Multiscale structural similarity for image quality assessment
The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index; none of the 14 sources its claims cite is checked for support yet.
Abstract
The structural similarity image quality paradigm is based on the assumption that the human visual system is highly adapted for extracting structural information from the scene, and therefore a measure of structural similarity can provide a good approximation to perceived image quality. This paper proposes a multiscale structural similarity method, which supplies more flexibility than previous single-scale methods in incorporating the variations of viewing conditions. We develop an image synthesis method to calibrate the parameters that define the relative importance of different scales. Experimental comparisons demonstrate the effectiveness of the proposed method.
Overview
CoChat Summary
TL;DR
This paper proposes a multi-scale structural similarity (MS-SSIM) image quality method that combines contrast and structure comparisons across scales and luminance comparison at the coarsest scale. On the LIVE JPEG/JPEG2000 database, MS-SSIM outperformed PSNR, the Sarnoff model, and single-scale SSIM variants on all reported evaluation criteria.
Key findings
- MS-SSIM achieved the best reported performance: CC 0.969, ROCC 0.966, MAE 3.86, RMS 4.91, and OR 1.16%, compared with the best single-scale SSIM at M=2 (CC 0.963, ROCC 0.959, MAE 4.21, RMS 5.38, OR 2.62%) and PSNR (CC 0.905, ROCC 0.901, MAE 6.53, RMS 8.45, OR 15.7%).
- Single-scale SSIM performance varied with scale; the best single-scale result was at M=2, and single-scale SSIM tended to give higher scores as scale increased.
- The calibrated cross-scale exponents were β1=γ1=0.0448, β2=γ2=0.2856, β3=γ3=0.3001, β4=γ4=0.2363, and α5=β5=γ5=0.1333.
- The authors conclude that multi-scale SSIM provides a meaningful balance between scales and outperforms the best single-scale SSIM and state-of-the-art image quality metrics.
Methods
- The MS-SSIM system iteratively applies a low-pass filter and downsamples by 2; at each scale it computes contrast and structure comparisons, computes luminance comparison only at Scale M, and combines them using exponents normalized so that ∑γj=1.
- Cross-scale parameters were calibrated by image synthesis: 5 scales and 12 distortion levels (range 2^3 to 2^14) were used with 10 original 64×64 images, producing 600 distorted images; 8 subjects selected images of equal perceived quality across scales at a viewing distance of 32 pixels per degree of visual angle.
- Evaluation used the LIVE database with 344 JPEG and JPEG2000 compressed images (bit rate 0.028 to 3.150 bits/pixel), MOS from 13–25 subjective scores, and compared PSNR, Sarnoff JNDmetrix 8.0, single-scale SSIM for M=1–5, and MS-SSIM using CC, MAE, RMS, OR, and ROCC after logistic mapping.
Limitations
- Stated: the image-synthesis calibration approach is “still rather crude,” and the authors are working on a more systematic approach.
- Stated: CSF cannot be directly used to derive the parameters because it is typically measured at visibility threshold with simplified stimuli, whereas the goal is comparing complex structured images at visible distortion levels.
- Evident from the text: calibration used only 10 original 64×64 grayscale images and 8 subjects, including one author, and evaluation was limited to the LIVE JPEG/JPEG2000 database.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 14 Not checked · 18 Own results and interpretations
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It can also be observed that the single-scale model tends to supply higher scores with the increase of scales.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Finally, for every one of the objective evaluation criteria, multi-scale SSIM model outperforms all the other models, including the best single-scale SSIM model, suggesting a meaningful balance between scales.”
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
“Experiments show that with an appropriate parameter settings, the multi-scale method outperforms the best single-scale SSIM model as well as state-of-the-art image quality metrics.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In general, the subjects agreed with each other on each image more than they agreed with themselves across different images.”
plainstates an associationn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The resulting parameters we obtained are β 1 = γ 1 = 0.0448, β 2 = γ 2 = 0.2856, β 3 = γ 3 = 0.3001, β 4 = γ 4 = 0.2363, and α 5 = β 5 = γ 5 = 0.1333, respectively.”
plainn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“However, the SSIM index algorithm introduced in [5] is a single-scale approach.”
What the sources say
- Not matchedcites [5] [5] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, 'Image quality assessment: From error measurement to structural similarity,' IEEE Trans. Image Processing , vol. 13, Jan. 2004. -
More than one paper fits this citation.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
10 more claims
Create a free account to see all 33 claims and every source verdict.
- It may state two separate claims as one. 8
- Probably not a substantial claim. 13
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Further, as research in quality assessment has demonstrated, incorporating multiscale information when assessing quality produces QA algorithms that perform better in terms of correlation with human perception [13], [29].”
