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 4,301–4,400 of 6,047
Yu, X., & Xie, W. C. (2022) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2022.3157766 · Cited by 2
“Furthermore, there is three reference image quality assessment (RIQA) methods for evaluating the performance of restored results, including PSNR, structural similarity index (SSIM), and multiscale structural similarity (MSSSIM) [46].”
MethodA gradient-based framework for 3D print appearance optimization
Nindel, T. K., Iser, T., Rittig, T., et al. (2021) · ACM Transactions on Graphics · doi:10.1145/3476576.3476767 · Cited by 2
“There are other SSIM alternatives available [Wang et al. 2003; Xue et al. 2013; Zhang et al. 2011], further improving its behavior.”
MethodStudy of Noise Removal Techniques for Digital Images
Patel, P., Jena, B., Sahoo, B., Patel, P., & Majhi, B. (2016) · IGI Global eBooks · doi:10.4018/978-1-5225-0983-7.ch044 · Cited by 3
Self-Supervised Variable Rate Image Compression using Visual Attention
Sinha, A. K., Moorthi, S. M., & Dhar, D. (2022) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw56347.2022.00179 · Cited by 3
“Following [4,5], The distortion is measured using MS-SSIM [21] and PSNR in RGB space.”
MethodInfluentialFractal pyramid low-light image enhancement network with illumination information
Sun, T., Fan, G., & Gan, M. (2022) · Journal of Electronic Imaging · doi:10.1117/1.jei.31.4.043050 · Cited by 2
Evaluating the Stability of Deep Image Quality Assessment with Respect to Image Scaling
Tsubota, K., Akutsu, H., & Aizawa, K. (2022) · IEICE Transactions on Information and Systems · doi:10.1587/transinf.2022edl8025 · Cited by 2
“For reference, we evaluated three traditional methods: PSNR, SSIM [3], and MS-SSIM [15].”
MethodSearching Lightweight Neural Network for Image Signal Processing
Lin, H., Li, L., Zheng, X., Chao, F., & Ji, R. (2022) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3503161.3548288 · Cited by 2
“Following previous works [7, 18], we use a mixed loss function that contains L1 pixel loss, perceptual loss [19], and the structural similarity (SSIM) loss [35].”
MethodA Self-Supervised Method for Infrared and Visible Image Fusion
Lin, X., Zhou, G., Zeng, W., Tu, X., Huang, Y., & Ding, X. (2022) · Proceedings - International Conference on Image Processing · doi:10.1109/icip46576.2022.9897731 · Cited by 5
“(3) multi-scale structural similarity MSSSIM [18].”
Wang, Y., Li, H., & Jiang, Q. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2022.1022041 · Cited by 3
“There have been many FR image quality assessment (IQA) metrics designed for 2D plane images (Wang et al., 2003; Zhou et al., 2004; Sheikh et al., 2006; Zhang et al., 2011; Xue et al., 2014).”
BackgroundSong, H., Kang, J., & Boctor, E. M. (2022) · Ultrasonics · doi:10.1016/j.ultras.2022.106908 · Cited by 2
Semi-Supervised Disentanglement of Tactile Contact Geometry from Sliding-Induced Shear
Gupta, A. K., Church, A., & Lepora, N. F. (2022) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros47612.2022.9982165 · Cited by 2
A Comparison of Quality Assessment Metrics on Image Resolution Enhancement Artifacts
Zheng, R., Jiang, X., Ma, Y., & Wang, L. (2022) · International Conference on Culture-oriented Science & Technology (ICCST) · doi:10.1109/cost57098.2022.00049 · Cited by 4
“Some schemes aim at improving SSIM by considering the human visual system, IQA metrics such as MS-SSIM [9] achieved by performing SSIM on multiple scales of images, information fidelity criterion with extension visual information fidelity VIF [10], gradient magnitudes combined with phase congruency…”
MethodMutual Information Estimation-Based Disentangled Representation Network for Medical Image Fusion
Huang, W., Zhang, H., Li, Z., et al. (2022) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm55620.2022.9995221 · Cited by 2
“, MS-SSIM [22], sum of the correlations of differences (SCD) [23], human visual system (HVS) [24] and visual information fidelity (VIFF) [25] to evaluate the fusion results.”
