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 4,701–4,800 of 6,047
Textural Detail Preservation Network for Video Frame Interpolation
Yoon, K., Huh, J., Kim, Y. H., Kim, S., & Jeong, J. (2023) · IEEE Access · doi:10.1109/access.2023.3294964 · Cited by 3
“To quantify perceptual similarity between the restored and original images, metrics such as SSIM [57], MSSIM [58], and FSIM [59] have been proposed.”
MethodLight Field Image Quality Assessment Based on Disentangling Bayesian Theory
Zhang, X., Ma, J., Chengjin, Ли, Ж., & Wang, J. (2023) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas46773.2023.10182128 · Cited by 1
“For instance, structural similarity (SSIM) [2], and multiscale SSIM (MS-SSIM) [3], feature similarity (FSIM) [4], information content weighting SSIM (IWSSIM) [5], information fidelity criterion (IFC) [6] and visual information fidelity (VIF) [7].”
BackgroundMethodRepVGGFuse: an approach for infrared and visible image fusion network based on RepVGG architecture
Xiong, Z., Zhang, X., Hu, Q., et al. (2023) · International Conference on Computing, Networks and Internet of Things (CNIOT) · doi:10.1145/3603781.3603847 · Cited by 1
“They are mutual information(MI) [16], fast mutual information (FMI) [17], peak signal-to-noise ratio(PSNR) [18], structural similarity index measure(SSIM) [19], multiscale structural similarity for image quality assessment (MSSSIM) [20], the sum of the correlations of differences(SCD) [21].”
BackgroundResidual Hybrid Attention Network for Single Image Dehazing
Li, H. M. (2023) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn54540.2023.10191742 · Cited by 1
“Inspired by [8], our loss function consists of three parts: MS-SSIM loss [43], smooth L1 loss [44] and perceptual loss [45].”
MethodZhang, X., & Li, S. (2023) · International Conference on Electronics, Communications and Control Engineering (ICECC) · doi:10.1145/3592307.3592313 · Cited by 1
“FR (or RR) methods often need to use all (or part) of the reference videos, such as some classic FR methods including Peak Signal-to-Noise Ratio (PSNR) methods [2], Structural Similarity (SSIM) index methods [3] and Multi-Scale SSIM (MS-SSIM) [4].”
MethodVariable-Rate Neural Image Compression with Joint Content-Channel Features and Accurate R-λ Model
Shi, Y., Wang, P., Wang, J., et al. (2023) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme55011.2023.00126 · Cited by 1
Rate-Controllable and Target-Dependent JPEG-Based Image Compression Using Feature Modulation
Jeong, S., Jeon, K. E., & Ko, J. H. (2023) · IEEE International Conference on Multimedia and Expo workshops · doi:10.1109/icmew59549.2023.00035 · Cited by 1
Decao, 马. M., Yong, 鲜. X., Juan, 苏. S., et al. (2023) · ACTA PHOTONICA SINICA · doi:10.3788/gzxb20235204.0410003 · Cited by 1
NUCQ: Non-Uniform Conditional Quantization for Learned Image Compression
Ge, Z., Jia, C., Ma, S., et al. (2023) · Proceedings - International Conference on Image Processing · doi:10.1109/icip49359.2023.10222198 · Cited by 1
“(1) be 117, 228, 436, 845, 1626, 3141 respectively (same to the official setting in CompressAI) for MSE-opt models and 2.40, 4.58, 8.73, 16.64, 31.73, 60.50 respectively for MS-SSIM [21] models.”
InfluentialSpatially-Adaptive Learning-Based Image Compression with Hierarchical Multi-Scale Latent Spaces
Brand, F., Kopte, A., Fischer, K., et al. (2023) · Proceedings - International Conference on Image Processing · doi:10.1109/icip49359.2023.10222756 · Cited by 1
“To measure the distortion, we use PSNR and MS-SSIM [21], as implemented in the pytorch msssim library.”
MethodLearned Structure-Based Hybrid Framework for Martian Image Compression
Li, S., Sun, X., Xu, M., Jiang, L. W., & Zou, X. (2023) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2023.3327687 · Cited by 1
“…by where R ( x d ) represents the bit-rates of compressing x d , λ is the hyperparameter that adjusts the bit-rates, and D ( x , ˆ x ) denotes the discrepancy between x and ˆ x , which is chosen by mean squared error (MSE) or multiscale structural similarity (MS-SSIM) [20] in this letter.”