From No-Reference Image Quality Assessment in the Spatial Domain (2012) · cites this paper as [13]
Citations
Cites
Image quality assessment: from error visibility to structural similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · IEEE Transactions on Image Processing · doi:10.1109/tip.2003.819861 · Cited by 57,892
Self-citationA universal image quality index
Wang, Z., & Bovik, A. C. (2002) · IEEE Signal Processing Letters · doi:10.1109/97.995823 · Cited by 5,757
Self-citationImage quality measures and their performance
Eskicioglu, A. M., & Fisher, P. S. (1995) · IEEE Transactions on Communications · doi:10.1109/26.477498 · Cited by 2,117 · cited 1×
“Traditional perceptual image quality assessment methods are based on a bottom-up approach which attempts to simulate the functionality of the relevant early human visual system (HVS) components.”
BackgroundA Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients
Portilla, J., & Simoncelli, E. P. (2000) · International Journal of Computer Vision · doi:10.1023/a:1026553619983 · Cited by 1,806 · cited 2×
“These synthesis methods provide a powerful and efficient means of testing a model, and have the added benefit that the resulting images suggest improvements that might be made to the model [11].”
MethodSelf-citation(1995) · Medical Entomology and Zoology · Cited by 1,212
Pyramid-based texture analysis/synthesis
Heeger, D. J., & Bergen, J. R. (1995) · Proceedings of the 22nd annual conference on Computer graphics and interactive techniques - SIGGRAPH '95 · doi:10.1145/218380.218446 · Cited by 1,095
Video quality assessment based on structural distortion measurement
Wang, Z., Lu, L., & Bovik, A. C. (2004) · Signal Processing: Image Communication · doi:10.1016/s0923-5965(03)00076-6 · Cited by 1,024
Self-citationImage Quality Assessment: From Error Measurement to Structural Similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · Cited by 919
Self-citationTeo, P. C., & Heeger, D. J. (2002) · Proceedings of 1st International Conference on Image Processing · doi:10.1109/icip.1994.413502 · Cited by 471
Objective video quality assessment
(2003) · Handbook of Video Databases · doi:10.1201/9780203489864-45 · Cited by 359
Research works printed but not linked: 2
- 0, JNDmetrix Technology, Sarnoff Corporation
- 2000, Final report from the video quality experts group on the validation of objective models of video quality assessment, VQEG
Cited by
Items 3,601–3,700 of 6,047
A Deep Learning Approach to No-Reference Image Quality Assessment For Monte Carlo Rendered Images
Whittle, J., & Jones, M. W. (2018) · Eurographics · doi:10.2312/cgvc.20181204 · Cited by 4
Full-Reference Video Quality Assessment Using Deep 3D Convolutional Neural Networks
Dendi, S. V. R., Krishnappa, G., & Channappayya, S. S. (2019) · National Conference on Communications · doi:10.1109/ncc.2019.8732265 · Cited by 5
“For spatial quality estimation we use the robust MS-SSIM index [11], and for spatio-temporal quality estimation, we measure the deviation between pristine and test video 3D ConvNet features.”
MethodA Conditional Generative Adversarial Network for Rendering Point Clouds
Atienza, R. (2019) · Cited by 6
Disentangled Human Action Video Generation via Decoupled Learning
Yang, L., Zhao, Z., Wang, S., Wang, S., Ma, S., & Gao, W. (2019) · IEEE International Conference on Multimedia & Expo Workshops (ICMEW) · doi:10.1109/icmew.2019.00091 · Cited by 3
“We adopt two commonly used objective evaluation metrics, MS-SSIM [22] and LPIPS [23] to account for both signal-level statistical fidelity and semantic-level perceptual quality.”
MethodPrivacy-Net: An Adversarial Approach For Identity-obfuscated Segmentation.
Kim, B. N., Desrosiers, C., Dolz, J., & Jodoin, P. (2019) · Cited by 4
Objective Quality Assessment for Light Field Based on Refocus Characteristic
Meng, C., An, P., Huang, X., et al. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-34113-8_17 · Cited by 4
Multi-scale Generative Adversarial Learning for Facial Attribute Transfer
Zhang, Y., Song, L., Xie, R., et al. (2020) · Communications in computer and information science · doi:10.1007/978-981-15-3341-9_8 · Cited by 4
Tao, T., Ding, L., & Huang, H. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.02.084 · Cited by 5
Feasibility Analysis of Deep Learning-Based Reality Assessment of Human Back-View Images
Kwon, Y. C., Jang, J. W., Lim, H., & Choi, O. (2020) · Electronics · doi:10.3390/electronics9040656 · Cited by 4
“In Table 1, ‘Masked SSIM’ and ‘Masked MS-SSIM’ are the structural similarity index [5] and its multi-scale version [24] of the masked regions of the images, excluding the background.”
