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 5,101–5,200 of 6,047
HVS Based Full Reference Video Quality Assessment Based on Optical Flow
Gujjunoori, S., & Oruganti, M. (2018) · International Conference on Pattern Recognition and Artificial Intelligence (PRAI) · doi:10.1145/3243250.3243271
“The spatial features in Figure 3 are computed using MS-SSIM index [11] as below in equation (1).”
MethodCombination of Image Quality Scores Based on Particle Swarm Optimization
Khaing, Y., Sugiura, Y., & Shimamura, T. (2018) · IEEE Region 10 Conference (TENCON) · doi:10.1109/tencon.2018.8650475
“The calculation formula is defined as Mcom = k1M ϕ ma + k2M υ ms +M ρ fs + C (1) where Mma, Mms and Mfs are the MAD, MS-SSIM and FSIM scores, respectively, and k1, k2, ϕ, υ, ρ and C { c |c1, c2, c3 } are adjustable parameters.”
BackgroundMethodInfluentialTensor Completion using Balanced Unfolding of Low-Rank Tensor RingPreprint
Huang, H., Liu, Y., & Zhu, C. (2019) · arXiv (Cornell University)
A Multi-metrics Integrated FR IQA Method Based on Machine Learning
Pan, Z., Xiao, J., Sun, L., & Chen, Y. (2019) · Lecture notes in electrical engineering · doi:10.1007/978-981-13-3663-8_28
Motion Trajectory based Spatial-Temporal Degradation Measurement for Video Quality Assessment
Wu, J., Liu, Y., & Shi, G. (2018) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip.2018.8698612
“, MOVIE [13], VQM VFD [14], stRRED [15], Vis3 [3], FePVQ [16], FLOSIM [4], SpEED [17] and Peng[5]), as well as two FR IQA models (GMSD [12] and MS-SSIM [18]) are employed for comparison.”
MethodCompressed Image Quality Assessment Based on Saak FeaturesPreprint
Zhang, X., Kwong, S., & Kuo, C. ‐. J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.02001
“In the past decades, to obtain more consistent quality measures with human visual perception, numerous image quality assessment (IQA) methods [4, 5, 6, 7, 8, 9, 10] and datasets [11, 12, 13] have been proposed for different distortion types.”
BackgroundMethodInfluentialQuality Assessment of Virtual Reality Videos
Wu, P. J., An, P., & Ma, J. (2019) · Communications in computer and information science · doi:10.1007/978-981-13-8138-6_23
Comparing CNNs and JPEG for Real-Time Multi-view Streaming in Tele-Immersive Scenarios
Konstantoudakis, K., Christakis, E., Drakoulis, P., et al. (2018) · International Conference on Signal-Image Technology & Internet-Based Systems (SITIS) · doi:10.1109/sitis.2018.00022
“Table I shows the correspondence between equivalent, in terms of MS-SSIM, modes of the three codecs.”
MethodInfluentialEfficient Data usage and Savingwhile Streaming Video
Prakash, O., Chakraborty, P., & Aggarwal, S. (2019) · International Conference on Communication Systems & Networks (COMSNETS) · doi:10.1109/comsnets.2019.8711395
Análise Exploratória de Dados de Imagens Digitais Noturnas
Steffens, C. R., Drews, P., Botelho, S. S. D. C., et al. (2019) · Scientia Plena · doi:10.14808/sci.plena.2019.049912
Trends and New Prospects of Video Coding Researches towards the IoT Era
Yashima, Y., & Takamura, S. (2019) · Denshi Jouhou Tsuushin Gakkai Tsuushin Sosaieti magajin/B-plus/Denshi Jouhou Tsuushin Gakkai Tsuushin Sosaeti magajin · doi:10.1587/bplus.13.43
Savioli, N. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.04681
“The evaluation metrics used are the PSNR and MS-SSIM [13] for both valid and test.”
MethodInfluentialDisentangling Latent Space for VAE by Label Relevant/Irrelevant DimensionsPreprint
Zheng, Z., & Sun, L. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.09502
“We adopt the metric proposed in [5], which measures the average negative MSSSIM [40] between all pairs in the generated image set X .”
