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 4,501–4,600 of 6,047
RAILS: Radar Range-Azimuth Map Estimation from Image, LiDAR and Semantic Descriptions
Rangaraj, P. A., Alkanat, T., & Pandharipande, A. (2025) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv64158.2025.11097772 · Cited by 1
“…loss are respectively given by: where the hyper-parameter α balances the two losses, influencing high frequency region and color/luminance preservation, M is the number of scales, with luminance l M computed at the first scale, and contrast c j and structure s j are computed at scale j [24].”
Šimka, M., Polák, L., Zizien, A., et al. (2025) · IEEE Access · doi:10.1109/access.2025.3597214 · Cited by 1
“…namely: Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM) [20], Multi-Scale Structural Similarity (MS-SSIM) [21], Feature Similarity chrominance (FSIMc) [22], Gradient Magnitude Similarity Deviation (GMSD) [23], Visual Information Fidelity in the pixel…”
Multi-Scale Contextual Coding for Human-Machine Vision of Volumetric Medical Images
Chen, J., Chen, W., Xing, Q., & Yu, F. N. (2025) · IEEE Access · doi:10.1109/access.2025.3597008 · Cited by 2
“Here we introduce another evaluation metric for lossy compression methods: the Multi-Scale Structural Similarity Index Measure (MS-SSIM) [72].”
Diffusion Model-Based Augmentation Using Asymmetric Attention Mechanisms for Cardiac MRI Images
Özdemir, M., & Eroğul, O. (2025) · Diagnostics · doi:10.3390/diagnostics15161985 · Cited by 4
StyleDemorpher: high-quality face demorphing via StyleGAN2’s latent space
Ismayilov, R., Spreeuwers, L., & Batskos, I. (2025) · Machine Vision and Applications · doi:10.1007/s00138-025-01735-3 · Cited by 5
“Additionally, the MS-SSIM [36] loss function was integrated to enhance identity preservation, as elaborated in Sect.”
Enhancement of Underwater Images Using Lattice Transformer at Multiple Scales and Layers
Hsu, W., Chen, I.-T., & Hsieh, Y.-M. (2025) · IEEE Journal of Oceanic Engineering · doi:10.1109/joe.2025.3583856 · Cited by 1
“MS-SSIM [30] is an image quality paradigm based on SSIM that offers more flexibility than single-scale SSIM [31] in accommodating variations in image resolution and observation conditions.”
Feature Quality Assessment: A Database and A Lightweight Objective Method
Wang, S., An, P., Yang, C., Li, G., Huang, X., & Wang, S. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2025.3599090 · Cited by 1
“Earlytraditionalworksfocusonextractingmanualfeatures from images, including SSIM [25], MS-SSIM [26], FSIM [27], PSNR [28], and VIF [29].”
Deep Learning for Position Angle Quantification Applied to Interstellar Filaments
Umetaliev, T., Alina, D., & Salmenova, A. (2025) · The Astronomical Journal · doi:10.3847/1538-3881/adf858 · Cited by 1
BPConvNet: A deep learning based ρ-Filtered layergram reconstruction method for computed tomography
Bauer, P., & Frikel, J. (2025) · AIP conference proceedings · doi:10.1063/5.0286063 · Cited by 1
Multi-View Video Semantic Communication
Ren, Y., Teng, L., Bao, Z., Fan, S., Chen, D., Xu, X., Li, L., & Li, Z. (2025) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/icccworkshops67136.2025.11148118 · Cited by 2
“The MVSC is trained based on the Mean Squared Error (MSE), and the performance is tested using PSNR and MS-SSIM [17], which are perceptual quality metrics.”
MFSR-GAN: Multi-Frame Super-Resolution with Handheld Motion Modeling
Khan, F., Ebenezer, J. P., Sheikh, H. R., & Lee, S.-J. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00084 · Cited by 2
“We train our generator model with weighted L1 loss, MS-SSIM loss [28], and relativistic generator loss (Section 4).”
TEFormer: Thermal Infrared Image Enhancement by Preserving Spatial Consistency and Details
Wu, Z., Li, Y., Zhang, R., Cao, S.-Y., Ying, J., Zhang, X., Bai, X., Chen, S., Yang, B., & Shen, H. (2025) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2025.3611503 · Cited by 3
“Moreover, we compare the multiscale SSIM (MS-SSIM) [55] to evaluate the enhanced multiscale structure.”