BackgroundInfluentialCho, Y., Lee, J., Yang, S., Kim, J., Park, Y., Lee, H., Khan, M. A., Kim, D., & Choo, J. (2023) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv56688.2023.00186 · Cited by 2
“As discussed in a previous work [32], we guess that the PSNR score is not sufficient to measure how naturally a model can generate if only partial conditions are given.”
CoNFies: Controllable Neural Face Avatars
Yu, H., Niinuma, K., & Jeni, L. A. (2023) · International Conference on Automatic Face and Gesture Recognition · doi:10.1109/fg57933.2023.10042763 · Cited by 6
“We use Peak Signal-toNoise Ratio (PSNR), Multi-scale Structural Similarity (MSSSIM) [39] and Learned Perceptual Image Patch Similarity (LPIPS) [45] to quantitatively evaluate our method compared with NeRF [25], NeRF + Latent, Nerfies [29], HyperNeRF [30], CoNeRF-M and CoNeRF [17], which is consistent with [17].”
MethodMulti-color super-resolution microscopy accelerated by a neural network
Narayanasamy, K. K., Rahm, J. V., Jang, S., & Heilemann, M. (2023) · Single Molecule Spectroscopy and Superresolution Imaging · doi:10.1117/12.2657442 · Cited by 2
Salient Object Detection Based on Transformer and Multi-scale Feature Fusion
Zhao, L., & Wang, H. (2023) · International Conference on Neural Networks, Information and Communication Engineering (NNICE) · doi:10.1109/nnice58320.2023.10105709 · Cited by 2
“…a mixed loss function is designed at each stage of the decoder for significance monitoring, including the classic pixel level BCE loss, patch level SSIM loss [13] and map level IoU loss [14]: (5) L bce (·) is BCE loss, S sal is predicted saliency map, and G sal is ground truth of saliency map.”
BackgroundQuality comparison of 360° 8K images compressed by conventional and deep learning algorithms
Kufa, J., & Budac, A. (2023) · International Conference Radioelektronika · doi:10.1109/radioelektronika57919.2023.10109066 · Cited by 2
“Six full reference objective quality metrics recommended by [1], [2], [14], and [15] are used, which are Weighted Spherical Peak Signal to Noise Ratio (WS-PSNR) [16], Multiscale Structural Similarity (MSSSIM) [17], Visual Information Fidelity pixel domain version (VIFp) [18], Feature Similarity Chrominance (FSIMc) [19…”
MethodLiNuIQA: Lightweight No-Reference Image Quality Assessment Based on Non-Uniform Weighting
Kim, W., Hahm, C.-H., Baijal, A., Kim, N.-U., Cho, I., & Koo, J. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10096440 · Cited by 2
“IQA techniques can be divided to 2 types of category, full-reference (FR) [1, 2] and no-reference (NR) [3, 4].”
BackgroundQuality Assessment of Screen Content Videos
Motamednia, H., Cheraaqee, P., Mansouri, A., & Mahmoudi-Aznaveh, A. (2023) · doi:10.1109/ipria59240.2023.10147176 · Cited by 3
“PSNR, VIF [32], SSIM [29], MSSIM [31], PSIM [54], and GMSD [30] are general IQA metrics; SQMS [35], ESIM [14], SVQI [36], and MDOGS [37] are SCI-specific QA methods; MOVIE [39], SpEED [43], ViS3 [55], STMAD [38], VMAF [41], and FAST [56] are the general video quality assessment metrics; and SGFTM [26] is the SCV-speci…”
MethodEffects of Different Full-Reference Quality Assessment Metrics in End-to-End Deep Video Coding
Xian, W., Chen, B., Fang, B., et al. (2023) · Electronics · doi:10.3390/electronics12143036 · Cited by 2
On the assessment of high-quality images: advances on the JPEG AIC-3 activity
Testolina, M., Upenik, E., & Ebrahimi, T. (2023) · doi:10.1117/12.2677347 · Cited by 2
An End-to-End Video Coding Method via Adaptive Vision Transformer
Yang, H., Zhou, M., Shang, Z., et al. (2023) · International Journal of Pattern Recognition and Artificial Intelligence · doi:10.1142/s021800142354023x · Cited by 2
An Evaluation of the Impact of Distance on Perceptual Quality of Textured 3D Meshes
Nguyen, D. V., Hien, T. T., Tran, H. T. T., et al. (2023) · IEICE Transactions on Information and Systems · doi:10.1587/transinf.2023mul0001 · Cited by 3
Thermal Image Super-Resolution Methods Using Neural Networks
Didenko, A., Oliinyk, А., & Субботин, С. А. (2023) · doi:10.32782/cmis/3392-18 · Cited by 2
“To evaluate trained models SSIM [28] (structural similarity index measure) and RSNR (peak signal-to-noise ratio) metrics were used.”