BackgroundQuantum Steganography: Hiding Secret Messages in Images using Quantum Circuits and SIFT
Azooz, H. J., Salah, K. B., Kherallah, M., & Naceur, M. S. (2023) · International Journal of Advanced Computer Science and Applications · doi:10.14569/ijacsa.2023.01410107 · Cited by 1
“The formula for determining the SSIM is as follows [17](6): x + σ 2 y + c 2 ) c represents the cover image, and s represents the stego image. µ c and µ s are the average pixel values of images c and s , respectively.”
BackgroundAttention-Guided Motion Estimation for Video Compression
Zhang, S., & Duan, P. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-46305-1_12 · Cited by 1
Neyigapula, B. S. (2023) · International Journal of Artificial Intelligence and Machine Learning · doi:10.51483/ijaiml.3.2.2023.45-56 · Cited by 3
CPIPS: Learning to Preserve Perceptual Distances in End-to-End Image Compression
Huang, C.-H., & Wu, J. (2023) · Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) · doi:10.1109/apsipaasc58517.2023.10317329 · Cited by 1
“addition, to mean square error (MSE) or PSNR, various FRIQA metrics, such as SSIM variants [36], [37], PIM [25], and DISTS [27], have been proposed to predict subjective image quality judgments.”
BackgroundA software test bed for sharing and evaluating color transfer algorithms for images and 3D objects
Potechius, H., Raja, G., Sikora, T., & Knorr, S. B. (2023) · Conference on Visual Media Production · doi:10.1145/3626495.3626509 · Cited by 1
“Multiple improvements were made, like the Multi-Scale SSIM (MS-SSIM) [Wang et al. 2003], which considers the image details in multiple resolutions and was used in [Yao et al. 2016], Gradient-based SSIM (GSSIM) [Chen et al. 2006] used in [Li et al. 2019; Xiao et al. 2013] which established gradient…”
MethodAccelerating spiral deblurring with square kernels and low‐pass preconditioning
Wang, D., Chao, T. C., & Pipe, J. G. (2023) · Magnetic Resonance in Medicine · doi:10.1002/mrm.29928 · Cited by 1
M-DVC: Moderate Deep Video Compression Embracing the Metaverse New Era
Li, Z., & Liu, Q. (2023) · IEEE International Workshop on Multimedia Signal Processing (MMSP) · doi:10.1109/mmsp59012.2023.10337722 · Cited by 1
“We employ two evaluation metrics, namely PSNR and MS-SSIM [27], to assess the distortion of the reconstructed frames.”
MethodBlind visual quality assessment for super-resolution images: database and model
Zheng, R., Jiang, X., Zhang, J., & Yu, H. (2023) · Multimedia Tools and Applications · doi:10.1007/s11042-023-17026-w · Cited by 1
“Among them, the following typical FR-IQA metrics are included: MSE-based metric (PSNR), multi-scale structural similarity index MS-SSIM [51], information fidelity criterion (IFC) [52], the fusion of phase congruency and gradient magnitude method (FSIM [7]), visual saliency-induced index (VSI [53]),…”
Generation of High-Quality Realistic Faces with StyleGAN
Ashwini, K., Nagajyothi, D., Ramakrishna, C., & Jyothi, V. (2023) · IEEE Global Conference for Advancement in Technology (GCAT) · doi:10.1109/gcat59970.2023.10353468 · Cited by 1
Motion-Based Convolutional Neural Networks for Super-Resolution from Compressed Videos
Chang, H.-Y., Hsieh, H.-Y., & Lin, Y.-C. (2023) · IEEE International Conference on Consumer Electronics - Berlin (ICCE-Berlin) · doi:10.1109/icce-berlin58801.2023.10375648 · Cited by 1
Nelson, J., Chen, X., Li, Y., Omoumi, F. H., Wong, M. D., & Liu, H. (2024) · Biophotonics and Immune Responses · doi:10.1117/12.3004490 · Cited by 1
“The SSIM metric is formally defined as: In our implementation, we leveraged a more advanced variant of SSIM known as Multiscale Structural Similarity (MS-SSIM) [11].”
MethodResultExploring synthesizing 2D mammograms from 3D digital breast tomosynthesis images
Chłędowski, J., Park, J., & Geras, K. J. (2023) · International Conference on Digital Image Computing: Techniques and Applications (DICTA) · doi:10.1109/dicta60407.2023.00083 · Cited by 1
“The Multi-Scale Structural Similarity (MS-SSIM) index [32] is a method for measuring the structural similarity between two images.”