MethodPrediction of the Influence of Navigation Scan-path on Perceived Quality of Free-Viewpoint VideosPreprint
Ling, S., Gutiérrez, J., Gu, K., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.04409
Patch redundancy in images: a statistical testing framework and some applicationsPreprint
Valentin, D. B., Desolneux, A., Galerne, B., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.06428
“In this paper we focus on the square Euclidean distance but other choices could be considered [65, 64, 15, 17].”
BackgroundLow Photon Budget Phase Retrieval with Perceptual Loss Trained Deep Neural NetworksPreprint
Deng, M., Goy, A., Li, S., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.05687
“It is commonly acknowledged that metrics such as Peak Signal-to-Noise Ratio (PSNR) [3,22,29], Structural Similarity index (SSIM) [23,24] do not reliably correlate with human visual perceptions, which is the reason why we refrained from drawing conclusions based on these metrics.”
BackgroundSemantically Interpretable and Controllable Filter SetsPreprint
Prabhushankar, M., Kwon, G., Temel, D., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.06334
“Previous studies [25, 26, 27, 28, 29] in this field, including our own [11, 12] concentrated on and provided results on all the distortion levels together.”
BackgroundAutomatic ISP image quality tuning using non-linear optimizationPreprint
Nishimura, J., Gerasimow, T., Rao, S., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.09023
“all zipper artifact and false color at high frequency (as indicated in red arrow on OV16860-ISO50) is properly addressed. Table 1 shows Mean Absolute Difference (MAD) in 8bit scale, SSIM, and MS-SSIM [10] between reference and different tunings of each key ISP block. ‘Auto-tuned’ provides significant IQ improvement ov…”
BackgroundTowards improved lossy image compression: Human image reconstruction with public-domain imagesPreprint
Bhown, A., Mukherjee, S., Yang, S., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.11137 · Cited by 1
“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.”
BackgroundPerceptual stereoscopic images quality assessment method based on visual adaptable characteristics
Zheng, Z., Liu, Y., & Liu, Y. (2019) · Optical Engineering · doi:10.1117/1.oe.58.7.073101
Thai, B. C., & Mokraoui, A. (2019) · doi:10.1109/ict.2019.8798770
“The map of SF1 is averaged on each scale and the final SF1 index is obtained in the same way as in the case of MS-SSIM.”
MethodViewpoint-agnostic Image Rendering
Aizawa, H., Kataoka, H., Satoh, Y., & Kato, K. (2021) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv48630.2021.00385 · Cited by 1
“Table 1 shows exploration study with SSIM/MSSSIM and LPIPS as evaluation metrics.”
MethodDefending Against Misclassification Attacks in Transfer Learning.Preprint
Wu, B. Y., Yang, X., Wang, S., Yuan, X., Wang, C., & Rudolph, C. (2019) · arXiv (Cornell University) · Cited by 1
Towards Object Shape Translation Through Unsupervised Generative Deep Models
Bollens, L., Tuytelaars, T., & Jose Oramas, M. (2019) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2019.8803457
“difference of Structural Similarity loss [10].”