BackgroundInpainting of Ring Artifacts on Microtomographic Images by 3D CNN
Kornilov, A., Safonov, I. V., & Yakimchuk, I. (2020) · Conference of Open Innovations Association (FRUCT) · doi:10.23919/fruct48808.2020.9087422 · Cited by 6
“We analyze an application of conventional mean-squared-error (MSE) loss L MSE , loss function based on multi-scale structural similarity index normalized to a range from 0 to 1 (MS-SSIM) (3) [29], and the weighted sum of this loss and mean absolute error (MAE) (4).”
MethodFeng, X., Su, X., Lian, X., et al. (2019) · doi:10.1109/icusai47366.2019.9124786 · Cited by 4
“Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM) are widely used to measure image quality [4].”
MethodDeep Multi-path Low-Light Image Enhancement
Li, S., Cheng, Q. S., & Zhang, J. (2020) · IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) · doi:10.1109/mipr49039.2020.00025 · Cited by 5
“Our quality loss provides a sensitivity of the human visual system (HVS) by using multi-scale structural similarity [21].”
BackgroundArtefact removal in ground truth deficient fluctuations-based nanoscopy images using deep learning
Jadhav, S., Acuña, S., Opstad, I. S., Ahluwalia, B. S., Agarwal, K., & Prasad, D. K. (2020) · Biomedical Optics Express · doi:10.1364/boe.410617 · Cited by 5
“In the original article [58], their values are set to 1, and we have used the same.”
MethodInfluentialSeemann, M., Bargsten, L., & Schlaefer, A. (2020) · Current Directions in Biomedical Engineering · doi:10.1515/cdbme-2020-0015 · Cited by 3
Parameters optimization of the Structural Similarity Index
Bakurov, I., Buzzelli, M., Castelli, M., et al. (2020) · London Imaging Meeting · doi:10.2352/issn.2694-118x.2020.lim-13 · Cited by 4
“Multiscale SSIM (MS-SSIM) [22] is an extension of SSIM based on its efficient application at different resolutions.”
MethodInfluentialDesign of an Intelligent Symptom Differentiation and Electrical Stimulation Rehabilitation System
Liu, J., & Kai, L. (2020) · Journal Européen des Systèmes Automatisés · doi:10.18280/jesa.530511 · Cited by 4
“For instance, many image quality assessment algorithms choose to decompose the original image into several sub-bands [10, 11] in the preprocessing stage, including visual information fidelity (VIF), multiscale structural similarity (MS-SSIM) [12], full reference-difference of Gaussian (FR-DOG) [13], and most apparent…”
BackgroundOn visual masking estimation for adaptive quantization using steerable filters
Paul, S., Norkin, A., & Bovik, A. C. (2021) · Signal Processing Image Communication · doi:10.1016/j.image.2021.116290 · Cited by 4
Self-citationVideo Quality Prediction Using Voxel-Wise fMRI Models of the Visual Cortex
Mahankali, N. S., & Channappayya, S. S. (2021) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp39728.2021.9414328 · Cited by 4
Monte Carlo Optimization of a Combined Image Quality Assessment for Compressed Images Evaluation
Bouida, A., Khelifi, M., Beladgham, M., & Hamlili, F.-Z. (2021) · Traitement du signal · doi:10.18280/ts.380204 · Cited by 6
Performance Assessment of the Adaptive GoP-size extension of the Wireless SoftCast Video Scheme
Trioux, A., Coudoux, F., Corlay, P., & Gharbi, M. (2021) · International Symposium on Signal, Image, Video and Communications (ISIVC) · doi:10.1109/isivc49222.2021.9487549 · Cited by 5
Lindskog, E., & Carlsson, N. F. (2021) · ACM Multimedia Systems Conference (MMSys) · doi:10.1145/3458305.3478453 · Cited by 4
“[27] to extract the MS-SSIM, PSNR, and other per-frame scores for every frame.”
BackgroundMethodGradient Cepstrum Combined with Simplified Extreme Channel Prior for Blind Deconvolution
Liu, J., Tan, J., & He, L. (2021) · Circuits Systems and Signal Processing · doi:10.1007/s00034-021-01827-1 · Cited by 4
“8, our approach exceeds the classical algorithms [5, 6, 27, 30, 45, 48, 49, 51] in terms of Table 3 demonstrates the result of average PSNR and SSIM values on this data-set.”
MethodInfluentialOriole: Thwarting Privacy Against Trustworthy Deep Learning Models
Chen, L., Hu, W., Zhao, B. Z. H., Xue, M., & Qian, H. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-90567-5_28 · Cited by 5
“Next, the calculation of a cloak δ ( x, x T i ) is defined as: where δ subjects to | δ ( x, x T i ) | ρ , and | δ ( x, x T i ) | is calculated by DSSIM (Structure Dis-Similarity Index) [42, 44] and ρ is the perturbation budget.”