MethodLow-complexity Transform Network Architecture for JPEG AI Image Codec
Pan, X., Ding, D., Wang, L., et al. (2023) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip59821.2023.10402641 · Cited by 2
“Objective quality metrics include MS-SSIM [15], FSIM [16], NLPD [17], IW-SSIM [18], VMAF [19], VIF [20] and PSNR-HVS-M [21], between compressed and original image, at these target bi-trates.”
MethodAutoencoder based image quality metric for modelling semantic noise in semantic communications
Samarathunga, P., Fernando, T., Gowrisetty, V. S. S., Atulugama, T., & Fernando, A. (2024) · Electronics Letters · doi:10.1049/ell2.13115 · Cited by 3
The Accuracy and Precision of Memory for Natural Scenes: A Walk in the Park
Westebbe, L., Liang, Y., & Blaser, E. (2024) · Open Mind · doi:10.1162/opmi_a_00122 · Cited by 1
“However, three different routes were used, each of which inherently had differing levels of inter-scene ‘ self-similarity ’— a measure of visual change per meter traveled based on a Multiscale Structural Similarity analysis (Wang et al., 2003).”
MethodAnis, M., Kim, Y.-J., & Yi, Y. K. (2023) · Building Simulation Conference proceedings · doi:10.26868/25222708.2023.1219 · Cited by 2
“SSIM is a measure of the similarity between two images based on the pixel values on various windows of the images (Wang et al., 2003).”
BackgroundDisparity-Guided Multi-View Interaction Network for Light Field Reflection Removal
Liu, Y., Weng, W., Gao, R., Xiao, Z., Zhang, Y., & Xiong, Z. (2024) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2024.3394773 · Cited by 2
“The loss function L ssim is defined using the structural similarity (SSIM) [79] and can better supervise the image structural similarity like illuminance, contrast, and etc .”
MethodA single image deblurring approach based on a fractional order dark channel prior
Yu, X., Xie, W., & Yu, J. (2022) · International Journal of Applied Mathematics and Computer Science · doi:10.34768/amcs-2022-0032 · Cited by 3
“Figure 3 shows the average PSNR, SSIM, and MSSSIM performance on this benchmark data set by some classical methods.”
MethodSynthetic X-ray Image Generation for Non-Destructive Testing using Generative Adversarial Networks
Gosswami, B. M., Chanda, R. K., & Wittenberg, T. C. (2024) · e-Journal of Nondestructive Testing · doi:10.58286/29930 · Cited by 3
“• Structural Similarity (SSIM) : SSIM [9] and MSSIM [10] metrics measure image similarities.”
