413502 · Cited by 471
Objective video quality assessment
(2003) · Handbook of Video Databases · doi:10.1201/9780203489864-45 · Cited by 359
Research works printed but not linked: 2
- 0, JNDmetrix Technology, Sarnoff Corporation
- 2000, Final report from the video quality experts group on the validation of objective models of video quality assessment, VQEG
Cited by
Items 3,901–4,000 of 6,047
Zero shot prediction of video quality using intrinsic video statistics
Mittal, A., Saad, M. A., & Bovik, A. C. (2014) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2036162 · Cited by 3
Self-citationChen, S., & Muntean, G. (2013) · IEEE Vehicular Technology Conference · doi:10.1109/vtcfall.2013.6692109 · Cited by 3
Objective video quality metrics for HDTV services: A survey
Farrugia, R. A., Galea, C., Zammit, S., & Muscat, A. (2013) · IEEE EUROCON - International Conference on Computer as a Tool (EUROCON) · doi:10.1109/eurocon.2013.6624982 · Cited by 3
“The Multi-Scale SSIM (MS-SSIM) [7] extends the capabilities”
MethodDelivering audiovisual content with MPEG-21-enabled cross-layer QoS adaptation
Ahmed, T., & Djama, I. (2006) · Journal of Zhejiang University. Science A · doi:10.1631/jzus.2006.a0784 · Cited by 3
An Attempt to Combine Structure and Color for Assessing Image Quality
Shi, Y., Ding, Y., Zhang, R., & Li, J. (2009) · International Conference on Digital Image Processing · doi:10.1109/icdip.2009.56 · Cited by 4
“[2] evaluates the performance of several prominent full reference image quality assessment algorithms, among which Multi-scale Structure Similarity (SSIM-MS) method[3] and Visual Information Fidelity (VIF)[4] have much better performance by being tested on the subjective database[5], which includes 5 types of distorti…”
MethodTest pattern based evaluation of ringing and blur in JPEG and JPEG2000 compressed images
Punchihewa, A., & Keerl, A. (2010) · International Conference on Signal Processing and Communication Systems · doi:10.1109/icspcs.2010.5709771 · Cited by 3
Multi-scale information fidelity approach for full reference image quality assessment
Feng, M., Zhao, S., & Cui, J. (2015) · International Conference on Wireless Communications and Signal Processing (WCSP) · doi:10.1109/wcsp.2015.7340986 · Cited by 3
“SSIM has been successfully improved based on both feature extraction of image and the theory of HVS for decades [3] [4] [5] [6].”
MethodBlind Image Quality Assessment Through Wakeby Statistics Model
Jenadeleh, M., & Moghaddam, M. E. (2015) · Lecture notes in computer science · doi:10.1007/978-3-319-20801-5_2 · Cited by 3
“We conducted performance comparisons between the proposed method and six other state of the art image quality assessment methods: Three full reference algorithms including (PSNR [19], SSIM [3], MS-SSIM [14]), and other three no-reference (blind) algorithms including (SSEQ [11], BRISQUE [5], BLIINDS-II [10]).”
BackgroundMethodA pretreatment to improve the quality metrics performance for encoding H264/AVC
Amor, M. B., Kammoun, F., & Masmoudi, N. (2016) · Journal of the Society for Information Display · doi:10.1002/jsid.428 · Cited by 3
A new color transfer quality measure
Bao, L., Panetta, K. A., & Agaian, S. С. (2016) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2224210 · Cited by 3
A study on the effects of RGB-D database scale and quality on depth analogy performance
Kim, S., Kim, Y., & Sohn, K. (2016) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2229600 · Cited by 3
Video quality assessment based on correlation between spatiotemporal motion energies
Yan, P., & Mou, X. (2016) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2236454 · Cited by 4
“Successful FR-IQA algorithms include VSI[1], GMSD[2], SSIM[3], MS-SSIM[4], FSIM[5], VSNR[6], rNSE[7], MAD[8], and VIF[9], and some of them have been used in VQA model design.”