MethodAn Efficient ConvNet for Learned Image Compression with Transformer-Style Architecture
Ruan, H., Wang, F. R., & Wang, Y. (2023) · IEEE International Conference on Visual Communications and Image Processing (VCIP) · doi:10.1109/vcip59821.2023.10402780 · Cited by 1
“• The proposed method gets competitive results by comparing existing learning-based methods and traditional codecs in terms of both PSNR [20] and MS-SSIM [21] quality metrics, and also gets visually pleasant results.”
MethodDegradation Aware Multi-Scale Approach to No Reference Image Quality Assessment
Roheda, S., Unde, A. S., Rashid, L., & Ameta, A. (2023) · Indian Conference on Computer Vision, Graphics and Image Processing · doi:10.1145/3627631.3627644 · Cited by 1
BPG-based compression of Poisson noisy images
Коваленко, Б., & Лукин, В. В. (2023) · International Conference on Dependable Systems, Services and Technologies (DESSERT) · doi:10.1109/dessert61349.2023.10416544 · Cited by 3
Ground-Penetrating Radar Random Missing Data Reconstruction Method Based on U-Net Network
Liu, W., Zhang, Q., & Sun, F.-Y. (2023) · Cross Strait Radio Science and Wireless Technology Conference (CSRSWTC) · doi:10.1109/csrswtc60855.2023.10426946 · Cited by 1
“Multi-scale structural similarity (MS_SSIM) is proposed by Zhou on the basis of SSIM, which introduces multi-scale to SSIM to better complement the detail information of the image [9].”
MethodMultiscale Augmented Normalizing Flows for Image Compression
Windsheimer, M., Brand, F., & Kaup, A. (2024) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp48485.2024.10446147 · Cited by 1
Efficient Joint Rectification of Photometric and Geometric Distortions in Document Images
Tang, H., Guo, J., Wang, T., Yu, Y., & Wang, C. (2024) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp48485.2024.10447446 · Cited by 3
“We use MS-SSIM[18] and LD[19] as quantitative evaluation metrics, where MS-SSIM measures the global similarity between the result and ground truth, while LD focuses more on the local details.”
MethodDeep Learning Empowers Next-Gen Video Compression: Bridging Quality and Efficiency
Khadir, M., Hashmi, M. F., & Khadir, M. (2023) · International Arab Conference on Information Technology (ACIT) · doi:10.1109/acit58888.2023.10453895 · Cited by 1
“Some of the most popular algorithms for picture compression are the Joint Photographic Experts Group's (JPEG) JPEG [7] and JPEG2000 [8] formats.”
MethodHierarchical Domain Prior CycleGAN: Translated Satellites ISAR Image from the Optical Domain
Liao, H., Xia, J., Yang, Z., Liu, Z., & Liu, Y. (2023) · China Automation Congress (CAC) · doi:10.1109/cac59555.2023.10450872 · Cited by 1
“The Inception Score (IS) [18] and Multi-Scale-Structural Similarity Index (MS-SSIM) [19] are applied to evaluate the obtained ISAR images.”
MethodHuman-oriented video retargeting via object detection and patch decision
Kim, D.-H., Lee, S., Bae, J., et al. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18878-6 · Cited by 1
“For an assessment of the quality, we have presented full-reference [28, 29, 32] and no-reference [15, 17, 26, 34, 35] assessments in Table 1, 2, 3, and 4.”
InfluentialTAMPAR: Visual Tampering Detection for Parcel Logistics in Postal Supply Chains
Naumann, A., Hertlein, F., Dörr, L., et al. (2024) · IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) · doi:10.1109/wacv57701.2024.00789 · Cited by 1
“…mean alignment As image similarity metrics, we consider • Learned Perceptual Image Patch Similarity (LPIPS) [35], • Structural Similarity (SSIM) [36], • Multiscale Structural Similarity (MS-SSIM) [37], • Complex Wavelet Structural Similarity (CW-SSIM) [38], • HOG [15] feature similarity 1”
BackgroundSparse-View Artifact Correction of High-Pixel-Number Synchrotron Radiation CT
Huang, M., Li, G., Sun, R., Zhang, J., Wang, Z., Wang, Y., Deng, T., & Yu, B. (2024) · Applied Sciences · doi:10.3390/app14083397 · Cited by 1
Self-citationME: Multi-Task Edge-Enhanced for Image Forgery Localization
Hao, X., Shao, L., He, X. D., et al. (2024) · International Conference on Image and Graphics Processing (ICIGP) · doi:10.1145/3647649.3647701 · Cited by 1
Leveraging Conv-Attention for Efficient and High-Quality JPEG AI Image Coding
Wang, M., Esenlik, S., Zhang, Z., et al. (2024) · DCC · doi:10.1109/dcc58796.2024.00012 · Cited by 1
“…to measure the coding distortion of the reconstructed image for each coding rate point, including Multi-Scale Structural SIMilarity (MS-SSIM) [8], Information Content Weighted Structural Similarity Table 3: Illustration of the rate-distortion performance and decoding complexity with five…”
MethodPanchal, H. D., & Shah, H. B. (2023) · International Conference on Image Information Processing (ICIIP) · doi:10.1109/iciip61524.2023.10537677 · Cited by 1
“The Structural Similarity Index Measurement (SSIM) is a metric used to evaluate the similarity between two frames [22].”