BackgroundRankDVQA-Mini: Knowledge Distillation-Driven Deep Video Quality Assessment
Feng, C., Danier, D., Wang, H., Zhang, F., Vallade, B. Q. A., Mackin, A., & Bull, D. (2024) · Picture Coding Symposium · doi:10.1109/pcs60826.2024.10566364 · Cited by 3
An Improved Neural Network Approach to End-to-end Video Compression
Fu, L., Wang, P., & Wang, X. (2024) · International Conference on Computer Information and Big Data Applications (CIBDA) · doi:10.1145/3671151.3671162 · Cited by 2
Image Quality Assessment Based on SIFT and SSIM
Lu, W., Li, C., Shi, Y., & Sun, X. (2014) · Communications in computer and information science · doi:10.1007/978-3-662-45498-5_1 · Cited by 2
Full-Reference Image Quality Assessment Measure Based on Color Distortion
Seghir, Z. A., & Hachouf, F. (2015) · IFIP advances in information and communication technology · doi:10.1007/978-3-319-19578-0_6 · Cited by 2
Objective No-Reference Video Quality Assessment Method Based on Spatio-Temporal Pixel Analysis
da Silva, W. B., Fonseca, K. V. O., & Pohl, A. D. A. P. (2015) · IEICE Transactions on Information and Systems · doi:10.1587/transinf.2014edp7155 · Cited by 2
“The performance results based on the calculation of the PLCC coe ffi cients of the proposed method are compared with the results of two other metrics: Peak Signal-to-Noise Ratio (PSNR) and Multi-Scale Structural SIMilarity index (MS-SSIM) [38], which are FR metrics.”
MethodFrequency structure analysis for IQA
Li, X., Lu, W. B., Tao, D., & Gao, X. (2008) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/icsmc.2008.4811627 · Cited by 2
Eyetracking based approach to objective image quality assessment
Fliegel, K. (2008) · IEEE International Carnahan Conference on Security Technology (ICCST) · doi:10.1109/ccst.2008.4751330 · Cited by 2
“Based on this fact a combination of both approaches has been introduced as a multi scale SSIM [7].”
Banterle, F., Artusi, A., Šikudová, E., et al. (2016) · Signal Processing Image Communication · doi:10.1016/j.image.2016.09.004 · Cited by 2
“, LDR images, we used classic LDR quality metrics/indices such as PSNR, SSIM [45], and multi-scale SSIM (MS-SSIM) [44].”
MethodImage quality assessment based on properties of HVS and principle of image structure
Mahamud, S. T., & Rahmatullah, B. (2015) · International Conference on Image and Vision Computing New Zealand (IVCNZ) · doi:10.1109/ivcnz.2015.7761557 · Cited by 2
Comments on objective quality assessment of JPEG images with visible differences
Bondžulić, B. P., Pavlović, B., Andrić, M., & Petrovic, V. S. (2017) · International Conference on Advanced Technologies, Systems and Services in Telecommunications (TELSIKS) · doi:10.1109/telsks.2017.8246323 · Cited by 2
“This paper provides objective measures performance comparison results including conventional PSNR and actual structural similarity index (SSIM) [3], as well as its relatives – universal image quality index (UIQI), multi-scale structural similarity (MS-SSIM) [4], gradient-based structural similarity (GSSIM) [5] and add…”
BackgroundOnly-Reference Video Quality Assessment for Video Coding Using Convolutional Neural Network
Dinh, K. Q., Lee, J., Kim, J., et al. (2018) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2018.8451262 · Cited by 2
“The VQANet is trained with PSNR, SSIM [3], and MSSSIM [4], which are three established image quality assessment metrics popularly reported for work in video coding community.”
BackgroundMethodInfluentialEilertsen, G. (2018) · Linköping studies in science and technology. Dissertations · doi:10.3384/diss.diva-147843 · Cited by 2
“There are also measures that better agree with perceived visual quality, such as the multi-scale structural similarity (MS-SSIM) index [256] or the HDR visual difference predictor (HDR-VDP-2) [172].”