MethodAdaptive Video Transition Detection Based on Multiscale Structural Dissimilarity
Santos, A. C. S. E., & Pedrini, H. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-50832-0_18 · Cited by 3
Optimal Image Quality Assessment based on Distortion Classification and Color Perception
Lee, J.-Y., & Kim, Y. (2016) · KSII Transactions on Internet and Information Systems · doi:10.3837/tiis.2016.01.015 · Cited by 4
An inverse tone mapping operator based on Reinhard's global operator
Kinoshita, Y., Sayaka, S., & Kiya, H. (2016) · International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS) · doi:10.1109/ispacs.2016.7824712 · Cited by 3
A 3D subjective quality prediction model based on depth distortion
Zhang, Y., Jin, X., & Dai, Q. (2016) · Visual Communications and Image Processing · doi:10.1109/vcip.2016.7805546 · Cited by 3
Wang, Z., Baroud, Y., Najmabadi, S. M., et al. (2016) · Picture Coding Symposium · doi:10.1109/pcs.2016.7906359 · Cited by 3
“It can be seen from the MS-SSIM and VDP curves that the proposed codec exhibits a clear gain in perceptual quality over JPEG-LS in a rate range between 1 and 3.5 bits-per-pixel (bpp).”
MethodInfluentialSelf-citationOptimal background modeling for cluttered scenes
Shahbaz, A., & Jo, K.-H. (2017) · Proceedings of the Annual Conference of the IEEE Industrial Electronics Society · doi:10.1109/iecon.2017.8216907 · Cited by 3
Visual Quality Assessment for Projected Content
Le, H. M., Marshall, C., Doan, T., Mai, L., & Liu, F. (2017) · Canadian Conference on Computer and Robot Vision · doi:10.1109/crv.2017.47 · Cited by 3
“These methods include SSIM [8], GSM [29], FSIM [30], VSI [10], MS-SSIM [9], IW-SSIM [31], VIF [11], and SRSIM [32].”
BackgroundMethodInfluentialHardware-friendly inter prediction techniques for AV1 video coding
Deng, Z., & Moccagatta, I. (2017) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2017.8296421 · Cited by 3
A Review Paper on Image Quality Assessment Metrics
Raijada, M. K., Patel, D., & Prajapati, P. (2015) · Journal of Emerging Technologies and Innovative Research · Cited by 3
Medical Image Lossy Compression With LSTM Networks
Nithin Prabhu, G., Nagavi, T. C., & Mahesha, P. (2018) · Advances in medical technologies and clinical practice book series · doi:10.4018/978-1-5225-6316-7.ch003 · Cited by 3
Comments on Objective Quality Assessment of MPEG-2, H.264 and H.265 Videos
Pavlović, B., Bondžulić, B. P., Stojanović, N., Novčić, M., & Bujaković, D. M. (2018) · New Trends in Signal Processing (NTSP) · doi:10.23919/ntsp.2018.8524085 · Cited by 3
Neural Network-based Prediction of Visual Quality for Noisy Images
Rubel, A., Rubel, O. S., & Лукин, В. В. (2019) · International Conference on Experience of Designing and Applications of CAD Systems in Microelectronics · doi:10.1109/cadsm.2019.8779325 · Cited by 3
“There are more than half of the considered FR IQA metrics possessing prediction accuracy in terms of adjusted R2 larger than 0.9 (SSIM, MS-SSIM, FSIM, HaarPSI, MAD, GSM, IW-SSIM, ADD-GSIM, DSS and SSIM4).”
BackgroundInfluentialJust-noticeable difference binary pattern for reduced reference image quality assessment
Miao, X., & Lee, D.-J. (2019) · Optical Engineering · doi:10.1117/1.oe.58.9.093105 · Cited by 3
Image Enhancement and Denoising in Extreme Low-Light Conditions
(2019) · International Journal of Innovative Technology and Exploring Engineering · doi:10.35940/ijitee.a9243.119119 · Cited by 3
“(5) In the present work, we have used the Multi-scale SSIM (MS-SSIM) [29] which typically computes SSIM at multiple”
BackgroundMethodInfluentialComprehensive measure for evaluating image fusion algorithm
Guo, R., Shen, X., & Zhang, X. (2020) · Journal of Electronic Imaging · doi:10.1117/1.jei.29.1.013008 · Cited by 3
“26 The model continuously performs low-pass filtering and subsampling on the image to obtain a set of layers at different scales.”