MethodGheisari, M., & Genovesio, A. (2024) · Biological Imaging · doi:10.1017/s2633903x24000084 · Cited by 1
Adapting Pretrained Networks for Image Quality Assessment on High Dynamic Range Displays
Chubarau, A., Yoo, H., Akhavan, T., & Clark, J. J. (2024) · Electronic Imaging · doi:10.2352/ei.2024.36.11.hvei-224 · Cited by 1
“The comparison can be based on error visibility [18, 19], structural similarity [20, 21, 22], information content [23, 24], contrast visibility [25, 11], or various other feature similarities inspired by the Human Visual System (HVS) [26, 27, 28, 29, 30] and optionally modulated by visual saliency…”
BackgroundKumar, V., Sharma, M. K., Jehadeesan, R., Venkatraman, B., & Sheet, D. (2024) · International Journal of Imaging Systems and Technology · doi:10.1002/ima.23113 · Cited by 2
Assessing objective quality metrics for JPEG and MPEG point cloud coding
Lazzarotto, D. N., Testolina, M., & Ebrahimi, T. (2024) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex61742.2024.10598277 · Cited by 1
“Other predictors are based on the Structural Similarity Index Measure (SSIM) [19], including different variations such as the Multi-Scale Structural Similarity Index (MS-SSIM) [20] or the Information Content Weighted Structural Similarity Measure (IW-SSIM) [21].”
MethodMiao, Y., Xiao, S., Pan, L., Zhang, L., & Yang, Z. (2024) · Signal Image and Video Processing · doi:10.1007/s11760-024-03444-2 · Cited by 2
“…image distortion metrics, such as Peak Signal-to-Noise Ratio (PSNR) [30], Structural Similarity Index (SSIM) [31], and Visual Information Fidelity (VIF) [32], are commonly used directly or through Multiscale Structural Similarity (MS-SSIM) [33] to assess the quality of projected 2D images.”
89‐2: An Evaluation Index for See‐Through Image Quality on Transparent MicroLED Displays
Tsai, Y., Wu, J., Huang, M., et al. (2024) · SID Symposium Digest of Technical Papers · doi:10.1002/sdtp.17766 · Cited by 1
Self-normalization for a 1 mm3 resolution clinical PET system using deep learning
Chin, M., Jafaritadi, M., Franco, A. B., Ullah, M. N., Chinn, G., Innes, D. R., & Levin, C. S. (2024) · Physics in Medicine and Biology · doi:10.1088/1361-6560/ad69fb · Cited by 1
“These values are set to C 1 = 1 × 10 − 4 , C 2 = 3 × 10 − 4 , and C 3 = C 2 / 2 as suggested by Wang et al (2003) The reformulated equivalent equation can be expressed as (10) We rescaled pixel values to 0-255 before calculating the PSNR and SSIM metrics.”
A New HDR Video Reconstruction Benchmark, Dataset and Metric
Voronin, M., Safonov, N., & Vatolin, D. S. (2024) · doi:10.1145/3653876.3653896 · Cited by 1
“Consequently, we opted to use metrics designed for SDR content: PSNR, SSIM [27], MS-SSIM [28], NIQE [18], and VMAF 3 .”
Super-Resolution of Multiple Sentinel-2 Images Using Composite Loss Function
Liu, S., Fonseca, J., & Mora, A. (2024) · International Young Engineers Forum on Electrical and Computer Engineering (YEF-ECE) · doi:10.1109/yef-ece62614.2024.10625336 · Cited by 2
“…[3], a composite function established using (cid:4) (cid:5) loss and mean gradient error, and the combination of loss with the multiscale structural similarity index (MS-SSIM) [5] based on structural similarity index (SSIM) [6] as proposed by Wang et al. [4], both achieved excellent results.”