Ganesan, R., Amirrajab, S., Scannell, C. M., Gietema, H. A., Quicken, S., Mattina, A. A. L., Boymans, T. A. E. J., van de Vosse, F. N., & Huberts, W. (2025) · Computer Methods in Biomechanics and Biomedical Engineering Imaging & Visualization · doi:10.1080/21681163.2025.2563726 · Cited by 1
Tian, Y., Liu, M., & Su, D. (2025) · Pattern Recognition Letters · doi:10.1016/j.patrec.2025.10.002 · Cited by 1
A Survey on Compression and Quality Assessment Techniques for 3D Gaussian Splatting
Wu, X., Liu, X., Wang, M., Wang, S., & Kwong, S. (2025) · International Symposium on Machine Learning and Media Computing (MLMC) · doi:10.1109/mlmc65154.2025.11189884 · Cited by 1
Hierarchical Semantic Compression for Consistent Image Semantic Restoration
Li, S., Zhang, Z., Xu, M., Jiang, L. W., Liu, Y., & Zhu, C. (2025) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3618379 · Cited by 6
“The above meth-ods are optimized mostly against the objective quality, e.g., minimizing the MSE or multi-scale structure similarity index measure (MS-SSIM) [35], [36] of reconstructed images, which now have outperformed the state-of-the-art standard VVC codec regarding pixel-wise accuracy [25].”
Barba, T., Bagley, B. A., Steyaert, S., Carrillo‐Pérez, F., Sadée, C., Iv, M., & Gevaert, O. (2025) · GigaScience · doi:10.1093/gigascience/giaf116 · Cited by 5
“…a loss function relying on the structural similarity index measure (SSIM) between the input (x) and the output (ˆ y) produced by the autoencoders [ 28 ], defined as follows (1): with ( μ x, σ x 2 ) and ( μ ˆ y, σ ˆ y 2 ) the respective pixel sample mean and variance of the input (x) and output (ˆ…”
DPCD: A Quality Assessment Database for Dynamic Point Clouds
Liu, Y., Zhang, Y., Yang, Q., et al. (2025) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme59968.2025.11210051 · Cited by 1
Generative Image Coding with Diffusion Prior
Chang, J. (2025) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme59968.2025.11210192 · Cited by 2
“Using MS-SSIM [20] to measure the reconstruction distortion L sdist , the pretext optimization objective is formulated as follows: During training with Eq.”
Trioux, A., Zhang, W., Valenzise, G., & Yang, F. (2025) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme59968.2025.11209177 · Cited by 4
Deep Generative Network for Cellular-Resolution Optical Coherence Tomography Image Denoising
Liu, C.-H., Lee, Y.-W., Chen, Y.-S., Chen, Y.-C., & Huang, S.-L. L. (2025) · IEEE Journal of Selected Topics in Quantum Electronics · doi:10.1109/jstqe.2025.3627951 · Cited by 1
Fine-Grained HDR Image Quality Assessment From Noticeably Distorted to Very High Fidelity
Jenadeleh, M., Sneyers, J., Lazzarotto, D. N., et al. (2025) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex65720.2025.11219975 · Cited by 1
“The dataset includes 100 [30] 0.674 -0.672 0.720 -0.713 0.876 -0.889 SSIM after TM 0.544 -0.533 0.635 -0.638 0.770 -0.767 MS-SSIM (PQ) [31] 0.757 -0.731 0.764 -0.749 0.918 -0.907 MS-SSIM after TM 0.679 -0.669 0.709 -0.707 0.905 -0.907 PSNR-HVS (PQ) [32] 0.520 -0.526 0.771 -0.798 0.635 -0.639…”
InfluentialDescriptor: Benchmark Dataset for Animated Multimedia Quality Assessment (BeDAMQA)
Ganavdiya, S., Appina, B., & Pachori, R. B. (2025) · IEEE data descriptions. · doi:10.1109/ieeedata.2025.3628283 · Cited by 1
“2) Multiscale SSIM (MS-SSIM) [19] performs the computation of SSIM scores at multiple resolutions, and parametric pooling was performed to estimate the overall quality score of a frame.”