BackgroundMethodLow Bitrate Image Compression with Discretized Gaussian Mixture Likelihoods
Cheng, Z., Sun, H., & Katto, J. (2020) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw50498.2020.00071 · Cited by 3
“We optimized our models using MS-SSIM quality metrics [34] to achieve better visual quality and distortion term is defined by D(x, x̂) = 1 − MS-SSIM(x, x̂), where the weights in mult-scale SSIM is defined as the default values [0.”
MethodBlind Tone-Mapped Image Quality Assessment Based on Regional Sparse Response and Aesthetics
He, Z., Yu, M., Chen, F., et al. (2020) · Entropy · doi:10.3390/e22080850 · Cited by 3
“Up to now, a large number of perceptual IQA methods designed for LDR images have been proposed [5,6], and can be usually divided into three categories: full-reference (FR), reduced-reference (RR) and no-reference/blind (NR).”
BackgroundYoussef, S. M., el Shehaby, M., & Fayed, S. F. (2020) · Multimedia Tools and Applications · doi:10.1007/s11042-020-09432-1 · Cited by 3
“Image quality is a subjective issue, heavily subjected to different studies [26, 33, 35, 42, 46, 47].”
BackgroundA Subjective Quality Assessment Database for Mobile Video Coding
Zhang, Y., Liu, Z., Wu, W., Yao, R., Chen, Z., & Liu, S. (2020) · IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) · doi:10.1109/mipr49039.2020.00053 · Cited by 4
“…no-reference (NR) or reduced-reference (RR) models, we mainly test some well-known FR models, including Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM) [9], Multiscale SSIM (MS-SSIM) [10], Visual Information Fidelity (VIF) [11], and Video Multi-method Assessment Fusion (VMAF) [12].”
MethodPredicting visible flicker in temporally changing images
Dénes, G., & Mantiuk, R. (2020) · Electronic Imaging · doi:10.2352/issn.2470-1173.2020.11.hvei-233 · Cited by 4
Inpainting-filtering for metal artifact reduction (IMIF-MAR) in computed tomography
Rodríguez-Gallo, Y., Orozco‐Morales, R., & Pérez-Díaz, M. (2021) · Physical and Engineering Sciences in Medicine · doi:10.1007/s13246-021-00990-8 · Cited by 3
“IWMSE is given as: where M is the number of scales, j is the weight given to the jth scale, and the weights are defined in similar ways as in the multiscale SSIM approach [47].”
MethodA Fast Noniterative Visually Lossless Compression of Dental Images Using AGU-M Coder
Krylova, O., Kryvenko, S., Krivenko, S., & Лукин, В. В. (2021) · International Conference on Experience of Designing and Applications of CAD Systems in Microelectronics · doi:10.1109/cadsm52681.2021.9385257 · Cited by 3
“For MS SSIM, distortions are practically invisible is MS SSIM is larger than 0.985 (the maximal value of MS SSIM equals to unity and relates to perfect quality).”
BackgroundInfluentialMP2020: Visual quality assessment database for macro photography images
Sang, Q., Cao, Y., Liu, L., Hu, C., & Wu, X. (2021) · IET Image Processing · doi:10.1049/ipr2.12198 · Cited by 3
Bucci, A., Topini, A., Franchi, M., Zacchini, L., Secciani, N., & Ridolfi, A. (2020) · Global Oceans 2020: Singapore – U.S. Gulf Coast · doi:10.1109/ieeeconf38699.2020.9389055 · Cited by 6
“The SSIM index mathematical formulation [20], [21] is reported in Eq.”
MethodSlice Reconstruction on 3D Medical Image using Optical Flow Approach
Supriyanto, M. I. A., Fajar, A., Sarno, R., et al. (2021) · IEEE Asia Pacific Conference on Wireless and Mobile (APWiMob) · doi:10.1109/apwimob51111.2021.9435238 · Cited by 3
“…Conference on Wireless and Mobile (APWiMob) 978-1-7281-9475-2/21/$31.00 ©2021 IEEE In this paper [16], there is an advanced version of SSIM called the Multi-Scale Structural Similarity Index Method (MS-SSIM) which evaluates various structural similarities of images at different image scales [17].”
MethodInfluentialA temporal attention based appearance model for video object segmentation
Wang, H., Liu, W., & Xing, W. (2021) · Applied Intelligence · doi:10.1007/s10489-021-02547-4 · Cited by 5
“The proposed boundary-aware loss is a mixture loss, which consists of the Euclidean loss of boundary pixels, the SSIM (Structural SIMilarity) [23] loss and the top-k loss [24].”