BackgroundMethodNo-Reference Quality Assessment for Stereoscopic 3D Images Based on Binocular Visual Perception
Chen, L., & Zhao, J. (2018) · IEEE International Symposium on Haptic, Audio and Visual Environments and Games (HAVE) · doi:10.1109/have.2018.8548066 · Cited by 2
A Robust Background Initialization Method Based on Stable Image Patches
Yu, L., & Guo, W. (2018) · Chinese Automation Congress · doi:10.1109/cac.2018.8623774 · Cited by 2
Zhang, L., Saidi, I., Tian, S., Barriac, V., & Déforges, O. (2019) · Journal of Electronic Imaging · doi:10.1117/1.jei.28.2.023001 · Cited by 2
Spatial Error Concealment by Jointing Gauss Bayes Model and SVD for High Efficiency Video Coding
Zhou, M., Mao, Q., Chen, Z., et al. (2019) · International Journal of Pattern Recognition and Artificial Intelligence · doi:10.1142/s0218001419540375 · Cited by 2
Chetouani, A., Hassouni, M. E., & Jennane, R. (2018) · International Symposium on Signal, Image, Video and Communications (ISIVC) · doi:10.1109/isivc.2018.8709203 · Cited by 2
“670 Wang [29] 0.”
Depth Image Quality Assessment for View Synthesis Based on Weighted Edge Similarity
Li, L., Chen, X., Zhou, Y., et al. (2019) · Cited by 4
Rajput, A. S., & Raman, B. (2019) · IEEE ... International Conference on Cloud Computing · doi:10.1109/cloud.2019.00024 · Cited by 3
“[16], MSSSIM evaluates structural similarity between two images and returns their matching score in the range [0,1].”
BackgroundImproved Noisy Image Quality Assessment Using Multilayer Neural Networks
Rubel, A., Rubel, O. S., Abramova, V., Проскура, Г., & Лукин, В. В. (2019) · IEEE Ukraine Conference on Electrical and Computer Engineering (UKRCON) · doi:10.1109/ukrcon.2019.8879950 · Cited by 2
“To date, various automatic image quality assessment (IQA) techniques have been proposed in order to replace subjective evaluations [4-7].”
MethodSüzme, N. Ö., & Güraksın, G. E. (2020) · Lecture notes on data engineering and communications technologies · doi:10.1007/978-3-030-36178-5_14 · Cited by 2
Enhanced Image Compression System
Mohan, P. (2019) · International Journal of Mobile Computing and Application · doi:10.14445/23939141/ijmca-v6i3p101 · Cited by 2
Natural Images Enhancement Using Structure Extraction and Retinex
Du, X., & Xia, Y. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-40605-9_35 · Cited by 2
Synthesis and Edition of Ultrasound Images via Sketch Guided Progressive Growing GANs
Liang, J., Yang, X. Y., Li, H., et al. (2020) · White Rose Research Online (University of Leeds, The University of Sheffield, University of York) · doi:10.48550/arxiv.2004.00226 · Cited by 2
“Table 1 further lists the numerical evaluation metrics, including Freshet Inception Distance (FID) [14], Kernel Inception Distance (KID) [15] and multi-scale structural similarity (MS-SSIM) [16].”
BackgroundAdaptive switching interpolation filter for restoring impulse corrupted digital images
Varghese, J., Subash, S., Sridhar, K. P., et al. (2020) · IET Image Processing · doi:10.1049/iet-ipr.2019.1445 · Cited by 2
“Experimental analysis in terms of various objectives and subjective quantitative metrics such as PSNR, MSSIM, IEF, and FSIM demarcated that the new ASIF produce better quality results than other comparative algorithms.”
MethodInfluentialBouida, A., Beladgham, M., Bassou, A., et al. (2020) · Traitement du signal · doi:10.18280/ts.370507 · Cited by 2
“This metric is presented by the following formula [40]: With the SSIM parameter, Eq.”
BackgroundMethodInfluentialNo-Reference Stereoscopic Image Quality Assessment Considering Multi-loss Constraints
Han, Y., Li, S., Yue, G., et al. (2020) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip49819.2020.9301832 · Cited by 2
Explainers in the Wild: Making Surrogate Explainers Robust to Distortions Through Perception
Hepburn, A., & Santos‐Rodríguez, R. (2021) · Proceedings - International Conference on Image Processing · doi:10.1109/icip42928.2021.9506711 · Cited by 5
“The three distances used were: cosine similarity between the binary interpretable representations x′ and z′, MS-SSIM and NLPD both in the image domain x and z.”