Chen, B., Shang, Z., Chung, J., Lerner, D., Robitza, W., Rao, R. R. R., Raake, A., & Bovik, A. C. (2025) · IEEE Access · doi:10.1109/access.2025.3631409 · Cited by 4
“Video quality assessment (VQA) models like SSIM [33], MS-SSIM [34], ST-RRED [35], and VMAF [36], BRISQUE [37], and NIQE [38] focus on pixel-level distortions but do not account for rebuffering or varying bandwidth dynamics.”
Self-citationAdaptive U-Shaped Split Federated Learning for Image Coding at Resource-Constrained UAV Network
Sun, Q., Guo, C., Yang, Y., Liu, C., & Hu, L. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3631437 · Cited by 2
“We evaluate model performance using both the widely adopted pixel-level metric PSNR and the perceptual quality metric, the MS-SSIM [44].”
Gapyak, V., März, T., & Weinmann, A. (2025) · Physics in Medicine and Biology · doi:10.1088/1361-6560/ae205c · Cited by 4
“We remark that the DF-FoV is outlined by the scanning trajectory and contains the scanning points, whereas the FoV contains a boundary area around the DF-FoV without data points.”
Pseudocylindrical Convolutions for Learned Omnidirectional Image Compression
Li, M., Ma, K.-D., Li, J.-X., & Zhang, D. (2021) · IEEE transactions on circuits and systems for video technology (Print) · doi:10.1109/tcsvt.2025.3638018 · Cited by 14
“For distortion evaluation, conventional metrics, including mean squared error (MSE), peak signal-to-noise ratio (PSNR), structural similarity (SSIM) [45], and multi-scale SSIM (MS-SSIM) [46], were employed to compute the “perceptual” distance between the original and compressed images.”
Bridging the Gap Between Saliency Prediction and Image Quality Assessment
Alexey, K., Moskalenko, A., & Vatolin, D. S. (2025) · European Signal Processing Conference (EUSIPCO) · doi:10.23919/eusipco63237.2025.11226437 · Cited by 5
Zero-3DCE: A Low-Light Video Enhancement for More Robust Computer Vision Tasks
Tatana, M. M., Maswanganyi, R. C., & Khumalo, P. (2025) · Algorithms · doi:10.3390/a18120775 · Cited by 1
Holla, V., Lane, B. M., Redford, J., Kopp, P., & Kollmannsberger, S. (2025) · Integrating materials and manufacturing innovation · doi:10.1007/s40192-025-00430-5 · Cited by 2
“Inspired by multi-scale SSIM (known as MS-SSIM, see [32]), this approach ensures that different scales influence the final measure in a controlled manner, which is particularly relevant for thermographic measurement in PBF-LB/M due to its multi-scale nature.”
MarsQE: Semantic-Informed Quality Enhancement for Compressed Martian Image
Liu, C., Xu, M., Xing, Q., & Zou, X. (2025) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2025.3643953 · Cited by 3
“…of several metrics, including Peak Signal-to-Noise Ratio (PSNR), Peak Signal-to-Noise Ratio including Blocking effects (PSNR-B) [49], Structural Similarity Index Measure (SSIM) [50], Multi-Scale Structural Similarity (MS-SSIM) index [51], and Learned Perceptual Image Patch Similarity (LPIPS) [52].”
VQ-DeepVSC: A Dual-Stage Vector Quantization System for Video Semantic Communication
Miao, Y., Li, Z., Wang, Y., et al. (2025) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3641237 · Cited by 1
“To evaluate the system performance, we employ MS-SSIM [37] to compare the multiscale structural similarity at the pixel level between each frame of the original video and its reconstructed counterpart, and utilize LPIPS [36] to assess the perceptual quality of the reconstructed video from the human…”
Efficient Approximation of Earth Mover’s Distance Based on Nearest Neighbor Search
Meng, G., Zhou, R., Liu, L., Liang, P., Liu, F., Chen, D. Z., Niemier, M., & Hu, X. S. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2025.3623529 · Cited by 1
“The 8 evaluation metrics – PSNR [66], MS-SSIM [67], FSIM [68], BA [69], HI [70], Corr [71], BRISQUE [72], and NIQE [73] – are selected from the 20 metrics in [74] to cover 3 different aspects: structural similarity, color fidelity, and perceptual quality.”