BackgroundLearning from Multiple Modalities of Imaging Data for Cancer Detection/Diagnosis
Obukhova, N. A., Motyko, A. A., & Pozdeev, A. A. (2021) · Intelligent systems reference library · doi:10.1007/978-3-030-71975-3_3 · Cited by 3
Joint model of gradient magnitude and Gabor features via Spatio-Temporal slice
Bediako, D. O., & Mou, X. (2021) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2021.103204 · Cited by 6
WavNet — Visual saliency detection using Discrete Wavelet Convolutional Neural Network
Sasibhooshan, R., Kumaraswamy, S., & Sasidharan, S. (2021) · Journal of Visual Communication and Image Representation · doi:10.1016/j.jvcir.2021.103236 · Cited by 3
A new objective metric to predict image quality using deep neural networks
Ebrahimi, T., & Akyazi, P. (2018) · SPIE Applications of Digital Image Processing · doi:10.1117/12.2322709 · Cited by 5
Ding, L., Wang, P., & Huang, H. (2021) · Signal Processing Image Communication · doi:10.1016/j.image.2021.116446 · Cited by 3
Quality Assessment of 3D Synthesized Images Based on Textural and Structural Distortion Estimation
Alvi, H. M. U. H., Farid, M. S., Khan, M. H., & Grzegorzek, M. (2021) · Applied Sciences · doi:10.3390/app11062666 · Cited by 3
“We compared the performance of the proposed metric with widely used 2D quality assessment metrics: PSNR, SSIM [60], VSNR [74], IFC [75], MSSIM [29], VIF [76], and UQI [77].”
MethodDNA Image Storage Using a Scheme Based on Fuzzy Matching on Natural Genome
Zhang, J., Chen, S., Zhang, H., Shen, Y., & Ping, Z. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-91415-8_48 · Cited by 3
Face Reconstruction with Variational Autoencoder and Face Masks
Toledo, R. S., & Antonelo, E. A. (2021) · doi:10.5753/eniac.2021.18282 · Cited by 3
“This work was later extended to Multi-Scale Structural Similarity (MS-SSIM) [Wang et al. 2003] as SSIM depends on the right scale of the viewing conditions like display resolution and viewing distance; differently than MS-SSIM, which incorporates easier image details at different resolutions.”
MethodNVSS: High-quality Novel View Selfie Synthesis
Bian, J.-W., Zhan, H., & Reid, I. D. (2021) · International Conference on 3D Vision (3DV) · doi:10.1109/3dv53792.2021.00116 · Cited by 3
“The PSNR and MS-SSIM [45] are used to evaluate the low-level image differences to the ground truth, and we use the LPIPS [47] metric to evaluate the high-level perception differences.”
MethodChaDRaL: RGB Image Encryption based on 3D Chaotic Map, DNA, RSA and LSB
Parekh, N., & D’Mello, L. (2021) · International Conference on Artificial Intelligence and Machine Vision (AIMV) · doi:10.1109/aimv53313.2021.9670921 · Cited by 5
A new paradigm for image quality assessment based on human abstract layers of quality perception
Khosravi, M. H., & Hassanpour, H. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-12478-y · Cited by 3
“The main reason for employing the multi-scale scheme is the fact that recognizability of image contents depends on its resolution, distance from the image screen to the observer, and the perceptual capability of the observer’s visual system [36].”
BackgroundMethodInfluentialKhandouzi, A., & Ezoji, M. (2022) · Iranian Conference on Machine Vision and Image Processing · doi:10.1109/mvip53647.2022.9738773 · Cited by 4
Improved U‐Net‐Like Network for Visual Saliency Detection Based on Pyramid Feature Attention
Gong, X., Qingge, L., Liu, Q., & Yang, P. H. (2022) · Wireless Communications and Mobile Computing · doi:10.1155/2022/1108462 · Cited by 3
“The SSIM loss function, which can capture structural information in images, was originally proposed for image quality assessment [41].”