InfluentialDistorted Representation Space Characterization Through Backpropagated GradientsPreprint
Kwon, G., Prabhushankar, M., Temel, D., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.09998
Kim, J., Jeong, Y.-J., Chung, H.-J., Lee, C., & Oh, C. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.10995 · Cited by 2
Geometry-Based Skin Colour Estimation for Bare Torso Surface Reconstruction
Monteiro, J., Zolfagharnasab, H., & Oliveira, H. P. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-30645-8_54
Scalable Double Regularization for 3D Nano-CT ReconstructionPreprint
Tang, W., & Li, M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.02256
“noise ratio (SNR) and the structural similarity (SSIM) index [45] between the reconstructed structure f̂ and the true structure f :”
BackgroundLearning Patterns of Latent Residual for Improving Video Compression
Chen, Y., Chang, K.-J., Tsai, Y., et al. (2019)
Empirical evaluation of full-reference image quality metrics on MDID databasePreprint
Varga, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.01050 · Cited by 2
“…0.7243 0.5327 MCSD [31] 2016 0.8386 0.8457 0.6622 MDSI (’mult’) [32] 2016 0.8130 0.8278 0.6441 MDSI (’sum’) [32] 2016 0.8249 0.8363 0.6527 MS-SSIM [33] 2003 0.7884 0.8292 0.6360 MS-UNIQUE [34] 2017 0.8604 0.8712 0.6893 NQM [35] 2000 0.6177 0.5869 0.4143 PerSIM [36] 2015 0.8282 0.8196 0.6296…”
InfluentialA Kernel Estimation For Blurry Image Using Maximum a Posteriori Framework with Elastic-Net Prior
Yu, X., Zhou, Y. L., Chen, D., & Xie, W. C. (2019) · Chinese Control Conference (CCC) · doi:10.23919/chicc.2019.8866046
Visual Expression of Volumetric Smoke Animation by Dynamic Luminance Information
Chiba, Y., & Koike, T. (2019) · International Conference on Intelligent Systems and Image Processing (ICISIP) · doi:10.12792/icisip2019.023
Performance Analysis of Quality Measurement for Biomedical Images
Rajyaguru, V. C., & Thanki, R. M. (2019) · IOP Conference Series Materials Science and Engineering · doi:10.1088/1757-899x/561/1/012101
“The standards like mean square error (MSE) [9], root MSE (RMSE), peak signal to noise ratio (PSNR) and its various types, structural similarity index measure (SSIM) [1] [3] [10], multi-scale SSIM (MS-SSIM) [3] [11] and feature similarity index measure (FSIM) [12] are given in next subsections.”
MethodProgressive Compressed Records: Taking a Byte out of Deep Learning DataPreprint
Kuchnik, M., Amvrosiadis, G., & Smith, V. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1911.00472
“Scan groups of 5 or higher have an MSSIM of 95%+, which is likely why they consistently perform well.”
MethodInfluentialSwitching GAN-based Image Filters to Improve Perception for Autonomous Driving
Masud, Z. (2019) · UWSpace (University of Waterloo)
Continuous Predictive Model for Quality of Experience in Wireless Video Streaming
Shi, W.-J., & Pan, J. (2019) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-030-36405-2_25
IEGAN: Multi-purpose Perceptual Quality Image Enhancement Using Generative Adversarial NetworkPreprint
Ghosh, S. S., Yang, H., Mukherjee, S. S., & Robertson, N. M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.09134
“However, the ability of PSNR to capture perceptually relevant differences is very limited as they are defined based on pixel-wise image differences [16, 36, 34].”
BackgroundLearning Structure-Appearance Joint Embedding for Indoor Scene Image SynthesisPreprint
Xue, Y., Zhou, Z., & Huang, X. (2019) · arXiv (Cornell University)
Deep Optimized Multiple Description Image Coding via Scalar Quantization LearningPreprint
Zhao, L., Bai, H., Wang, A., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2001.03851
“As described in [38], SSIM is a good approximation to assess image quality from perspective of human visual perception, but this method only considers single-scale image information.”
MethodInfluentialA Foreground-background Parallel Compression with Residual Encoding for Surveillance VideoPreprint
Wu, L., Huang, K., Shen, H., & Gao, L. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2001.06590
“The experimental results show that our model outperforms the widely used algorithms, such as H.264 and H.265 when measured by the Multiscale Structural Similarity Index (MS-SSIM) [7] and Peak Signal to Noise Ratio (PSNR).”
Research on Influence of Content Diversity on Full-Reference Image Quality Assessment
Zhang, H., Li, S., Xia, Z., et al. (2020) · Communications in computer and information science · doi:10.1007/978-981-15-3341-9_23
Discernible Compressed Images via Deep Perception ConsistencyPreprint
Yang, Z., Wang, Y., Xu, C., et al. (2020) · arXiv (Cornell University)
Crowd Counting via Enhanced Feature Channel Convolutional Neural Network
Bian, Y., Shen, J., Xiong, X., Li, Y., He, W., & Li, P. (2019) · IEEE International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai.2019.00118
“MS-SSIM is an image quality evaluation index proposed by [23].”