Personalized Reward Modeling for Text-to-Image Generation
Lee, J., Heo, R., & Lee, D. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37356-4_25 · Cited by 6
Learned image compression with dual frequency branch modulation
Qiu, L., Bai, E., Wu, Y., et al. (2025) · Signal Image and Video Processing · doi:10.1007/s11760-025-04978-9 · Cited by 1
“Ballé et al. [7] first introduced a variational autoencoder (VAE)–based framework trained via rate–distortion optimization [8], surpassing traditional codecs in PSNR and MS-SSIM [9].”
A Noise Constrained Diffusion (NC-Diffusion) Framework for High-Fidelity Image Compression
Du, Z., Gao, Y., Li, S., Li, Y., Yuan, H., & Ye, M. (2025) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2025.3642340 · Cited by 3
“PSNR is used for distortion measurements, while MS-SSIM metric [65], LPIPS metric [66] and FID metric [67] are used for perceptual quality evaluation.”
Learning Land-Cover Change Inpainting Network for Spatiotemporal Remote Sensing Fusion
Zhu, B., Song, H., Zhang, X., et al. (2025) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2025.3644329 · Cited by 1
“For pixel loss, we utilize the Charbonnier loss from LapSRN [63], while for structural loss, we calculate the multiscale structural similarity (MS-SSIM) [64].”
Integrated Optical Camera Communication and Scene Sensing Based on Generative Adversarial Networks
Liu, W., Deng, Q., Yang, Z., Xiao, F., Zhou, B., & Geng, X. (2025) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2025.3649981 · Cited by 1
Xu, C., Xie, P., Kong, F., et al. (2026) · Multimedia Systems · doi:10.1007/s00530-025-02130-8 · Cited by 1
“The Peak Signal-to-Noise Ratio (PSNR) and Multi-Scale Structural Similarity (MS-SSIM) [18] of many frameworks outperform those of BPG and VVC.”
Syntax-Driven Multi-Realism Image Compression With Consistency Guided Diffusion Model
Kuang, H., Yang, W., Guo, Z., & Liu, J. (2026) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2026.3651547 · Cited by 1
“…[11], where PSNR quantifying the mean square error (MSE) between the original and compressed images, multi-scale structural similarity (MS-SSIM) [11] comparing patch-level similarities and visual information fidelity (VIF) o ff er insights into the fidelity of the reconstructed images in four…”
InfluentialA Superimposed Pilot Design Method for Semantic Transmission Over Doubly Selective Channel
Liang, H., Liu, X., Dong, C., et al. (2026) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2025.3650161 · Cited by 1
“4) Evaluation Metrics: The end-to-end image transmission performance is evaluated using the commonly adopted metric MS-SSIM [24], which is defined as follows, where µ x , σ x and σ xy are the mean, standard deviation, and cross-correlation between the two patches x , y , respectively.”
Compressed Video Quality Assessment With Fine-Grained Artifact Perception and Evaluation
Lin, L., Zhang, Y., Chen, B., Yang, J., & Zhao, T. (2026) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2026.3651043 · Cited by 2
“The comparison algorithms include PSNR [4], SSIM [5], MS-SSIM [6], 3D-PSD [61], VMAF [42], SpEED-QA [11], NIQE [62], VSFA [24], VIIDEO [43], TLVQM [16], MDTVSFA [63], Fast_Deep_NR-VQA [44], STFEE [45], HSTVQA [46], FAST-VQA [47], STAN [48], HVS-5M [49], and NR-RankDVQA [23].”
Boosting Neural Video Representation via Online Structural Reparameterization
Li, Z., Mao, Q., Liu, S., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5679-3_35 · Cited by 1
Zhang, W., Yu, Z., Huang, H., et al. (2026) · IEEE Transactions on Communications · doi:10.1109/tcomm.2026.3653875 · Cited by 1
“We adopted the MS-SSIM as the metric to evaluate the perceptual quality of the reconstructed images after transmission [44].”
A General Image Fusion Approach Exploiting Gradient Transfer Learning and Fusion Rule Unfolding
Wang, W., Deng, L., Cao, Q., & Vivone, G. (2026) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2026.3655694 · Cited by 2
“For the MEIF task with reference images, we used PSNR, MEF-SSIM [86], and MS-SSIM [84] to evaluate the fidelity between the fused image and the reference image.”