MethodAtencia, J. R., López-Granado, O., Malumbres, M. P., Martínez-Rach, M., Ruiz, D., Fernández-Escribano, G., & Wallendael, G. V. (2024) · Electronics · doi:10.3390/electronics13163341 · Cited by 1
Deep Generative Modeling for Neural Compression
Tomczak, J. M. (2024) · Springer eBooks · doi:10.1007/978-3-031-64087-2_10 · Cited by 2
Deep Generative Model based Rate-Distortion for Image Downscaling Assessment
Liang, Y., Garg, B., Rosin, P. L., & Qin, Y. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01832 · Cited by 5
“For example, Mittal et al. [27] propose BRISQUE, an NR-IQA metric that uses the natural scene statistics (NSS) to quantify loss of “naturalness” in distorted images.”
BackgroundXu, J., Hong, D., Su, S., et al. (2024) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2024.3460942 · Cited by 1
“…six objec-tive image fusion evaluation metrics, including information entropy (EN) [39], visual information fidelity (VIF) [40], edge retentiveness ( Q abf ) [41], the sum of the correlations of differences (SCDs) [42], the multiscale structural similarity (MS-SSIM) [43], and FMI pixel [44].”
BackgroundA Feasibility Study on Novel View Synthesis of Underwater Structures using Neural Radiance Fields
Too, Y. M., Vishnu, H., Chitre, M., et al. (2024) · doi:10.1109/oceans51537.2024.10682127 · Cited by 1
“For quantitative analysis, we compared the camera images with the synthesized images based on PSNR, MS-SSIM [12], and LPIPS [13] as summarized in Table I.”
MethodRedefining Visual Quality: The Impact of Loss Functions on INR-Based Image Compression
Catania, L., & Allegra, D. (2024) · Proceedings - International Conference on Image Processing · doi:10.1109/icip51287.2024.10647328 · Cited by 1
Mobile Aware Denoiser Network (MADNet) for Quad Bayer Images
Madhusudana, P. C., Li, J., Nadir, Z., Sheikh, H. R., & Lee, S.-J. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00009 · Cited by 1
Neural Radiance Field-Assisted Static-Scene Video Coding
Yang, R., Liu, D., Wu, F., et al. (2024) · Proceedings - International Conference on Image Processing · doi:10.1109/icip51287.2024.10647411 · Cited by 1
Utilizing a Two Planes Model to Rectify Documents With a Single Arbitrary Crease
Ershov, A., Tropin, D. V., Kazimirov, D. D., et al. (2024) · IEEE Access · doi:10.1109/access.2024.3474099 · Cited by 1
Annotation Pipeline for Railway Track Segmentation
Sagastiberri, I., Larrazabal, M., Sánchez, M., Aranjuelo, N., Nieto, M., & de Eribe, D. O. (2024) · IEEE International Conference on Intelligent Systems · doi:10.1109/is61756.2024.10705179 · Cited by 1
“Two metrics are used to compute the QA: • MS-SSIM (Multi-Scale Structural Similarity) [34]: this metric is an extension of the Structural Similarity (SSIM) index.”
MethodComputationally-Efficient Neural Image Compression Using Asymmetric Autoencoders
Augusto, L., da Silveira, T. L. T., & Grellert, M. (2024) · doi:10.1109/sbcci62366.2024.10704000 · Cited by 1
“…networks, we adopt the methodology used in [10], which measures processing time, compression measured in BPP, and image quality using PSNR, MS-SSIM [19], and the more recent Learned Perceptual Image Patch Similarity (LPIPS) [20], which uses learning-based methods and correlates well with human…”
MethodGene-to-Image: Decoding Brain Images from Genetics via Latent Diffusion Models
Jeon, S., Song, Y., & Kim, W. H. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-74561-4_5 · Cited by 1
A TransISP Based Image Enhancement Method for Visual Disbalance in Low‐light Images
Wu, J., Guo, J., Jing, R., Zhang, S., Tian, Z., Chen, W., & Wang, Z. (2024) · Computer Graphics Forum · doi:10.1111/cgf.15209 · Cited by 1
Self-citationZhang, W., Wang, Q., Huang, K., Huang, X., Guo, F., & Gu, X. (2024) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3664647.3681548 · Cited by 4
G-SemTMO: Tone Mapping With a Trainable Semantic Graph
Goswami, A., Bernard, E., Singh, A. R., et al. (2024) · IEEE Access · doi:10.1109/access.2024.3491494 · Cited by 1
“To assess the tone mapping results, we use 6 objective metrics: PSNR, Multi-scale Structural Similarity Index (MS-SSIM) [42], Visual Difference Predictor for HDR images HDR-VDP-3 [43], hybrid perceptual colour distance metrics HyAB [37] and CIEDE2000 ( ∆ E 00 ) [44] and Colourfulness-based contrast…”
Overcomplete U-Net Networks for Psoriasis Detection in Digital Color Images
Kovvuru, A. K., Londhe, N. D., Raj, R., et al. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-78312-8_10 · Cited by 1
Ma, Y., Yi, C., Zhou, Y., et al. (2024) · doi:10.1117/12.3047797 · Cited by 1
Subjective and Objective VQA of Video Codecs for UHD Video Streaming
Valiandi, I., Mahmoud, M., Rizou, S., & Panayides, A. S. (2024) · International Conference on Frontiers of Signal Processing (ICFSP) · doi:10.1109/icfsp62546.2024.10785449 · Cited by 3
Dehazing CNNs Loss Functions Analysis
Dobre-Baron, R., & Ancuți, C. (2024) · International Symposium on Electronics and Telecommunications (ISETC) · doi:10.1109/isetc63109.2024.10797441 · Cited by 1
“Multiscale SSIM [23] is SSIM done on multiple scales and sampling frequencies, providing better results than SSIM, having occasionally been used as a loss function [6].”