BackgroundNo-Reference Quality Assessment of Stereoscopic Video Based on Deep Frequency Perception
Xiao, S., Wen, J., Yang, J., & Zhou, Y. (2022) · Workshop on Quality of Experience in Visual Multimedia Applications (QoEVMA) · doi:10.1145/3552469.3555711 · Cited by 3
“State-of-the-art IQA/VQA models, including peak-signal-to-noise ratio (PSNR), structural similarity (SSIM) [26], MS-SSIM, and Video Quality Metric (VQM) [21], etc., are applied to predict the quality of single view of the stereo video.”
MethodInfluentialEfficient Human Vision Inspired Action Recognition Using Adaptive Spatiotemporal Sampling
Mac, K.-N. C., Do, M. N., & Vo, M. P. (2023) · IEEE Transactions on Image Processing · doi:10.1109/tip.2023.3310661 · Cited by 3
“We minimize the negative SSIM score since the default SSIM ranges from 0 to 1.”
MethodInfluentialVideo Quality Assessment Based on Content-Partitioned Multi-Scale Structural Similarity
Yao, J., Xie, Y., Tan, M. J., Li, Z., Qi, J., & Gao, L. (2011) · Lecture notes in electrical engineering · doi:10.1007/978-3-642-25541-0_33 · Cited by 3
“MSSIM [3] is an extension of SSIM, which is accomplished over multiple scales of the reference and distorted image patches by iteratively low-pass filtering and down-sampling the signals.”
MethodOn the Usefulness of Combined Metrics for 3D Image Quality Assessment
Okarma, K. (2014) · Advances in intelligent systems and computing · doi:10.1007/978-3-319-10662-5_17 · Cited by 3
A visual consistent adaptive image thresholding method
Jiao, X., & Wu, T. (2015) · The Imaging Science Journal · doi:10.1080/13682199.2015.1115193 · Cited by 3
“SG(i) = 2G1(xi) · G2(yi)+ t2 G1(2)(xi)+ G2(2)(yi)+ t2 , (30)”
Investigation of the effectiveness of video quality metrics in video pre-processing
Chilukamari, J., Kannangara, S., & Maxwell, G. M. (2013) · IEEE International Conference on Consumer Electronics - Berlin (ICCE-Berlin) · doi:10.1109/icce-berlin.2013.6698005 · Cited by 5
“The objective of this work is to investigate the effectiveness of PSNR and a number of state-of-the-art full reference objective video quality metrics, namely, Structural Similarity Index (SSIM) [10], Multi-Scale Structural Similarity Index (MS-SSIM) [11], Video Quality Metric (VQM) [12] and no reference video quality…”
MethodInfluentialLi, A., Wu, C., Chen, Y., & Ni, B. (2020) · ACM International Symposium on Theory, Algorithmic Foundations, and Protocol Design for Mobile Networks and Mobile Computing (MobiHoc) · doi:10.1145/3397166.3409140 · Cited by 4
“es and interpolated intermediate frames by attempting to predict human perceptual similarity judgments. In this experiment, we adopt the widely applied metric multi-scale structuralsimilarity(MS-SSIM)[39, 50] for measuring the frame similarity. MS-SSIM is a multi-scale perceptual similarity metric that attempts to pay…”
MethodPark, H., Park, S.-C., & Joo, Y. H. (2020) · IEEE Access · doi:10.1109/access.2020.3036893 · Cited by 3
“…distance between the real image and the reconstructed fake image within the main network as follows : In general, full-reference IQA algorithms such as SSIM, MS-SSIM [48], CW-SSIM [49], and FSIM [50] obtain the final quality score from the average or weighted average of the regional quality map.”
MethodOptimum HEVC Quantization Parameter for Cloud Gaming
Ghafari, M., Ajoodani, A., & Hemami, S. M. S. (2021) · doi:10.1109/ipria53572.2021.9483484 · Cited by 4
“org/wiki/Peak_signal-to-noise_ratio [12] Wang, Z.”
BackgroundDigital Video Quality Assessment Algorithms
Moorthy, A. K., Seshadrinathan, K., & Bovik, A. C. (2009) · Springer eBooks · doi:10.1007/978-0-387-89024-1_6 · Cited by 3
“In order to better evaluate quality over multiple resolutions, the multi-scale SSIM (MS-SSIM) index was proposed in [ 31 ].”