BackgroundMethodActived Edge Strength for Image Quality Assessment
Gao, M., & Zhang, X. (2019) · International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS) · doi:10.1109/ispacs48206.2019.8986261
Similarity-Aware EQUALS and IN Operator in Cassandra and Its Application in Agriculture
Ramesh, D., Kumar, A., & Kumar, A. (2019) · IEEE International Symposium on Smart Electronic Systems (iSES) · doi:10.1109/ises47678.2019.00021
Active Lighting Recurrence by Parallel Lighting Analogy for Fine-Grained Change DetectionPreprint
Zhang, Q., Feng, W., Wan, L., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.09663
“1 1 SSIM MS-SSIM 0 5.”
InfluentialAn Optimized Image Fusion Method for Fume Removal in Automated Welding Robots Field
Vaz, E. S., Weis, Á. A., Drews, P., Botelho, S. S. D. C., Steffens, C. R., & Pereira, L. (2019) · Latin American Robotics Symposium, Brazilian Symposium on Robotics and Workshop on Robotics in Education (LARS/SBR/WRE) · doi:10.1109/lars-sbr-wre48964.2019.00055
Single Image Super ResolutionUSingDeep Convolutional Neural Networks
Venkat, V. S., & Lokanath, M. (2018) · IEEE International Conference on Recent Trends in Electronics, Information & Communication Technology · doi:10.1109/rteict42901.2018.9012445
Regional bit allocation with visual attention and distortion sensitivity
Pak, M., & Bayazıt, U. (2020) · Multimedia Tools and Applications · doi:10.1007/s11042-020-08686-z
“We adapt the MS-SSIM (multi-scale structural similarity index which is defined in [54]) to the VAM model in order to assess the reconstructed images in an attention-based manner (exponential parameter k is set to 5.”
MethodScene Completeness-Aware Lidar Depth Completion for Driving ScenarioPreprint
Wu, C.-Y., & Neumann, U. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.06945
“Scale SSIM (MS-SSIM)[13] as a variant to measure structural similarity for completed depth maps.”
MethodContinuous QoE Prediction Based on WaveNetPreprint
Tan, P. X., Duc, T. N., Tran, C. M., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.09249
Mokhtar, F., & Ngadiran, R. (2020) · IOP Conference Series Materials Science and Engineering · doi:10.1088/1757-899x/767/1/012015
Co-occurrence Background Model with Superpixels for Robust Background InitializationPreprint
Zhou, W., Deng, Y., Peng, B., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.12931
“• MS-SSIM ( Multi-scale Structural Similarity Index ): estimate of the perceived visual distortion defined in [25].”
MethodImage Compression for Constrained Aerial Platforms: A Unified Framework of Laplacian and cGAN
Faheema, A. G. J., Lakshmi, A., & Priyanka, S. (2020) · Communications in computer and information science · doi:10.1007/978-981-15-4018-9_19
Low Bitrate Image Compression with Discretized Gaussian Mixture LikelihoodsPreprint
Cheng, Z., Sun, H., & Katto, J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.04318
“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 Quality Assessment for Image Superresolution Using Deep Two-Stream Convolutional NetworksPreprint
Zhou, W., Jiang, Q., Wang, Y., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.06163
“For the SR quality database [17], we compare the performance of our proposed DeepSRQ method with three classical FR-IQA metrics, namely PSNR, SSIM [33], and MS-SSIM [35].”
MethodInfluentialNovel solutions toward high accuracy automatic brain tissue classification in young childrenPreprint
Portman, N., Toussaint, P.-J., & McConnell, A. C. E. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.03261
“It has been shown that the SSIM is well-matched to the subject ratings of image databases, and therefore, it is a good approximation to the perceived image quality (Wang et al., 2003).”
BackgroundConvolutional auto-encoder based multiple description coding network
Meng, L., Li, H., Zhang, J., et al. (2020) · KSII Transactions on Internet and Information Systems · doi:10.3837/tiis.2020.04.015
DP-GAN: Dual Pathway Generative Adversarial Network for Face Recognition in Degraded Scenarios
Bhattacharjee, A., Banerjee, S., & Das, S. (2018) · Indian Conference on Computer Vision, Graphics and Image Processing · doi:10.1145/3293353.3293379 · Cited by 1
“Since the images captured by the surveillance cameras appear at different resolutions and distances, a multi-scale SSIM based technique proposed by Wang et al. [42] has been adopted for computing the MS-SSIM loss, as described later in section 4.1.4.”