Dynamic SpectraFormer for Ultra-High-Definition Underwater Image Enhancement
Hu, Z., Yu, T., Huang, S., & Ishikawa, M. (2024) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros58592.2024.10802529 · Cited by 3
“WaterNet[21] FUnIE-GAN[17] U-ShapeTransformer [29] Ucolor[20] UHD[38] Ours The DSWG module, given the input size represented by M Q i freq and weight matrices W 1 ∈ R K × V and W 2 ∈ V × K , the DSWG’s computational complexity is calculated as O ( KV 2 C + K 2 V C ) .”
Wirabudi, A. A., & Choi, H. (2024) · Journal of Broadcast Engineering · doi:10.5909/jbe.2024.29.7.1103 · Cited by 1
An Exploration of How Volume Rendering is Impacted by Lossy Data Reduction
Etchi, Y., Wang, D., Grosset, P., et al. (2024) · doi:10.1109/scw63240.2024.00040 · Cited by 1
“This is reflected in the SSIM metric, a well-known metric for quantifying perceptual image difference [11], [12], that has a score of 0.997 in Figure 1(b) but only 0.937 in Figure 1(a); an SSIM score of 1 indicates that two images compared are identical.”
InfluentialAMDIS: Amplitude Dissimilarity Reduced Reference IQA Metric for Neural Radiance Field
Togo, R., Yanagi, R., Kawai, M., OGAWA, T., & HASEYAMA, M. (2025) · IEICE Transactions on Fundamentals of Electronics Communications and Computer Sciences · doi:10.1587/transfun.2024imp0003 · Cited by 2
RenderGAN: Enhancing Real-time Rendering Efficiency with Deep Learning
Mameli, M., Paolanti, M., Mancini, A., Zingaretti, P., & Pierdicca, R. (2025) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3712263 · Cited by 4
“SSIM, as defined in [57], incorporates the concept of MSSSIM, which applies SSIM on M scales, with scale 1 representing the input image.”
Spatial-Temporal Motion Compensation for Learned Video Compression
Huang, Q., Liu, W., Lü, H., et al. (2024) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc54092.2024.10831688 · Cited by 1
“…used to encode the motion map O t and the context r t , respectively. λ is a hyper parameter used to control the tradeoff between bit rate and distortion. d ( X t , ˆ X t ) represents the distortion between X t and ˆ X t , where d ( · ) represents the mean squared error (MSE) or 1 - MS-SSIM [24].”
Advances in Light Field Spatial Super-Resolution: A Comprehensive Literature Survey
Lyu, W., Sheng, H., Ke, W., et al. (2025) · IEEE Access · doi:10.1109/access.2025.3532610 · Cited by 1
Perceptual QP optimization for VVC with dual hybrid neural networks
Atencia, J. R., López-Granado, O., Malumbres, M. P., et al. (2025) · The Journal of Supercomputing · doi:10.1007/s11227-025-06954-z · Cited by 1
“The MS-SSIM (multi-scale structural similarity) metric [30] is designed to assess structural fidelity by evaluating luminance, contrast, and structural information across multiple scales, aligning with how HVS processes images at various resolutions.”