MethodQuality Assessment of Images Using SSIM Metric and CIEDE2000 Distance Methods in Lab Color Space
Chandrakanth, T., & Sandhya, B. (2014) · Advances in intelligent systems and computing · doi:10.1007/978-3-319-12012-6_50 · Cited by 4
Kadir, R., Saha, R., Akhter, M. M., Awal, M. A., & Kadir, M. I. (2021) · Array · doi:10.1016/j.array.2021.100107 · Cited by 4
GADO-Net: An Improved AOD-Net Single Image Dehazing Algorithm
Chen, S., Cheng, J., & Huang, Z. (2021) · International Academic Exchange Conference on Science and Technology Innovation (IAECST) · doi:10.1109/iaecst54258.2021.9695731 · Cited by 5
Evaluation of Image Quality Assessment Metrics: Color Quantization Noise
Hassan, M. I., & Bhagvati, C. (2015) · International Journal of Applied Information Systems · doi:10.5120/ijais15-451367 · Cited by 7
Optimal server bandwidth for mobile video on demand
Buchinger, S., & Hlavacs, H. (2010) · Annals of Telecommunications · doi:10.1007/s12243-009-0152-8 · Cited by 6
“Amongst others, the MSU tool computes three video quality metrics called structural similarity index (SSIM) [19, 20 ], discrete cosine transform-based video quality metric (VQM) 3 and mean squared error (MSE) [21].”
MethodThe impact of skull bone intensity on the quality of compressed CT neuro images
Urbaniak, I. A., Vrscay, E. R., Wang, Z., et al. (2012) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.912467 · Cited by 4
Self-citationImage Multidistortion Estimation
Caron, A. L., & Jodoin, P. (2011) · IEEE Transactions on Image Processing · doi:10.1109/tip.2011.2159233 · Cited by 4
Enhancing the Intra-prediction in JPEG-XR by Using Edge Information
Maalouf, A., & Larabi, M. (2009) · International Conference on Signal-Image Technology and Internet-Based Systems (SITIS) · doi:10.1109/sitis.2009.32 · Cited by 4
SSIM and their dynamic range for image quality assessment
Silvestre-Blanes, J., & Pérez-Lloréns, R. (2011) · Proceedings ELMAR-2011 · Cited by 5
Quantitative assessment mechanism transcending visual perceptual evaluation for image dehazing
Zhu, Q., Hu, Z., & Ivanov, K. (2015) · IEEE International Conference on Robotics and Biomimetics (ROBIO) · doi:10.1109/robio.2015.7418869 · Cited by 6
Image quality assessment algorithms: Study and performance comparison
Ruikar, J. D., Sinha, A. K., & Chaudhury, S. (2014) · International Conference on Electronics and Communication Systems (ICECS) · doi:10.1109/ecs.2014.6892744 · Cited by 4
Objective Video Quality Assessment Based on Neural Networks
Menor, D. P. A., Mello, C. A. B., & Zanchettin, C. (2016) · Procedia Computer Science · doi:10.1016/j.procs.2016.08.202 · Cited by 5
Efficient lightweight video packet filtering for large-scale video data delivery
Corbillon, X., Boyrivent, F., Williencourt, G. A. D., et al. (2016) · IEEE International Conference on Multimedia & Expo Workshops (ICMEW) · doi:10.1109/icmew.2016.7574700 · Cited by 4
“We evaluate the Multiscale - Structural Similarity (MS-SSIM) video quality metric [18] that can be achieved when our packet filtering algorithm is implemented on a set of videos for various bandwidth shortage conditions.”
MethodEvaluation of JPEG XT for high dynamic range cameras
Hanhart, P., & Ebrahimi, T. (2016) · Signal Processing Image Communication · doi:10.1016/j.image.2016.10.004 · Cited by 4
Towards Generating Personalized Volumetric Phantom from Patient’s Surface Geometry
Wu, Y., Singh, V. K., Teixeira, B., Ma, K., Tamersoy, B., Krauß, A., & Chen, T. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-00928-1_20 · Cited by 4
PERFORMANCE EVALUATION OF OBJECTIVE QUALITY METRICS ON HLG-BASED HDR IMAGE CODING
Sugito, Y., & Bertalmı́o, M. (2018) · IEEE Global Conference on Signal and Information Processing · doi:10.1109/globalsip.2018.8646673 · Cited by 5
A Novel Method for Quality Assessment of Image Stitching Based on the Gabor Filtering
Zhu, S., Zhang, Y., Liang, T., Liu, T., & Liu, Y. (2018) · IEEE International Conference on Information and Automation (ICIA) · doi:10.1109/icinfa.2018.8812565 · Cited by 4
“[6] was more effective than a single scale.”