InfluentialZhang, M., Wu, Z., Li, H., Li, J., Pang, L., & Qian, C. (2025) · IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom) · doi:10.1109/trustcom66490.2025.00171 · Cited by 1
“The MS-SSIM [23] is an extension of the structural similarity index (SSIM), designed to evaluate image quality by incorporating multi-scale analysis to better align with human visual perception.”
Skillful Short‐Term Forecasting of Clouds With a Cascade Diffusion Model
Chen, H., Zhong, X., Zhai, Q., Li, X., Chan, Y. W., Chan, P. W., Yang, M., Huang, Y., Li, H., & Shi, X. (2026) · Journal of Geophysical Research Machine Learning and Computation · doi:10.1029/2025jh000976 · Cited by 10
Qi, X., Ma, N., Lin, Y., et al. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3662328 · Cited by 1
“The FIVSSC is tested using peak signal-to-noise ratio (PSNR) or multi-scale structure similarity (MS-SSIM) [43] as evaluation metrics.”
Han, L., Liu, H., Zhang, W., et al. (2026) · Acta Geotechnica · doi:10.1007/s11440-026-02923-9 · Cited by 1
Shen, K., Zhou, X., & Liu, Z. (2026) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2026.3664887 · Cited by 1
“L bce and L ssim represent the bce loss and ssim loss [31], respectively.”
Image Coding for Machines via Feature-Preserving Rate-Distortion Optimization
Fernández-Menduiña, S., Pavéz, E., & Ortega, A. (2026) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2026.3664944 · Cited by 3
“We report 1) mean average precision (mAP@[0.5:0.05:0.95]), 2) Y-PSNR, and 3) Y-MS-SSIM [47].”
InfluentialA Novel Image Similarity Metric For Scene Composition Structure
Haque, M. R., Murshed, M., Paul, M., & Lee, T.-K. (2025) · IEEE International Conference on Image Processing Workshops (ICIPW) · doi:10.1109/icipw68931.2025.11386242 · Cited by 2
“However, developing such models also requires metrics capable of evaluating their success in preserving SCS. Commonly used image similarity metrics such as PSNR, SSIM [7], MS-SSIM [8], FID [9], LPIPS [10], and CLIP Score [11] are ill-suited for this task.”
Enhanced image compression using wavelet-based encoder-decoder architectures
Roshan, R., Aghazadeh, N., & Farajzadeh, N. (2026) · Signal Image and Video Processing · doi:10.1007/s11760-026-05134-7 · Cited by 1
“To evaluate the performance of the models, we employed three key metrics: Peak Signal-to-Noise Ratio (PSNR), Compression Ratio (CR), and Multiscale Structural Similarity (MS-SSIM) [22].”
3DGS-VBench: A Comprehensive Video Quality Evaluation Benchmark for 3DGS Compression
Xing, Y., Gordon, W., Yang, Q., Yang, K., Wang, J., & Xu, Y. (2025) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip67698.2025.11396862 · Cited by 1
“Our evaluation of 15 IQA/VQA metrics reveals limitations on 3DGS-specific distortions and establishes a standardized benchmark comparing 6 algorithms across storage efficiency and visual quality.”
Contrast-free super-resolution ultrasound imaging based on two-stage diffusion probabilistic models
Zhou, B., Zhang, G., Ren, X., Gu, W., Zhu, Q., Wang, J., & Liu, X. (2026) · Ultrasonics · doi:10.1016/j.ultras.2026.108033 · Cited by 1
Liu, J., Sun, G., Dong, L., & Zhang, H. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.115602 · Cited by 3
Huang, L., Zhu, J., Su, Q., Song, J., Ren, K., Ni, W., & Shi, X. (2026) · Remote Sensing · doi:10.3390/rs18050710 · Cited by 1
An Efficient FPGA Implementation for Semantic Communication Encoder
Li, Z., Qin, T., Wang, Q., et al. (2025) · doi:10.1109/ftcs68006.2025.11405794 · Cited by 1
“Using the MS-SSIM [11] and PSNR of the image generated by floating-point model as the evaluation metric.”
DOOMGAN: High-Fidelity Dynamic Identity Obfuscation Ocular Generative Morphing
Krishnamurthy, B., & Rattani, A. (2025) · IEEE International Joint Conference on Biometrics (IJCB) · doi:10.1109/ijcb65343.2025.11410742 · Cited by 1
“The D produces multi-scale real ( y i ) and fake (˜ y i ) scores, providing rich feedback to the generator that encourages the synthesis of more realistic images.”