Foundation Models Boost Low-Level Perceptual Similarity Metrics
Ghildyal, A., Barman, N., & Zadtootaghaj, S. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10889302 · Cited by 8
“PSNR measures direct pixel value differences, while SSIM [15] and its variants such as MS-SSIM [16], and FSIMc [17] leverage the understanding of HVS and measure differences in luminance, contrast, and structure across local and global image regions, enhancing robustness to misalignments.”
Overview of Variable Rate Coding in JPEG AI
Jia, P., Brand, F., Yu, D., et al. (2025) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2025.3552971 · Cited by 1
“To evaluate the performance, JPEG AI computed the average Bjøntegaard Delta rate (BD-Rate) [59] performance across 7 metrics, including MS-SSIM [47], Visual Information Fidelity (VIF) [48], feature similarity (FSIM) [49], Normalized Laplacian Pyramid (NLPD) [50], Information Content Weighted…”
Research on the Intelligent Color Matching Algorithm Integrating Visual Aesthetics
Lin, Y. (2025) · doi:10.1109/citsc64390.2025.00156 · Cited by 1
“Building on this, researchers have successively proposed evaluation models such as Multi-Scale Structural Similarity (MS-SSIM)[6] and Information-Weighted Structural Similarity (IW-SSIM)[7].”
Bezerra, S. A. C., Júnior, S. A. C. B., Pio, J. L. D. S., Carvalho, J. R. H., & Fonseca, K. V. O. (2025) · IEEE Access · doi:10.1109/access.2025.3560918 · Cited by 3
“Meanwhile, its extended version, called multi-scale SSIM (MS-SSIM) [18], produces better predictions than its single-scale counterpart.”
Arias-Castro, E., & Qiao, W. (2025) · The Annals of Statistics · doi:10.1214/24-aos2471 · Cited by 6
AFHNet: Attention-Free Hybrid Network for Salient Object Detection in Underwater Images
Tang, Q., Wang, Z., Wang, X., & Zhang, S. (2025) · IEEE Access · doi:10.1109/access.2025.3567606 · Cited by 2
“During the training process of the AFHNet model, we design a hybrid loss function l total , based on the methodologies presented in [35], [36], [37], and [38], to achieve more accurate depth map predictions.”
Self-citationDynamic Fusion: Merging Structural and Functional Connectivity Dynamics Via Joint CMICA
Wu, L., Duda, M., Iraji, A., & Calhoun, V. D. (2025) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi60581.2025.10980932 · Cited by 1
“We also applied a multiscale 3D similarity measure to match components and assess stability [12].”
Zhao, Q., Ji, L., Su, Y., Yu, K., & Yongchao, Z. (2025) · International Journal of Crowd Science · doi:10.26599/ijcs.2024.9100025 · Cited by 1
Dynamic Fusion of Structural and Functional Connectivity via Joint Connectivity-Matrix ICA
Wu, L., Duda, M., Iraji, A., & Calhoun, V. D. (2026) · Brain Connectivity · doi:10.1177/21580014261479990 · Cited by 1
“In this study, we adapted the 3D multiscale structural similarity index measure (MS-SSIM) (Wang et al., 2003) as the similarity function q for both source matching and stability measure.”
InfluentialOffloading-based Power-Efficient Mobile VTuber Live Streaming
Zhu, Z., Petrangeli, S., Swaminathan, V., et al. (2025) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3742787 · Cited by 1
Kawaoto, R., Namba, T., Ohtsuki, Y., et al. (2025) · Results in Optics · doi:10.1016/j.rio.2025.100865 · Cited by 1
EagleEye : Balancing Latency, Accuracy, and Power on Edge-Assisted UAVs for Urban Crowd Surveillance
Xiang, C., Li, Z., Zhang, Q., Wu, X., Zheng, X., & Guo, Y. (2025) · IEEE Transactions on Mobile Computing · doi:10.1109/tmc.2025.3586860 · Cited by 4
“(cid:2) CFRC [50]: Clipped Fixed-RoI Compressing method first clips the UAV-captured video with structural similarity and then compresses frames with fixed RoIs.”
Wu, Z., Li, W., & Chen, X. (2025) · MTS/IEEE OCEANS Conference (OCEANS) · doi:10.1109/oceans58557.2025.11104715 · Cited by 2
“L 1 -Loss to constrain the intensity of the reconstruction image r ′ and multi-scale structural similarity (MS-SSIM) [12] loss (denoted as L MS ) to ensure the structural similarity with original image x ′ in the logarithmic domain.”