BackgroundPeng, J., Gao, R., Nguyen, L., Liang, Y., Thng, S. T. G., & Lin, Z. (2019) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2019.8802993 · Cited by 7
“…results; iii) A redundancy reduction scheme is developed such that the similarity between the generated images and the original images is assessed using Structural Similarity index (SSIM index) [11, 12], and images having higher similarity with the original dataset are deleted to avoid redundancy.”
MethodSingle View Distortion Correction using Semantic Guidance
Molnár, S.-B. L., Pável, S., & Csato, L. (2019) · doi:10.1109/ijcnn.2019.8852065 · Cited by 4
“The proposed loss is given by Lr = − MS-SSIM(I, I ′) + 1 2 .”
MethodInfluentialTang, Y., & Zhu, Y. (2020) · Symmetry · doi:10.3390/sym12040661 · Cited by 4
“Image zooming results using different methods will be compared by the visual quality, edge detection by Prewitt operator and calculating peak signal to noise ratio (PSNR), structural similarity index (SSIM), feature similarity (FSIM) [28], multiscale structural similarity (MS-SSIM) [29].”
MethodInfluentialFull‐reference tone‐mapped images quality assessment
Faraji, M. R. (2020) · IET Image Processing · doi:10.1049/ipr2.12060 · Cited by 4
Benchmarking Single Image Dehazing Methods
Nair, D., & Sankaran, P. (2021) · SN Computer Science · doi:10.1007/s42979-021-00925-w · Cited by 6
“To evaluate image quality at multiple resolutions, the multi-scale SSIM (MSSSIM) index can be used [110].”
MethodZhang, F., & Bull, D. (2014) · Academic press library in signal processing · doi:10.1016/b978-0-12-420149-1.00007-7 · Cited by 4
Deviation Based Pooling Strategies For Full Reference Image Quality AssessmentPreprint
Nafchi, H. Z., Hedjam, R., Shahkolaei, A., & Cheriet, M. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1504.06786 · Cited by 4
“A number of popular and/or high performance IQAs are SSIM [14], MS-SSIM [17], VIF [11], VSNR [2], MAD [4], FSIM [23], GS [7], GMSD [20], VSI [21], and others [5].”
BackgroundMethodEffective Visual Masking Techniques in JPEG2000
Richter, T. (2008) · DCC · doi:10.1109/dcc.2008.43 · Cited by 4
Embedded Systems and Applications
(2022) · doi:10.5121/csit.2022.1206 · Cited by 4
Full-reference image quality assessment-based B-mode ultrasound image similarity measurePreprint
Xu, K., Liu, X., Cai, H., & Gao, Z. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1701.02797 · Cited by 5
“ (8) In the above expressions, i is an index to the”
Contrast and visual saliency similarity induced index for image quality assessment.Preprint
Jia, H., & Wang, T. (2017) · arXiv (Cornell University) · Cited by 4
PAD-Net: A Perception-Aided Single Image Dehazing NetworkPreprint
Liu, Y., & Zhao, G. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.03146 · Cited by 4
“Therefore, in this work, inspired by [13], we propose to use loss functions that match with human perception (e.g., SSIM [12], MS-SSIM [14]) as training objectives of a dehazing neural network developed based on the AOD-Net [10].”
MethodInfluentialA Robust Background Initialization Algorithm with Superpixel Motion DetectionPreprint
Xu, Z., Min, B., & Cheung, R. C. C. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.06737 · Cited by 4
Cai, J., Cao, Z., & Zhang, L. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.03470 · Cited by 4
Comparative evaluations of image encryption algorithms
Liu, Z. (2018) · Tuwhera (Auckland University of Technology) · Cited by 7
Dual Generator Generative Adversarial Networks for Multi-Domain Image-to-Image TranslationPreprint
Tang, H., Xu, D., Wang, W., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1901.04604 · Cited by 4
“To make the model benefit from multi-scale deep information, we refer to a multi-scale implementation of SSIM [45] which constrains SSIM over scales.”
Deep Inverse Tone Mapping Using LDR Based Learning for Estimating HDR Images with Absolute LuminancePreprint
Kinoshita, Y., & Kiya, H. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.01277 · Cited by 4
Streaming Video over HTTP with Consistent QualityPreprint
Li, Z., Begen, A. C., Gahm, J., et al. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1401.5174 · Cited by 4
“We define the optimization objective to be an alpha-fairness utility function [11, 18] of some quality metric (for example, MSE/PSNR [1], Sarnoff/PQR [4], MS-SSIM [20, 21], VQM [16], STRRED [17]), which is generic enough to cover a range of candidates.”