BackgroundMethodInfluentialEncoding in the Dark Grand Challenge: an Overview
Anantrasirichai, N., Zhang, F., Malyugina, A., Hill, P., & Katsenou, A. (2020) · IEEE International Conference on Multimedia & Expo Workshops (ICMEW) · doi:10.1109/icmew46912.2020.9106011 · Cited by 2
Joint-SRVDNet: Joint Super Resolution and Vehicle Detection NetworkPreprint
Mostofa, M., Ferdous, S. N., Riggan, B. S., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.00983
“…results of our experiments using several objective image quality metrics such as Peak Signal-to-Noise ratio (PSNR), Multi-scale Structural Similarity (MSSIM) [57], Universal image Quality Index (UQI) [58] and Visual Information Fidelity (VIF) [59] on a validation subset of images for each dataset.”
MethodEfficient Training Procedures for Multi-Spectral Demosaicing
Shopovska, I., Jovanov, L., & Philips, W. R. (2020) · Sensors · doi:10.3390/s20102850
“Similarly to the approach in paper [9], our method calculates multi-scale SSIM (MS-SSIM) on two different scales, approximated by Gaussian windows with different standard deviations, σ1 and σ2.”
MethodInfluentialOmnidirectional Images as Moving Camera Videos.Preprint
Sui, X., Ma, K., Yao, Y., et al. (2020) · arXiv (Cornell University)
ОЦІНКА ЕФЕКТИВНОСТІ МОДЕЛЕЙ МЕРЕЖ ГЛИБИННОГО НАВЧАННЯ ПІДВИЩЕННЯ РОЗДІЛЬНОЇ ЗДАТНОСТІ ЗОБРАЖЕНЬ
Demchyshyn, A. (2019) · System technologies · doi:10.34185/1562-9945-5-124-2019-15
One Size Fits All: Can We Train One Denoiser for All Noise Levels?Preprint
Gnanasambandam, A., & Chan, S. H. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.09627 · Cited by 21
“, 2004) and MSSSIM (Wang et al., 2003) also as long as they are convex.”
BackgroundJeon, S.-Y., & Kim, D.-Y. (2020) · Journal of Electrical Engineering and Technology · doi:10.1007/s42835-020-00431-8
Local semi-supervised approach to brain tissue classification in child brain MRIPreprint
Portman, N., Toussaint, P.-J., & Evans, A. C. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.09871
“It has been shown that the SSIM is well-matched to the subject ratings of image databases, and therefore, it is a good approximation to the perceived image quality (Wang et al., 2003).”
BackgroundAnalyzing the performance of autoencoder-based objective quality metrics on audio-visual content
Martinez, H. B., Farias, M. C. Q., & Hines, A. (2020) · Electronic Imaging · doi:10.2352/issn.2470-1173.2020.9.iqsp-167
“PSNR, SSIM, and BIQI seemed to produce scores that are sparser, while DIIVINE, BRISQUE, and NAVE scores correlated better with the subjective scores.”
MethodInfluentialDuan, S., Chen, H., & Gu, J. (2020) · arXiv (Cornell University)
Objective quality assessment of a mammography accreditation phantom
Valentina Corchuelo, G. (2020)
Neural Wireframe Renderer: Learning Wireframe to Image TranslationsPreprint
Xue, Y., Zhou, Z., & Huang, X. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.03840
“More details of MS-SSIM can be found in [49].”
Event Enhanced High-Quality Image RecoveryPreprint
Wang, B., He, J., Lei, Y., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.08336
“For the metrics, we use PSNR and SSIM [34] for quantitative comparison while the visual effect of the reconstructed images for qualitative comparison.”