Semantic Communication for High-Speed Mobility Scenarios: A Semantic-OTFS Scheme
Liu, Y., Dong, C., Liang, H., Li, W., Xu, X., & Zhang, P. (2026) · IEEE Transactions on Vehicular Technology · doi:10.1109/tvt.2026.3670829 · Cited by 3
“2) Evaluation Metrics: SSIM [31], MS-SSIM [35] and LPIPS [27] are used as evaluation metrics, where SSIM and MS-SSIM focus on image structural similarity evaluation, while LPIPS is based on deep learning more focused on the evaluation of image perceptual quality, note that this metric is in the…”
Fukushima, N., Maeda, Y., Kameda, Y., & Watanabe, O. (2025) · Picture Coding Symposium · doi:10.1109/pcs65673.2025.11417494 · Cited by 2
“The IQA metrics used are as follows: Grayscale IQA ( PSNR-Y, IW-PSNR [16], PSNR-HVS [9], [10], SSIM [11], MS-SSIM [12], IW-SSIM [16], VIF [17], GMSD [18], FSIM [19], NLPD [20] ), Color IQA ( PSNR-RGB, MDSI [21], Butteraugli (p=1,2,3, max) [22], [23], SSIMULACRA2 [24]) and VMAF [25], a video IQA,…”
TextFuse: Text Information Guided Infrared and Visible Image Fusion
Lv, G., Sima, C., Wang, J. Y., Gao, Y., Wei, Z., & Cheng, J. (2026) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2026.3673589 · Cited by 4
“2) Evaluation Metrics: We select some objective evaluation metrics, including PSNR [40], CC [41], Q AB/F [42], SSIM [36], MS − SSIM [43], N ab/f [44], FMI pixel [45], SCD [46], V IF [47].”
Satellite-GS: Enhanced 2D Gaussian Splatting for Robust Satellite Reconstruction
Wang, J., Liu, M., Yang, X., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3675851 · Cited by 1
“…patch similarity (LPIPS) [46] captures perceptual di ff erences that correlate well with human visual perception, and multi-scale SSIM (MS-SSIM) [47] evaluates structural consistency across multiple scales, which is particularly important for preserving both coarse-grained satellite geometry…”
Generation of Synthetic Laryngoscopy Images Using StyleGAN3
Morrison, D. A., Mohanty, A. S., Dehzangi, I., et al. (2025) · 2018 52nd Asilomar Conference on Signals, Systems, and Computers · doi:10.1109/ieeeconf67917.2025.11443461 · Cited by 1
“MS-SSIM was computed in two forms [12]: 1.”
ML‐PEA: Machine Learning‐Based Perceptual Algorithms for Display Power Optimization
Chen, K., Matsuda, N., Wan, T., et al. (2026) · Computer Graphics Forum · doi:10.1111/cgf.70369 · Cited by 1
Finger Vein Sample Compression using Recent AI-Based Still Image Compression Schemes
Kauba, C., Angerer, G., & Uhl, A. (2025) · IEEE International Workshop on Information Forensics and Security · doi:10.1109/wifs66636.2025.00030 · Cited by 1
“To quantify the image quality of the finger vein samples, MS-SSIM [22] as a representative of a full-reference image quality metric, NIQE [23] as a representative of no-reference image quality metrics and GCF [24] and Wang17 [25] as vein pattern specific quality metrics are employed.”
Deep Learning Methods for SAR and Optical Image Fusion: A Review
Guo, C., Zhang, Z., Huang, K., et al. (2026) · Remote Sensing · doi:10.3390/rs18081196 · Cited by 1
Wang, M., Chen, X., Wang, G., Wang, R. X., Du, L., & Zhang, Z. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3685195 · Cited by 2
“It can be calculated using the multiscale structural similarity (MS-SSIM) [50], formulated as where I is a tensor of ones, N is the index of the highest scale, and l i , c i , and s i represent the luminance, contrast, and structure comparisons at the i th scale, respectively. α i , β i , and γ i…”
NDVI Change-Guided Transformer–Mamba Hybrid Network for Spatiotemporal Fusion
Liang, Y., Zeng, Z., Li, F., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3686510 · Cited by 1
Zhou, X., Mo, Z., Li, G., Bao, L., Xu, X., & Zhang, J. (2026) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2026.3684146 · Cited by 1
“Our loss function can be expressed as where l bce , l iou , and l ssim denote BCE loss [66], IOU loss [67], and SSIM loss [68], respectively.”