BackgroundMethodSelf-citationAccurate and discernible photocollages
Miller, J. (2010) · University Library - University of Saskatchewan (University of Saskatchewan) · Cited by 3
RETINAL IMAGE QUALITY ASSESSMENT USING SUPERVISED CLASSIFICATION
Fasih, M. (2014) · PolyPublie (École Polytechnique de Montréal) · Cited by 4
Humans are Still the Best Lossy Image Compressors
Bhown, A., Mukherjee, S., Yang, S., et al. (2019) · DCC · doi:10.1109/dcc.2019.00070 · Cited by 3
“Indeed, there has been a large body of work in the computer vision community [7][8][9] in order to better understand human perception, and hence a loss function governing human vision.”
BackgroundBlind Motion Deblurring with Cycle Generative Adversarial NetworksPreprint
Yuan, Q., Li, J., Zhang, L., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1901.01641 · Cited by 3
“use PSNR, SSIM[44] and some new full reference methods MS-SSIM [45], IFC[40], VIF[39] and one No Reference image quality assessment NIQE [30] in all experiments.”
Technical Report on Visual Quality Assessment for Frame InterpolationPreprint
Men, H., Lin, H., Hosu, V., Maurer, D., Bruhn, A., & Saupe, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1901.05362 · Cited by 4
“that consider the HVS, such as SSIM [5], MS-SSIM [6], FSIM [7], VSI [8].”
MethodExtreme Image Coding via Multiscale Autoencoders With Generative Adversarial OptimizationPreprint
Huang, C., Liu, H., Chen, T., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.03851 · Cited by 3
“This is mainly due to the reason that visual sensitive information (i.e., perceptual significance) can not be well preserved using conventional quality optimization criteria, such as peak signal-to-noise ratio (PSNR) and multiscale structural similarity (MS-SSIM) [8], at such extreme compression scenario.”
BackgroundGenerative Collaborative Networks for Single Image Super-ResolutionPreprint
Seddik, M. E. A., Tamaazousti, M., & Lin, J. Y. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.10467 · Cited by 3
“In fact, since MSE consists in a pixel-wise images differences, its ability to recover high texture details is limited [1, 5, 6, 7].”
BackgroundPerceptual Quality Study on Deep Learning based Image CompressionPreprint
Cheng, Z., Akyazi, P., Sun, H., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.03951 · Cited by 3
“PSNR or MS-SSIM [16].”
Transforming Thermal Images to Visible Spectrum Images Using Deep Learning
Nyberg, A. (2018) · KTH Publication Database DiVA (KTH Royal Institute of Technology) · Cited by 3
Structural Decompositions for End-to-End Relighting.Preprint
Nestmeyer, T., Matthews, I. A., Lalonde, J., & Lehrmann, A. (2019) · arXiv (Cornell University) · Cited by 2
Stereoscopic Omnidirectional Image Quality Assessment Based on Predictive Coding TheoryPreprint
Chen, Z., Xu, J., Lin, C., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.05165 · Cited by 3
“Firstly, 2D IQA metrics are used for conventional 2D image quality assessment, we choose five commonly used metrics for comparison, namely PSNR, SSIM [11], MSSSIM [12], FSIM [13], and VSI [64].”
BackgroundMethodDeep Convolutional AutoEncoder-based Lossy Image CompressionPreprint
Cheng, Z., Sun, H., Takeuchi, M., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.09535 · Cited by 3
“The quality of the reconstructed images is measured using the quality metrics PSNR and MSSSIM [12], which measure the objective quality and perceived quality, respectively.”
NOVA: QoE-driven Optimization of DASH-based Video Delivery in NetworksPreprint
Joseph, V., & Veciana, G. A. D. (2013) · arXiv (Cornell University) · doi:10.48550/arxiv.1307.7210 · Cited by 6
“We obtained proxy subjective VQA metric for the representations based on the corresponding value of MSSSIM-Y metric ([22]).”
MethodAdversarial Pixel-Level Generation of Semantic ImagesPreprint
Ghelfi, E., Galeone, P., De Simoni, M., & Mattia, F. D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.12195 · Cited by 6
“MultiScale Structural Similarity The third metric is the MultiScale Structural Similarity (Wang et al., 2003) (MSSIM).”