BackgroundInfluentialAre conditional GANs explicitly conditional?
Boulahbal, H. E., Voicila, A., & Comport, A. I. (2021) · British Machine Vision Conference (BMVC) · doi:10.5244/c.35.354 · Cited by 5
“Only a few measures have been designed to explicitly evaluate this issue [39, 51, 3].”
BackgroundEvaluating Foveated Video Quality Using Entropic Differencing
Jin, Y., Patney, A., & Bovik, A. C. (2021) · Picture Coding Symposium · doi:10.1109/pcs50896.2021.9477465 · Cited by 2
“We compared FED with several leading existing IQA algorithms on both the 2D and 3D LIVE-FBT-FCVR databases: PSNR [20], SSIM [20], MS-SSIM [21], VIF [13], S-RRED [16], SpeedIQA [17], FSIM [22], FWQI [9], and FA-SSIM [11].”
MethodSelf-citationEfficient Immersive Video Compression using Screen Content Coding
Samelak, J., Dziembowski, A., Mieloch, D., et al. (2021) · Computer Science Research Notes · doi:10.24132/csrn.2021.3002.22 · Cited by 2
Band Selection for Dehazing Algorithms Applied to Hyperspectral Images in the Visible Range
Fernández-Carvelo, S., Martínez‐Domingo, M. Á., Valero, E. M., Romero, J., Nieves, J. L., & Hernández‐Andrés, J. (2021) · Sensors · doi:10.3390/s21175935 · Cited by 2
“The combined metric is defined as shown in Equation (5): CM-DIE = W1 |1−VIF|+ W2(MSiCID) + W3(1−MSSSIM), (5) where, W1 = 0.146, W2 = 0.309 and W3 = 0.545 are relative weights, VIF represents the VIF value, MSiCID the MS-iCID value, MSSIM the MS-SSIM value for the image pair, and the straight bars represent the absolut…”
BackgroundMethodInfluentialBlind Image Restoration with CNN Denoiser Prior
Repala, H. K., Christopher, A., & Sudeep, P. V. (2021) · Lecture notes in networks and systems · doi:10.1007/978-981-16-5348-3_58 · Cited by 2
Mastering Quantization is Key for Video Compression
Ropert, M., Tanou, J. L., & Blestel, M. (2022) · SMPTE Motion Imaging Journal · doi:10.5594/jmi.2022.3167782 · Cited by 2
DefOff: Defensive/Offensive System based on Hiding Technologies
Almomani, I., Ahmed, M., & El‐Shafai, W. (2022) · International Conference of Smart Systems and Emerging Technologies (SMARTTECH) · doi:10.1109/smarttech54121.2022.00054 · Cited by 2
Multi-scale and Scalable Video Quality Assessment
Seshadrinathan, K., & Bovik, A. C. (2008) · International Conference on Consumer Electronics · doi:10.1109/icce.2008.4588117 · Cited by 2
Self-citationEffective visual masking techniques in JPEG200
Richter, T. (2008) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2008.4712395 · Cited by 2
“One of them is the Structural Similarity Index [15] (SSIM) and its multi-scale variant [16] (M-SSIM) by Simoncelli et al. The new feature of this metric is that unlike traditional approaches like VDP [6], it does not try to address the properties of the human visual system directly.”
MethodA parametric method for edge detection based on recursive mean-separate image decomposition
Nercessian, S. C., Panetta, K. A., & Agaian, S. С. (2008) · International Conference on Machine Learning and Cybernetics · doi:10.1109/icmlc.2008.4621046 · Cited by 2
“The edge map which yields the maximum SSIM can be taken to be best edge map by the objective evaluation method.”