MethodEnhancing high-content imaging for studying microtubule networks at large-scalePreprint
Lee, H.-C., Cherng, S. T., Miotto, R., & Dudley, J. T. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.00662 · Cited by 6
“Structural similarity index (SSIM, Wang et al. (2004)) and the multi-scale structural similarity index (MSSSIM, Wang et al. (2003)).”
MethodEffective Data Augmentation with Multi-Domain Learning GANsPreprint
Yamaguchi, S., Kanai, S., & Eda, T. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.11597 · Cited by 4
LC-GAN: Image-to-image Translation Based on Generative Adversarial Network for Endoscopic ImagesPreprint
Lin, S., Qin, F., Li, Y., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.04949 · Cited by 4
“Similar to [36], [37], we propose a multi-scale method that compares the image structural information at different resolutions.”
MethodWang, H., Wang, G., Li, Y., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.04199 · Cited by 4
“We can see that at the same magnitude, the adversarial example generated under the supervision of MS-SSIM are much better than those generated under the supervision of SSIM and without any supervision.”
MethodInfluentialCan Un-trained Neural Networks Compete with Trained Neural Networks at Image Reconstruction?Preprint
Darestani, M. Z., & Heckel, R. (2020) · arXiv (Cornell University) · Cited by 5
Neural Video Coding using Multiscale Motion Compensation and Spatiotemporal Context ModelPreprint
Liu, H., Lu, M., Ma, Z., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.04574 · Cited by 4
“The NVC demonstrated the leading performance with consistent gains across all popular test sequences for both PSNR (Peak signal-to-noise ratio) and MS-SSIM (multiscale structural similarity) [6] distortion metrics.”
BackgroundFew-shot Scene-adaptive Anomaly DetectionPreprint
Lu, Y., Yu, F., Reddy, M. K. K., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.07843 · Cited by 4
“We define L(·) by combine the least absolute deviation (L1 loss) [28], multi-scale structural similarity measurement (Lssm loss) [39] and gradient difference (Lgdl loss) [21]:”
MethodConditional Entropy Coding for Efficient Video CompressionPreprint
Liu, J., Wang, S., Ma, W.-C., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2008.09180 · Cited by 4
“Moreover we plot the rate-distortion curve for multiscale structural similarity (MS-SSIM) [35], which is a commonly-used perceptual metric that captures the overall structural similarity in the reconstruction.”
MethodBeauferris, Y., Teuwen, J., Karkalousos, D., Moriakov, N., Caan, M. W. A., Rodrigues, L., Lopes, A., Pedrini, H., Rittner, L., Dannecker, M., et al. (2020) · arXiv (Cornell University) · Cited by 5
Intrinsic Decomposition of Document Images In-the-WildPreprint
Das, S., Sial, H. A., Ma, K., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2011.14447 · Cited by 4
“The quantitative results for image quality before and after applying our approach is reported in table 1, in terms of the the unwarping metrics: multi-scale structural similarity (MS-SSIM) [45] and Local Distortion (LD) [49].”
Self-Supervised Representation Learning for RGB-D Salient Object Detection.Preprint
Zhao, X., Pang, Y., Zhang, L., Lu, H., & Ruan, X. (2021) · arXiv (Cornell University) · Cited by 4
Space-Time Video Regularity and Visual Fidelity: Compression, Resolution and Frame Rate AdaptationPreprint
Lee, D. Y., Ko, H., Kim, J.-H., & Bovik, A. C. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.16771 · Cited by 7
“We compared the performances of nine relevant popular Reference video quality models, including PSNR, SSIM [7], MSSSIM [17], VIF [18], ST-RRED [26], SpEED [27], FRQM [32], VMAF [8], and the new model, VSTR. VMAF and VSTR are learning based models that each combines 6 and 8 spatio-temporal features,…”
MethodInfluentialSelf-citationNTIRE 2021 Challenge on Quality Enhancement of Compressed Video: Methods and ResultsPreprint
Yang, R., Timofte, R., Liu, J., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.10781 · Cited by 4
“They improve the average PSNR and MS-SSIM by 1.98 dB and 0.0257, respectively.”
MethodInfluentialMulti-Source Fusion and Automatic Predictor Selection for Zero-Shot Video Object SegmentationPreprint
Zhao, X., Pang, Y., Yang, J., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.05076 · Cited by 4
“For depth stream, we follow the related works [23, 31, 34] to adopt the combination loss of L1 and SSIM [49].”
Deep Learning for Improving Numerical Weather Prediction of Rainfall ExtremesPreprint
Hess, P., & Boers, N. (2021) · doi:10.1002/essoar.10507827.1 · Cited by 4
SurroundNet: Towards Effective Low-Light Image EnhancementPreprint
Zhou, F., Sun, X., Dong, J., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.05098 · Cited by 4
“Zhao et al. [52] investigate various loss functions in the image restoration task and suggests the combination of L1 loss and MS-SSIM loss [45].”