BackgroundUGC-VIDEO: perceptual quality assessment of user-generated videosPreprint
Li, Y., Meng, S., Zhang, X., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.11517 · Cited by 3
Predictive Auto-Encoding Network for Blind Stereoscopic Image Quality Assessment.Preprint
Xu, J., Zhou, W., Chen, Z., Ling, S., & Callet, P. L. (2019) · arXiv (Cornell University) · Cited by 4
Learning End-to-End Lossy Image Compression: A BenchmarkPreprint
Hu, Y., Yang, W., Ma, Z., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.03711 · Cited by 3
“Because end-to-end learned models can be tuned with both PSNR and MS-SSIM, we are able to investigate the relationship between different metrics and objectives.”
MethodInfluentialPatel, Y., Appalaraju, S., & Manmatha, R. (2020) · arXiv (Cornell University) · Cited by 4
Perceptually inspired weighted MSE optimization using irregularity-aware graph Fourier transformPreprint
Lu, K.-S., Ortega, A., Mukherjee, D., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.08558 · Cited by 3
Learning an Adaptive Model for Extreme Low-light Raw Image ProcessingPreprint
Fu, Q., Di, X., & Zhang, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.10447 · Cited by 3
“By introducing SSIM into loss function, our method shows more promising results in images with text and has advantages over SID in keeping more edge information, e.g. the images in the first row of Fig.”
BackgroundMethodInfluentialEnd-to-end Optimized Video Compression with MV-Residual PredictionPreprint
Wu, X., Zhang, Z., Feng, J., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.12945 · Cited by 3
“In TUCODEC SSIM the loss D = 1 − Lmsssim is defined for the perceptional loss where Lmsssim is as defined in [16].”
BackgroundRahim, T., Usman, M. A., & Shin, S. Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.02671 · Cited by 6
“The multi-scale SSIM index metric is an improved form of the SSIM metric designed to compute the quality of a frame of clip on multiple scales [35].”
MethodTowards Defeating Misclassification Attacks Against Transfer LearningPreprint
Wu, B. Y., Wang, S., Yuan, X., Wang, C., Rudolph, C., & Yang, X. (2019) · arXiv (Cornell University) · Cited by 3
Continual Learning: Tackling Catastrophic Forgetting in Deep Neural Networks with Replay ProcessesPreprint
Lesort, T. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.00487 · Cited by 37
“• Multi-scale structural similarity Multi-scale structural similarity Wang et al. (2003) (MSSIM) is a measurement that gives a way to incorporate image details at different resolutions in order to compare two images.”
BackgroundChannel-Level Variable Quantization Network for Deep Image CompressionPreprint
Zhong, Z., Akutsu, H., & Aizawa, K. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.12619 · Cited by 3
“Evaluating on the Kodak dataset, this toy model yields an average MSSSIM [Wang et al., 2003] of 0.922 at an average rate of 0.112 bits per pixel (BPP).”
MethodInfluentialAIBench Training: Balanced Industry-Standard AI Training BenchmarkingPreprint
Tang, F., Gao, W., Zhan, J., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.14690 · Cited by 3
“Reference Quality : The metric is 0.99 MS-SSIM (Multi-Scale-Structural Similarity Index [53]).”
InfluentialSelf-citationAWNet: Attentive Wavelet Network for Image ISPPreprint
Dai, L., Liu, X., Li, C., & Chen, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.09228 · Cited by 13
“We also employ the structural similarity (SSIM) loss LSSIM [36] that is aiming to reconstruct the RGB images by enhancing on structural similarity index.”
Improving the Reconstruction of Disentangled Representation Learners via Multi-Stage ModelingPreprint
Srivastava, A., Bansal, Y., Ding, Y., Hurwitz, C., Xu, K., Egger, B., Sattigeri, P., Tenenbaum, J. B., Sudjianto, A., Le, P., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.13187 · Cited by 7
“We choose FID instead of reconstruction error for our main metric as L1/L2 metrics are known to not correlate well with perceived image quality (Wang et al., 2003, 2004).”
MethodWang, Y., Feng, M., Ward, R. K., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.05254 · Cited by 3
“To reliably evaluate the perceptual improvement of the proposed structure-aware attacks, we adopt four image quality assessment (IQA) metrics: multiSim3 [39], Feature Similarity for color images (FSIMc) [40], Natural Image Quality Evaluator (NIQE) [41] and Mean Opinion Score (MOS) [42].”