ATDFusion: Adapter-Tuned Dual-Branch Network for multimodal medical image fusion
Li, C., Zhu, R., Zhao, H., Li, X., & Liu, Z. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.113848 · Cited by 1
End-to-End RGB-D Image Compression With Efficient Mamba Cross-Attention
Zheng, H., & Gao, W. J. (2026) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2026.3688935 · Cited by 1
“Additionally, MS-SSIM [76], a commonly used metric in image compression field, is also applied to evaluate image quality.”
Neural Geometry Image-Based Representations with Optimal Transport (OT)
Gao, X., Liu, Y., Li, J., Wang, X., Guo, M., Guo, Y., Song, X., Yu, H., Lao, Z., & Xianfeng, D. (2026) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61042.2026.00645 · Cited by 1
Agentic Generative Semantic Communication for Adaptive Image Transmission
Hao, J., Yue, C., Yang, S., Shi, Y., Liu, W., & Vucetic, B. S. (2025) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps68340.2025.11590981 · Cited by 1
“2) Evaluation Metrics: We employ three complementary metrics to evaluate our framework: CLIP score [17] (higher is better), LPIPS [18] (lower is better), and MS-SSIM [19] (higher is better).”
Hierarchical Optimization for Energy-Efficient Semantic Communications With Heterogeneous Users
Song, K., Fan, R., Zhan, C., Qin, Z., Hu, H., Yang, J., & Guo, S. (2026) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2026.3713752 · Cited by 1
“We adopt the Multi-Scale Structural Similarity Index (MS-SSIM) as the primary performance metric, as it correlates well with human visual perception of image quality [31].”
Möller, B., Li, Z., Stelzer, M., Graave, T., Bettels, F., Ataya, M., & Fingscheidt, T. (2026) · IEEE Intelligent Vehicles Symposium Workshops (IV Workshops) · doi:10.1109/ivworkshops71735.2026.11623987 · Cited by 1
“Metrics: To assess image reconstruction quality, we use PSNR [41], SSIM [42], MS-SSIM [43], and LPIPS [44], comparing decoded images to ground-truth references.”
DINOLight: Robust Ambient Light Normalization with Self-supervised Visual Prior Integration
Oh, Y., Kwon, J., & Cho, N. I. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31335-5_8 · Cited by 4
SIRR-LMM: Single-Image Reflection Removal via Large Multimodal Model
Guo, Y., Lao, Z., Song, X., Zhou, Y., & Yu, H. (2026) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw68408.2026.00181 · Cited by 2
Cheng, X., Liu, Y., Wang, Q., et al. (2026) · Applied Optics · doi:10.1364/ao.606833 · Cited by 1
Gao, H., Wu, X., Li, J., et al. (2026) · Geophysical Journal International · doi:10.1093/gji/ggag295 · Cited by 1
Automatic Bound Estimation for JPEG 2000 Compressing Leukocytes Images
Falcón-Ruíz, A., Paz-Viera, J. E., Taboada-Crispí, A., et al. (2012) · World Congress on Medical Physics and Biomedical Engineering, September 7 - 12, 2009, Munich, Germany · doi:10.1007/978-3-642-21198-0_140 · Cited by 1
Methods of 3D images quality assesment
Sakowicz, B., Kamiński, M., Ritter, R., Kotas, R., Marciniak, P., Kupis, J., Sękalski, P., & Napieralski, A. (2015) · International Conference Mixed Design of Integrated Circuits and Systems · doi:10.1109/mixdes.2015.7208494 · Cited by 1
Binocular combination and fractional differential based 3D image quality assessment
Xu, H., Jiang, G., Mei, Y., et al. (2015) · International Workshop on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex.2015.7148147 · Cited by 1
A novel image quality assessment metric using singular value decomposition
Ali, S. S. (2015) · Canadian Workshop on Information Theory (CWIT) · doi:10.1109/cwit.2015.7255178 · Cited by 1
Local Structure Divergence Index for Image Quality Assessment
Gao, F., Tao, D., Li, X., et al. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-34500-5_40 · Cited by 1
A no-reference video quality metric using a Natural Video Statistical Model
Galea, C., & Farrugia, R. A. (2015) · IEEE EUROCON - International Conference on Computer as a Tool (EUROCON) · doi:10.1109/eurocon.2015.7313754 · Cited by 1
Depth map post-processing for depth-image-based rendering: a user study
Nezveda, M., Brosch, N., Seitner, F. H., & Gelautz, M. (2014) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2039771 · Cited by 1
A global image fidelity metric: Visual distance and its properties
Richter, T. (2013) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2013.6738076 · Cited by 2
Fusion of imprecise data applied to image quality assessment
Guettari, N., Capelle-Laizé, A.-S., & Carré, P. (2014) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2014.7025104 · Cited by 2
Correlation analysis for quality assessment of Region-of-Interest-coded images
Garcia-Alvarez, J. C., & Diaz-Solarte, P. C. (2013) · Computing Colombian Conference · doi:10.1109/colombiancc.2013.6637521 · Cited by 1
“SIMilarity (MS–SSIM) [12], the Visual Image Fidelity (VIF) [13], the Quality Index based on Local Variance (QILV) [14], the Reflection Factor (RF) [15], the Visual Signal to Noise Ratio (VSNR) [16], and the Divisive Normalization (DN) [17].”