Error Correction for DNA-Based Image Storage
Lazzarotto, D. N., Testolina, M., & Ebrahimi, T. (2025) · Proceedings - International Conference on Image Processing · doi:10.1109/icip55913.2025.11084599 · Cited by 1
Attribute guided adversarial editing for face privacy protection
Xu, Y., Wang, Z., Tang, F., Cao, J., Li, X., & Li, J. (2025) · Visual Informatics · doi:10.1016/j.visinf.2025.100267 · Cited by 2
Self-citationKnowledge Distillation Enhanced DeepJSCC for Edge-Optimized Image Transmission Systems
Dai, S., Chen, X., & Deng, X. (2025) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/icccworkshops67136.2025.11147224 · Cited by 1
“Specifically, we calculate the Peak Signal-to-Noise Ratio (PSNR) and Multi Scale Structural Similarity Index Measure (MS-SSIM) [19] between the original and reconstructed images at different channel signal-to-noise ratio (CSNR).”
CDVS: Compressed Domain On Device Memory Efficient 8K Video SlowMo
Li, J., Wang, C., Sheikh, H. R., & Lee, S.-J. (2025) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw67362.2025.00182 · Cited by 2
Optimal Condition Evaluation for WNN Based Speckle Filtering Algorithms in SAR Image Denoising
Preethisree, G., Vivekanand, V., Mishra, D., et al. (2025) · Sensing and Imaging · doi:10.1007/s11220-025-00647-w · Cited by 1
“(vi) The structural similarity index (SSIM) (17) and (vii) the multi scale structural similarity index (MS-SSIM) [21] are also used in the evaluation of the image despecking algorithm. where c 1 , c 2 are constants and µ X , σ X and µ ˆ X , σ ˆ X are the mean and variances of the reference image X…”
Nature-1k: The Raw Beauty of Nature in 4K at 60FPS
Ghasempour, M., Amirpour, H., & Timmerer, C. (2025) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3746027.3758258 · Cited by 3
“…quality of the compressed videos in Nature-1k we use four widely adopted full-reference metrics: Peak Signal-to-Noise Ratio (PSNR) [24], Structural Similarity Index Measure (SSIM), Multi-Scale Structural Similarity Index Measure (MS-SSIM) [25], and Video Multi-Method Assessment Fusion (VMAF) [26].”
Optimizing Product Deduplication in E-Commerce with Multimodal Embeddings
Kulunk, A., Taskin, B., Eseoğlu, M. F., & Sahin, H. B. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11401681 · Cited by 3
“This section details the experimental setup, presents the main performance results, and concludes with an error analysis to identify areas for future improvement.”
Color Image Quantization Quality Assessment
Hassan, A. M., & Bhagvati, C. (2012) · Communications in computer and information science · doi:10.1007/978-3-642-31686-9_16 · Cited by 1
“In [9] the single-scale structural similarity index was extended in to a multi-scale structural similarity index (MSSIM) that works in multi scales of an image and achieved a better result than SSIM.”
MethodBursty interference-oriented video quality assessment method
Zhou, S., Lu, Z., Wen, X., Fu, B., Shao, H., & Chen, Y. (2015) · Multimedia Tools and Applications · doi:10.1007/s11042-015-2787-y · Cited by 1
“Based on SSIM, SS-SSIM, CW-SSIM and MS-SSIM [40], other evaluation algorithms also were proposed, promoting the development of objective assessment algorithm.”
InfluentialComponent Weight Tuning of SSIM Image Quality Assessment Measure
Skurowski, P., & Janiak, M. (2014) · Lecture notes in computer science · doi:10.1007/978-3-319-11331-9_8 · Cited by 1
“The evaluation procedure, after [12], involved following criteria:”
BackgroundMethodImage and Video Quality Assessment
Gao, W., & Ma, S. (2014) · Springer eBooks · doi:10.1007/978-3-319-14243-2_11 · Cited by 1
Image quality assessment scheme based on gradient similarity and color distortion
Seghir, Z. A., & Hachouf, F. (2015) · International Symposium on Programming and Systems (ISPS) · doi:10.1109/isps.2015.7244985 · Cited by 2
Revealing the dark side of a subjective study: Learnings from noise and sharpness ratings
Saad, M. A., Corriveau, P. J., & Jaladi, R. (2015) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex.2015.7148121 · Cited by 2
High dynamic range tone mapping optimization on perceptual quality and processor specific efficiency
Wang, T. L., & Zhang, X. (2013) · IEEE International Conference on Consumer Electronics - China (ICCE-China) · doi:10.1109/icce-china.2013.6780857 · Cited by 1
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.