MethodAnalysis of SSIM performance for Digital Cinema applications
Rahayu, F. N., Reiter, U., Nielsen, M. T. M., et al. (2009) · International Workshop on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex.2009.5246981 · Cited by 2
The Daala Directional Deringing FilterPreprint
Valin, J.-M. (2016) · arXiv (Cornell University) · Cited by 4
Range Image Quality Assessment by Structural Similarity
Malpica, W., & Bovik, A. C. (2008) · Encyclopedia of Multimedia · doi:10.1007/978-0-387-78414-4_80 · Cited by 2
Self-citationA Note on Efficiency of Downsampling and Color Transformation in Image Quality AssessmentPreprint
Nafchi, H. Z., & Cheriet, M. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1606.06152 · Cited by 2
“If a metric only uses a luminance channel [3], [4], [17], the old strategy should be used.”
MethodImpact of buffering on quality of experience
Hohlfeld, O. (2014) · DepositOnce · doi:10.14279/depositonce-4011 · Cited by 3
“Subjective tests with JPEG images show slightly higher correlation with human quality perception at the cost of higher computational complexity for MS-SSIM [259].”
MethodOn Nearest Neighbors in Non Local Means DenoisingPreprint
Frosio, I., & Kautz, J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1711.07568 · Cited by 2
“tch size, the number of negihbors N n, and the filtering parameter hare the optimal ones in [2]. We resort to a large set of image quality metrics to evaluate the filtered images: PSNR, SSIM[16], MSSSIM[17], GMSD[19], FSIM and FSIM C[21]. PSNR is proportional to the average squared error in the image; SSIM and MSSSIM ta…”
BackgroundInfluentialPredicting Encoded Picture Quality in Two Steps is a Better WayPreprint
Yu, X., Bampis, C. G., Gupta, P., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1801.02016 · Cited by 2
“Full reference models such as SSIM [3], MS-SSIM [4], FSIM [5] and VSI [6] deliver excellent quality prediction performance on subjective quality databases such as LIVE IQA [7], TID2008 [8] and CSIQ [9].”
BackgroundMethodSelf-citationEstimating Depth-Salient Edges And its Application To Stereoscopic Image Quality AssessmentPreprint
Khan, S., & Channappayya, S. S. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1802.03883 · Cited by 2
“The quality score is computed by applying the MS-SSIM [9] index to the reference and test cyclopean images.”
MethodNeural Multi-scale Image CompressionPreprint
Nakanishi, K. M., Maeda, S., Miyato, T., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.06386 · Cited by 2
Full-reference image quality assessment by combining global and local distortion measuresPreprint
Saha, A., & Wu, Q. M. J. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1412.5488 · Cited by 2
Adversarial Learning of Semantic Relevance in Text to Image SynthesisPreprint
Cha, M., Gwon, Y., & Kung, H. T. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.05083 · Cited by 2
Deep Reflectance MapsPreprint
Rematas, K., Ritschel, T., Fritz, M., Gavves, E., & Tuytelaars, T. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1511.04384 · Cited by 4
“The first is plain L2 error between all defined pixels of the reflectance map in RGB and the second the SSIM structural difference [36].”
Auto-Embedding Generative Adversarial Networks for High Resolution Image SynthesisPreprint
Guo, Y. F., Chen, Q., Chen, J., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.11250 · Cited by 2
“For quantitative evaluation, we use Frechet Inception Distance (FID) [36], Inception Score [7] and MS-SSIM [22], [47] to evaluate the generated samples.”
MethodDSIC: Deep Stereo Image CompressionPreprint
Liu, J. T., Wang, S., & Urtasun, R. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.03631 · Cited by 2
“Metrics: We report results on both peak signal-to-noise ratio (PSNR) and multi-scale structural similarity (MSSSIM) [46] as a function of bitrate.”
BackgroundMethodInfluentialQuality Assessment of Stereoscopic 360-degree Images from Multi-viewportsPreprint
Xu, J., Luo, Z., Zhou, W., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.10079 · Cited by 2
“For 2D metrics such as PSNR, SSIM [11], MS-SSIM [12], FSIM [27], S-PSNR [19], WS-PSNR [20], CPP-PSNR [21], the qualities of left and right view images are averaged to obtain the final perceptual score.”