MethodSelf-citationEmotional Semantics-Preserved and Feature-Aligned CycleGAN for Visual Emotion AdaptationPreprint
Zhao, S., Chen, X., Yue, X., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.12470 · Cited by 3
“According to [73], MS-SSIM is defined as”
MethodPrivacy Preserving for Medical Image Analysis via Non-Linear Deformation ProxyPreprint
Kim, B. N., Dolz, J., Desrosiers, C., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.12835 · Cited by 3
“In practice, we use a multi-scale structural similarity (MS-SSIM) which computes the SSIM at multiple image scales via subsampling [34].”
MethodLearned Block-based Hybrid Image CompressionPreprint
Wu, Y., Li, X., Zhang, Z., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.09550 · Cited by 3
Counterfactual Generation with KnockoffsPreprint
Popescu, O.-I., Shadaydeh, M., & Denzler, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.00951 · Cited by 8
“We quantify the correlation using the multi-structural similarity index (MS-SSIM) [22] between original images and perturbed images, and the in-distribution-ness as the classifier target probability.”
MethodEvent-based Synthetic Aperture Imaging with a Hybrid NetworkPreprint
Zhang, X., Wei, L., Yu, L., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.02376 · Cited by 3
“In the dense occlusion experiment, the metrics PSNR and SSIM [22] are employed for quantitative comparison, where the aligned APS images captured by DAVIS346 are considered as the ground truth.”
MethodInfluentialLearning-based Compression for Material and Texture RecognitionPreprint
Deng, Y., & Karam, L. J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.10065 · Cited by 4
“Specifically, when compared to JPEG/JPEG 2000 standard, learning-based codecs can produce higher compression quality in terms of some common perceptual objective quality metrics such as PSNR (peak signal-to-noise ratio) and SSIM (structural similarity) [13] / MS-SSIM (multi-scale structural similarity) [14] for some F…”
ResultGündüzalp, D., Cengiz, B., Unal, M. O., & Yıldırım, İ. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.14130 · Cited by 14
“Therefore, the SSIM is used as a second image quality metric to evaluate the similarity of luminance, contrast, and structure [32]. where µ x and µ y represent the average of reconstructed and ground truth images, respectively. σ 2 x and σ 2 y indicate the variance of reconstructed and ground truth…”
MethodHLIC: Harmonizing Optimization Metrics in Learned Image Compression by Reinforcement LearningPreprint
Sun, B., Gu, M. H., He, D., Xü, T., Wang, Y., & Qin, H. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.14863 · Cited by 4
“In image and video coding tasks, peak signal-to-noise ratio (PSNR) and multiscale structural similarity (MS-SSIM) [39] are two most common metrics.”
BackgroundNeRV: Neural Representations for VideosPreprint
Chen, H., He, B., Wang, H., Ren, Y., Lim, S.-N., & Shrivastava, A. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.13903 · Cited by 345
“First, to achieve the comparable PSNR and MS-SSIM performances, the training time of our proposed approach is longer than the encoding time of traditional video compression methods.”
MethodInfluentialGeological Facies Modeling Based on Progressive Growing of Generative Adversarial Networks (GANs)Preprint
Song, S., Mukerji, T., & Hou, J. (2020) · doi:10.31223/osf.io/ycufs · Cited by 3
Metadata
- Authors
- Wang, Z., Simoncelli, E. P., & Bovik, A. C.
- Affiliations
- Courant Institute of Mathematical Sciences (US)
- The University of Texas at Austin (US)
- Venue
- The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003
- Publisher
- IEEE
- Published
- 2004-07-08
- Pages
- 1398-1402
- ISSN
- 1058-6393 (ISSN-L)
- OpenAlex
- W1580389772
- Cited by
- 6,253
- References
- 15
- Citations by year
- 2026: 344 · 2025: 675 · 2024: 729 · 2023: 668 · 2022: 662 · 2021: 641 · 2020: 563 · 2019: 438 · 2018: 305 · 2017: 267 · 2016: 206 · 2015: 173 · 2014: 166 · 2013: 112 · 2012: 89
- FWCI
- 3.98
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Asilomar Conference on Signals, Systems and Computers (ACSSC) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image and Video Quality Assessment
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image Retrieval and Classification Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Digital Media Forensic Detection
- Keywords
- structural similarity, image quality assessment, multiscale analysis, human visual system, perceptual image quality, image generation, parameter calibration, viewing conditions
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.