MethodTavousi, P., Ahmadi, B., May, N., et al. (2020) · Microscopy and Microanalysis · doi:10.1017/s1431927620019133
“Here, we assume that the restored images taken by 20 microsecond dwell time are the reference traditional methods such as “the peak signal to noise ratio (PSNR)” and “the structural similarity index (SSIM)”, and some novel methods like “the feature similarity index (FSIM)”[6], “the multi-scale structural similarity sc…”
MethodDeep Controllable Backlight DimmingPreprint
Duan, L., Marnerides, D., Chalmers, A. G., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.08352
“valuation of the metrics. None of these 105 HDR images were not used in the training of DBLD. The metrics used for comparison were the Perceptually Uniform (PU) [2] versions of PSNR, Multi-Scale SSIM [41], along with HDR-VDP-2.2 [28]. The power saving ratio (PSR) [4] corresponds to the percentage of power savings with…”
MethodInfluentialDeep Local and Global Spatiotemporal Feature Aggregation for Blind Video Quality AssessmentPreprint
Zhou, W., & Chen, Z. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.03411
“As shown in Table 1, the proposed method is compared with 14 state-of-the-art quality assessment metrics, including PSNR, SSIM [4], MS-SSIM [5], FSIM [6], MOVIE [7], ST-MAD [8], ViS3 [3], JVQ [9], VQM [10], STRRED [11], V-CORNIA [12], V-BLIINDS [14], VIIDEO [16], and DeepBVQA [17].”
MethodInfluentialImproving the Prediction Accuracy of Objective Video Quality Evaluation
Stojanović, N., Bondžulić, B. P., Pavlović, B., Novčić, M., & Bujaković, D. M. (2020) · Acta Polytechnica Hungarica · doi:10.12700/aph.17.7.2020.7.12
67‐2: Lightweight Tone‐mapped HDRNET with Exposure Stack Generation
Jo, S. Y., Ahn, N., & Kang, S. (2020) · SID Symposium Digest of Technical Papers · doi:10.1002/sdtp.14040
Spatial Association Between Images
Vallejos, R. O., Osorio, F., & Bevilacqua, M. (2020) · Springer eBooks · doi:10.1007/978-3-030-56681-4_8
Image Compression with Laplacian Guided Scale Space Inpainting
Zhang, L., Kumar, P., Sabharwal, M., Kuzma, A., & Shi, J. (2020) · Proceedings - International Conference on Image Processing · doi:10.1109/icip40778.2020.9191041 · Cited by 1
“We focus on laplacian image features because it matches human perception of natural image quality [8] and human attention is scale dependent [9].”
BackgroundA Salience & Motion State Based Quality Adjustment for 360-Degree Video Transmission
Zong, N., Lu, J., Li, Y., et al. (2020) · International Conference on Big Data and Internet of Things (BDIOT) · doi:10.1145/3421537.3421555
“Under the circumstance of VR Video transmission, supposing the benchmark video frame is Image_X as an image in the server, and the following frame compared is Image_Y, the MS SSIM value is calculated [13]: (1) With: (2) Where in (1): M, the reduction factor, meaning the times of reduction.”
MethodAdaptive Importance Channel Selection for Perceptual Image Compression
He, Y., Li, F., Bai, H., & Zhao, Y. (2020) · KSII Transactions on Internet and Information Systems · doi:10.3837/tiis.2020.09.014
“In this paper, image distortion is evaluated by Multi-Scale Structure Similarity (MS-SSIM) [28] and Peak Signal-to-Noise Ratio (PSNR), while compression ratio is evaluated by bits per pixel (bpp), which calculates as the number of bits used to code the original image divided by the number of pixels.”
MethodDeep Autofocus for Synthetic Aperture SonarPreprint
Gerg, I. D., & Monga, V. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.15687
“The measures evaluated are peak signal-to-noise ratio (PSNR) [22], structural similarity metric (SSIM) [22], multiscale SSIM (MS-SSIM) [23], visual information fidelity (VIF) [24], and spatial correlation coefficient (SCC) [25].”
MethodR-MSSIM: image quality assessment while performing object detection
Lin, Q., Haichao, H., Wang, Z., et al. (2020) · International Conference on Image, Video Processing and Artificial Intelligence (IVPAI) · doi:10.1117/12.2579495
“9 In the decision-making of the global complex image quality, we need to train and make decisions on the weight of the selected object image block relative to the overall quality.”
BackgroundMethodInfluentialSelf-citationGuamán, A. V., Pico, P. A., Nunez, N. E., et al. (2020) · IEEE ANDESCON · doi:10.1109/andescon50619.2020.9272091
“The Structural Similarity Index (SSIM) was used to predict where are the most important or more probable areas with substance [17].”
MethodObject-centered image stitchingPreprint
Herrmann, C., Wang, C., Bowen, R. S., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.11789
“We provide some examples and their evaluation values (maximum MS-SSIM with direct object comparison) in the captions of the figures above.”