BackgroundHaze Level Evaluation Using Dark and Bright Channel Prior InformationPreprint
Chu, Y., Chen, F., Fu, H. C., et al. (2022) · SSRN Electronic Journal · doi:10.2139/ssrn.4004967 · Cited by 4
DeepSR: A Deep Learning Tool for Image Super Resolution
Temiz, H. (2022) · SoftwareX · doi:10.1016/j.softx.2022.101261 · Cited by 4
Micro𝕊plit: Semantic Unmixing of Fluorescent Microscopy DataPreprint
Ashesh, A., Carrara, F., Zubarev, I., Galinova, V., Croft, M., Pezzotti, M., Gong, D., Casagrande, F., Colombo, E., Giussani, S., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.02.10.637323 · Cited by 4
Neural Activity Shaping in Visual Prostheses with Deep Learning
Castro, D., Grayden, D. B., Meffin, H., & Spencer, M. (2023) · doi:10.1101/2023.12.20.572123 · Cited by 4
Discrete Representation Learning for Modeling Imaging-based Spatial Transcriptomics Data
Yarlagadda, D. V. K., Massagué, J., & Leslie, C. S. (2023) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw60793.2023.00416 · Cited by 4
“To learn structure awareness, i.e. to encode the structure of the input sample such as density and homogeneity within microenvironment, we additionally use multi-scale spectral similarity index (MSSIM) [45, 46] loss in our model, which is defined as follows: where μ x , μ ˆ x , σ x , σ ˆ x and σ x ˆ x are the local mea…”
MethodA Tri-Dynamic Preprocessing Framework for UGC Video Compression
Zhao, F., Guo, M., Zhao, S., Li, J., Zhang, L., & Xie, X. (2024) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp48485.2024.10445837 · Cited by 4
“In our experiments, we adopt MS-SSIM loss [15] as the measurement of distortion D and use the learnable factorized prior entropy model of Balle et al. [16, 17] to calculate R .”
Baek, W., Kang, S., & Yang, Y.-H. (2024) · IEEE Access · doi:10.1109/access.2024.3432607 · Cited by 4
Color quantization using an accelerated Jancey k-means clustering algorithm
Bounds, H., Celebi, M. E., & Maxwell, J. (2024) · Journal of Electronic Imaging · doi:10.1117/1.jei.33.5.053052 · Cited by 5
“MS-SSIM, 45 on the other hand, is a multi-scale variant of SSIM that incorporates variations in image resolution and viewing conditions.”
Chan, C.-Y., Siu, W.-C., Chan, Y., & Chan, H. A. (2024) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2024.3516366 · Cited by 3
Learned Focused Plenoptic Image Compression With Local-Global Correlation Learning
Liu, G., Yue, H., Wen, B., et al. (2025) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2024.3521815 · Cited by 3
“Quality evaluation of the decoded FPIs is conducted using PSNR and MS-SSIM [58] metrics in RGB colorspace.”
Pathological Section Staining Transferring with Tailored Metric-based Model Selection
Ji, Y., Zhu, S., Zhang, D., & Li, S. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10889542 · Cited by 3
“For medical slice translation, besides expert evaluation, several criteria are used for final model selection, including SSIM[17], MS-SSIM[18], and FID[8].”
Sippel, F., Seiler, J., & Kaup, A. (2025) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3556602 · Cited by 8
“The l 1 - loss in occluded areas as well as the multiscale structural similarity index [50] between the network output N v [ x , y ] and the ground-truth images ˆ M v [ x , y ] are used to train the network, where MS − SSIM( · ) calculates the multiscale structural similarity index.”
An End-to-End Spatially Scalable Light Field Image Compression Method
Lei, J., Li, H., Peng, B., Zhao, B., & Ling, N. (2025) · IEEE Transactions on Broadcasting · doi:10.1109/tbc.2025.3553295 · Cited by 2
“Meanwhile, to conduct further validation, the comparison results of all layers are also provided by calculating the PSNR and multi-scale structural similarity (MS-SSIM) [43] between original lenslet images and reconstructed lenslet images.”
Comparative evaluation of image registration techniques in functional ultrasound imaging
Zhong, S., Agyeman, K. A., Syed, S., Tobing, R., Choi, W., Liu, C. Y., Lee, D. J., & Christopoulos, V. (2025) · Imaging Neuroscience · doi:10.1162/imag.a.47 · Cited by 3
“The MS-SSIM is a similarity measure that extends the classic SSIM by evaluating structural similarity across multiple image resolutions ( Z. Wang et al., 2003).”