Quality Assessment of DIBR-synthesized views: An OverviewPreprint
Tian, S., Zhang, L., Zou, W., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1911.07036 · Cited by 2
“Most of the 2D objective quality metrics [4, 5, 6, 7, 8] which focus on the traditional distortions will fail to evaluate the quality of DIBR-synthesized views.”
BackgroundA deep learning approach to image quality assessment
Feng, Y. (2019) · DR-NTU (Nanyang Technological University) · doi:10.32657/10356/85493 · Cited by 2
Robust Invisible Hyperlinks in Physical Photographs Based on 3D Rendering AttacksPreprint
Jia, J., Gao, Z., Chen, K., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.01224 · Cited by 2
“To solve this problem, several methods to measure image structural or perceptual similarity were proposed, such as SSIM [32], MSSIM [33], FSIM [39].”
MethodLearned Variable-Rate Image Compression with Residual Divisive NormalizationPreprint
Akbari, M. A., Liang, J., Han, J., & Tu, C. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.05688 · Cited by 3
“We use both PSNR and MS-SSIM [9] as the evaluation metrics.”
Conditional Generative ConvNets for Exemplar-based Texture SynthesisPreprint
Wang, Z., Li, M., & Xia, G.-S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.07971 · Cited by 2
“For quantitative evaluation, we measure multi-scale structural similarity [47] (MS-SSIM) between the synthesized tex-”
MethodSelf-citationIntelligent Home 3D: Automatic 3D-House Design from Linguistic Descriptions OnlyPreprint
Chen, Q., Wu, Q., Tang, R., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.00397 · Cited by 2
“The generated interior textures are evaluated with popular image generation metrics such as Fr´echet Inception Distance (FID) [12] and Multi-scale Structural Similarity (MS-SSIM) [37].”
MethodInfluentialDasgupta, T., Das, N., & Nasipuri, M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.06872 · Cited by 2
“Thus, as another structural measure of dewarping accuracy, multi-scale structural similarity (MS-SSIM) [44] has been calculated for the dewarped images.”
MethodInfluentialXu, J., Zhou, W., Chen, Z., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.01738 · Cited by 2
End-to-End Learning for Video Frame Compression with Self-AttentionPreprint
Zou, N., Zhang, H., Cricri, F., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.09226 · Cited by 2
“We train our model by using the following objective: (11) where D 1 is the negative multi-scale structural similarity (MS-SSIM) [14], D 2 is the mean-squared error (MSE), R is the rate-loss provided by the probability model, M is the constraint on the importance map defined in Eq.”
InfluentialFrequency-Weighted Robust Tensor Principal Component AnalysisPreprint
Wang, S., Liu, Y., Feng, L., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.10068 · Cited by 2
“Defined in [57], [58].”
Cai, J., Meng, Z., & Ho, C. M. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.13674 · Cited by 2
“Besides, the multi-scale structural similarity (MS-SSIM) [34] supplies more flexibility than traditional SSIM in incorporating the variations of viewing conditions.”
MetaInv-Net: Meta Inversion Network for Sparse View CT Image ReconstructionPreprint
Zhang, H., Liu, B., Yu, H., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.00171 · Cited by 2
“reas-Test set. It has a different distribution to the training set. Quantitative results are measured by the multi-scale SSIM index (with level=5, Gaussian kernel size=11, and standard deviation=1.5) [91], [92] and PSNR. D. Quantitative Measure at Different Scanning Settings In this experiment, we evaluate the perform…”
MethodInfluentialA Gated and Bifurcated Stacked U-Net Module for Document Image DewarpingPreprint
Bandyopadhyay, H., Dasgupta, T., Das, N., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.09824 · Cited by 2
“1333 respectively following the original implementation [30].”
Fully Trainable and Interpretable Non-Local Sparse Models for Image RestorationPreprint
Lecouat, B., Ponce, J., & Mairal, J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.02456 · Cited by 2
Subjective Quality Study and Database of Compressed Point Clouds with 6DoF Head-mounted DisplayPreprint
Wu, X., Zhang, Y., Fan, C., et al. (2020) · arXiv (Cornell University) · Cited by 2
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.