MethodInfluentialRobust image stitching with multiple registrationsPreprint
Herrmann, C., Wang, C., Bowen, R. S., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.11784
“We then calculate MS-SSIM [24] or PSNR.”
Quan, Q., Wang, Q., Li, L., Du, Y., & Zhou, S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.09491
“of hard examples and real photos shown in Figure8. 5.1. Setup Evaluation Metrics. To evaluate the similarity between the original and the dewarped CT film image, we choose PSNR, SSIM [28], and MS-SSIM [29] as our evaluation metrics. Peak signal-to-noise ratio (PSNR), is the ratio between the maximum possible power of a…”
BackgroundInfluentialJalilian, E., Hofbauer, H., & Uhl, A. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-68821-9_40
“The luminescence comparison is computed only at scale M , and the final MS-SSIM evaluation is obtained by combining the measurements at different scales [26].”
MethodA Quantization Matrix Design using Total Variation for Sparse Representation-based Image Coding
Wang, J., & Yashima, Y. (2020) · IEEE Global Conference on Consumer Electronics (GCCE) · doi:10.1109/gcce50665.2020.9291947
Unsupervised Image Classification System for Mobile Devices Based on Incremental Isomap
Tiance, W. U., Tong, X., & Tian, X. (2020) · International Conference on Wireless Communications and Signal Processing (WCSP) · doi:10.1109/wcsp49889.2020.9299719
“Here, Structural Similarity (SSIM) [21] is used to describe the quality of the compressed images.”
MethodChen, C., & Mou, X. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.11525
Learned Multi-Resolution Variable-Rate Image Compression with Octave-based Residual BlocksPreprint
Akbari, M. A., Liang, J., Han, J., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.15463
“It outperforms the BPG codec (RGB4:4:4) in both PSNR and MS-SSIM [15].”
MethodInfluentialPerceptual Quality Assessment of Omnidirectional Images as Moving Camera VideosPreprint
Sui, X., Ma, K., Yao, Y., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.10547
“Over the years, SSIM and its multi-scale extension [34] have been regarded as standard “perceptual” metrics to guide the optimization of computational methods for image denoising [35], image compression [36], image synthesis [37], and video coding [38].”
Perceptual Image Quality Assessment Technique for Color Images based on HSI Colour Space
Yalman, Y. (2020) · Indonesian Journal of Electrical Engineering and Informatics (IJEEI) · doi:10.11591/ijeei.v8i4.2180 · Cited by 2
“On the other hand, some other algorithms are used for measuring images, such as PSNR-HVS-M [10], SSIM (Structural SIMilarity) [11], MS-SSIM (MultiScale Structural SIMilarity) [12], UQI (Universal Quality Index) [13], VIF (Visual Information Fidelity) [14] and so on.”
MethodInfluentialStructure-preserving dynamic texture generation algorithm
Wu, L.-C., Dong-yi, Y., & Chen, Z. (2021) · Neural Computing and Applications · doi:10.1007/s00521-020-05583-2
“Quantitative evaluation for texture generation was a particularly important task, so we used a multi-scale structural similarity index [22] (MS-SSIM for short) to evaluate the image quality of the generated results.”
MethodAdaptively Truncating Gradient for Image Quality Assessment
Gao, M., Dang, H., & Zhang, X. (2020) · International Journal of Advanced Network Monitoring and Controls · doi:10.21307/ijanmc-2020-034
Sasaki, H., & Takenouchi, T. (2021) · arXiv (Cornell University)
Quality Assessment of Super-Resolved Omnidirectional Image Quality Using Tangential Views
Ozcinar, C., & Rana, A. (2021) · Electronic Imaging · doi:10.2352/issn.2470-1173.2021.9.iqsp-295
“• Structural similarity measure: Structural similarity is used for measuring the similarity between the distorted and the reference image [28, 29].”
MethodInfluentialBoosting Image and Video Compression via Learning Latent Residual Patterns
Chen, Y., Chang, K.-J., Tsai, Y., et al. (2020) · British Machine Vision Conference (BMVC) · doi:10.5244/c.34.48
“The PSNR and SSIM [27] metrics are adopted for evaluation.”
Method
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.