Improved pix2pix Method for Clutter Removal From Complex Real Ground-Penetrating Radar Images
Zhao, Y., Chai, Z., Xu, Z., et al. (2025) · IEEE Sensors Journal · doi:10.1109/jsen.2025.3575421 · Cited by 3
“This article utilizes a composite loss function that combines adversarial loss, L 1 loss, and multiscale structural similarity (MS-SSIM) loss [40] to drive the improved pix2pix network.”
InfluentialRevisit Point Cloud Quality Assessment: Current Advances and a Multiscale-Inspired Approach
Zhang, J., Chen, T., Ding, D., & Ma, Z. (2025) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2025.3582309 · Cited by 3
“Typical examples include PSNR, SSIM [37], and MS-SSIM [38], which have been well devised by comparing pixel values, structural similarity, and multiscale structural similarity, respectively, be-tween the reference and the distorted images.”
Song, Q., Liu, Z., Zhong, Y.-P., et al. (2025) · Infrared Physics & Technology · doi:10.1016/j.infrared.2025.106052 · Cited by 2
Zhang, Y., Zhang, S., Li, N., et al. (2025) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2025.3595542 · Cited by 3
“…predicted QF / Q and the corresponding image with several quality metrics, including QF / Q, PSNR, Multi-Scale Structural Similarity (MS-SSIM) [46], Feature Similarity Index Measurement (FSIM) [47], Gradient Magnitude Similarity Deviation (GMSD) [48], Visual Saliency Index (VSI) [49], and…”
StarINN: An efficient invertible neural network for image steganography
Shi, W., Wang, Z., Zhang, X., & Feng, G. (2025) · Displays · doi:10.1016/j.displa.2025.103175 · Cited by 11
Self-citationSCSC: The Super Compressed Semantic Communication Transmission Method for Images and Video Frames
Guo, J., Sun, B., Zhu, J., Feng, W., He, S., & Xue, W. (2025) · IEEE Internet of Things Journal · doi:10.1109/jiot.2025.3598074 · Cited by 4
“For MS-SSIM [27], it evaluates the structural similarity between two images across multiple scales.”
Rate-Distortion Optimization with Non-Reference Metrics for UGC Compression
Fernández-Menduiña, S., Xiong, X., Pavéz, E., Ortega, A., Birkbeck, N., & Adsumilli, B. (2025) · Proceedings - International Conference on Image Processing · doi:10.1109/icip55913.2025.11084420 · Cited by 3
“1 shows that FRMs like the SSE and This work was funded in part by a gift from YouTube. multi-scale structural similarity index measure (MS-SSIM) [5] approach zero distortion–i.e., perfect quality–at high bitrates, whereas NRMs, such as video score feature aggregation (VSFA), plateau at the input…”
A comprehensive features representation for no-reference image quality assessment
Wu, X., & Shi, Z. (2025) · Multimedia Systems · doi:10.1007/s00530-025-02016-9 · Cited by 3
“The typical methods of FR-IQA include Peak Signal-to-Noise Ratio (PSNR) [3 ], Structural Similarity Index (SSIM) [ 4], Multi-Scale Structural Similarity Index (MS-SSIM) [5 ], Visual Information Fidelity (VIF) [ 6 ] and Feature Similarity (FSIM) [ 7].”
A Multidimensional Measurement of Photorealistic Avatars Quality of Experience
Cutler, R., Naderi, B., Gopal, V., & Palle, D. R. (2025) · Proceedings of the ACM on Human-Computer Interaction · doi:10.1145/3757542 · Cited by 6
“P.911 is a counterpart of P.910 but for audiovisual signals.”
Despeckling of Synthetic Aperture Radar Images Using Linear–Angular Attention Transformer
Kundu, S., Bhuyan, M. K., MacDorman, K. F., Sharma, N. K., & Bhatt, M. (2025) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2025.3628887 · Cited by 3
“The tested losses are L 1 loss with multiscale SSIM loss ( L SSIM ) [33], total variation loss ( L TV ) [34], gradient di ff erence loss ( L GD ) [35], loss with a linear combination of PSNR and SSIM called PSNR-SSIM loss ( L PSNR SSIM ), and L 1 loss.”
Diffusion Models for Neuroimaging Data Augmentation: Assessing Realism and Clinical Relevance
Mallardi, G., Calefato, F., Lanubile, F., Logroscino, G., & Tafuri, B. (2025) · Journal of Medical Systems · doi:10.1007/s10916-025-02300-1 · Cited by 6
Hwang, C., Yang, S., Kim, Y., Park, S., & Choi, H. (2025) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.70142 · Cited by 6
“By combining SSIM values calculated at multiple scales, MS-SSIM provides a more comprehensive assessment of both global image quality and local details (Moorthy & Bovik, 2009; Z. Wang et al., 2003).”
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.