Towards a new image quality metric for evaluating the effects of tiled displays
McFadden, S. B., & Ward, P. A. S. (2014) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2014.7025112 · Cited by 1
Perceptual tools for quality-aware video networks
Bovik, A. C. (2014) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2046685 · Cited by 1
Self-citationPerceptual color difference assessment using histogram distance on hue histogram descriptor
Lee, D., & Plataniotis, K. N. (2014) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2014.7025109 · Cited by 1
“Index Terms— color image, perceptual image difference, hue histogram descriptor, circular histogram, histogram distance”
BackgroundA perceptual distortion measure for video quality assessment using wavelet and statistical measures
Shadiya, P., & Balachandran, K. P. (2015) · IEEE International Conference on Signal Processing, Informatics, Communication and Energy Systems (SPICES) · doi:10.1109/spices.2015.7091376 · Cited by 1
A web-based digital analysis interface for image quality assessment
Yalman, Y., Akar, F., & Bayılmış, C. (2013) · The Imaging Science Journal · doi:10.1179/1743131x12y.0000000047 · Cited by 2
Assessment of random-noise contamination in digital images via testing on wavelet coefficients
Maldonado, D., Silwal, S., & Wang, H. (2013) · Statistics and Its Interface · doi:10.4310/sii.2013.v6.n1.a11 · Cited by 1
“3.3 MS-SSIM, VIF and IW-SSIM Among all the full-reference image quality assessment methods in the literature, the three multi-scale decomposition methods MS-SSIM (Wang et al., 2003), VIF (Sheikh and Bovik, 2006) and IW-SSIM (Wang and Li, 2011) are most relevant to our article.”
MethodInfluentialStudy of Multi-Scale temporal extension and Minkowski a MSSIM
Dhungel, A., & Weeks, M. (2010) · International Symposium on Communications and Information Technologies · doi:10.1109/iscit.2010.5665114 · Cited by 2
No-reference perceptual blur metric for stereoscopic images
Fezza, S. A., & Larabi, M. (2014) · International Conference on D Imaging · doi:10.1109/ic3d.2014.7032601 · Cited by 1
“The first three lines in the table are obtained by averaging the left and right 2DIQA scores, obtained with PSNR, SSIM [14], MS-SSIM [43].”
MethodUsing JPEG and JPEG2000 compressions for fast image quality metrics based on free energy theory
Gu, K., Zhai, G., Liu, M., Yang, X., & Zhang, W. (2014) · IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB) · doi:10.1109/bmsb.2014.6873472 · Cited by 1
SSPQ - spatial domain perceptual image codec based on subsampling and perceptual quantization
Wang, Z., Simón, S., Klaiber, M. J., Ahmed, S., & Richter, T. (2012) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2012.6467046 · Cited by 1
“given by MS-SSIM [11] have shown closely comparable results between JPEG-LS and SSPQ in our tests, it has been found that lossy JPEG-LS (especially with δ>3) tends to cause artifacts that are clearly visible in local regions of certain images whereas error visibility is much less in the same regions of SSPQ-compressed…”
MethodInfluentialSelf-citation
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.