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 4,901–5,000 of 6,047
Deep Multiple Description Coding by Learning Scalar QuantizationPreprint
Zhao, L., Bai, H., Wang, A., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.01504 · Cited by 1
“Meanwhile, to better measure image distortion for structural preservation, we introduce multi-resolution structural similarity index (MR-SSIM) fMR(·, ·) as an image quality evaluation factor according to [15].”
MethodInfluentialLearned Scalable Image Compression with Bidirectional Context Disentanglement NetworkPreprint
Zhang, Z., Chen, Z., Lin, J., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.09443 · Cited by 1
“Here, we take L1 norm and MS-SSIM (proposed in [21]) as the mentioned distance functions to train our model in this paper.”
MethodLing, S., Li, J., Che, Z., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.12107 · Cited by 1
Weighted Dark Channel DehazingPreprint
Zhu, M., He, B., & Jiantao, L. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.12245 · Cited by 1
“MSE (mean square error), CIEDE2000 [39], SSIM [40] and MS-SSIM [41] are employed as metrics.”
MethodLearned Quality Enhancement via Multi-Frame Priors for HEVC Compliant Low-Delay ApplicationsPreprint
Lu, M., Cheng, M., Xu, Y., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.01025 · Cited by 1
“For fairness, we follow [3] and evaluate the 3 rd frame of each clip in the Vimeo dataset in terms of PSNR and Struc-( tural SIMilarity (SSIM) [20].”
Reconstruction of Natural Visual Scenes from Neural Spikes with Deep Neural NetworksPreprint
Zhang, Y., Jia, S., Zheng, Y., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.13007 · Cited by 1
“Therefore, these measures are a general guideline rather than the precise characteristic, as it seems that there is no perfect measure [42, 43].”
BackgroundTowards Unsupervised Single-Channel Blind Source Separation using Adversarial Pair Unmix-and-RemixPreprint
Hoshen, Y. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.07504 · Cited by 1
Learning Image and Video Compression through Spatial-Temporal Energy CompactionPreprint
Cheng, Z., Sun, H., Takeuchi, M., & Katto, J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.09683 · Cited by 3
“Measurements To achieve high subjective quality, we used a popular MS-SSIM [30] as distortion term defined by D = 1 − MS-SSIM(x, x̂).”
BackgroundMethodTowards a Semantic Perceptual Image MetricPreprint
Chinen, T. T., Ballé, J., Gu, C., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.00447 · Cited by 1
“Many existing image metrics such as SSIM [1], MS-SSIM [2], PSNR-HVS [3], PSNR-HVS-M [4], FSIM [5], Butteraugli [6], HaarPSI [7] are designed to model specific documented aspects of the human visual system (HVS), such as contrast or color sensitivity, but are unable to access features that would be able to discern objec…”
BackgroundA comprehensive evaluation of full-reference image quality assessment algorithms on KADID-10kPreprint
Varga, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.02096 · Cited by 1
“Surprisingly, some FR algorithms, such as UQI [9], SUMMER[13], QILV [11], MS-SSIM [14], MAD [15], IFC [10], give significantly better results on images with lower distortion intensity levels than on images with higher distortion intensity levels.”
BackgroundEfficient Structurally-Strengthened Generative Adversarial Network for MRI ReconstructionPreprint
Zhou, W., Du, H., Mei, W., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.03858 · Cited by 1
“According to [25], MS-SSIM outperforms SSIM under an appropriate parameter settings.”
MethodNo-Reference Light Field Image Quality Assessment Based on Micro-Lens ImagePreprint
Luo, Z., Zhou, W., Shi, L., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.10087 · Cited by 1
“We conducted comparison experiments between our proposed model and several FR, RR and NR metrics, including nine 2D-FR metrics [10–12, 28–32], three 2D-NR metrics [15,”
MethodResidual-Recursion Autoencoder for Shape Illustration ImagesPreprint
Zhou, Q., Tao, P., Li, X., Chen, S., Zhang, F., & Hu, H. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.02063 · Cited by 1
“) is MS-SSIM (Wang et al., 2003).”
Boundary Guidance Hierarchical Network for Real-Time Tongue SegmentationPreprint
Zeng, X., Zhang, Q., Chen, J., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.06529 · Cited by 1
“After that 𝑃 65 and 𝑅 65 can be calculated as follows: Finally, the boundary loss function is defined as: SSIM [45] can capture the structural information in images.”
MethodObject-Based Image Coding: A Learning-Driven RevisitPreprint
Xia, Q., Liu, H., & Ma, Z. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.08033 · Cited by 1
“07 bpp (or >340x compression ratio), MS-SSIM offers more meaningful quality measurement close to our subjective sensation [21].”
BackgroundMethodInfluentialFuture Video Synthesis with Object Motion PredictionPreprint
Wu, Y., Gao, R., Park, J., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.00542 · Cited by 1
“And the improvement of MS-SSIM is 5%, 7%, 10%, respectively.”
MethodInfluentialCatSIM: A Categorical Image Similarity MetricPreprint
Thompson, G. Z., & Maitra, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.09073 · Cited by 2
“SSIM [13] is an image quality assessment index conceptually designed to account for structural similarities in images as visualized by a human rater.”
InfluentialSpectrally Consistent UNet for High Fidelity Image TransformationsPreprint
Marnerides, D., Bashford‐Rogers, T., & Debattista, K. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2004.10696 · Cited by 2
“The evaluation method uses the perceptually uniform (PU) [27] encoded metrics, SSIM [28], MS-SSIM [29] and HDRVDP-2 [30], traditionally used for evaluating HDR images, for both optimal and culling (clipping of top and bottom 10% of values) exposures, and the scene-referred (scaling to original HDR image range) and dis…”
MethodLinné, A. (2019) · Diva portal (Dalarna University Library) · Cited by 1
Capturing Video Frame Rate Variations through Entropic Differencing.Preprint
Madhusudana, P. C., Birkbeck, N., Wang, Y., et al. (2020) · arXiv (Cornell University) · Cited by 1
Self-citationDisentangle Perceptual Learning through Online Contrastive LearningPreprint
Mei, K., Lu, Y., Yi, Q., Wu, H., Li, J., & Huang, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.13511 · Cited by 2
“The evaluation metrics include pixel-based PSNR, perception related MS-SSIM [22] and LPIPS [24].”
Method3D EIT Reconstructions from Electrode Data using Direct Inversion D-bar and Calderon MethodsPreprint
Hamilton, S. J., Isaacson, D., Kolehmainen, V., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.03018 · Cited by 1
Variational Bayesian QuantizationPreprint
Yang, Y., Bamler, R., & Mandt, S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.08158 · Cited by 17
“As common in image compression work, we measure the distortion between the original and compressed image under two quality metrics (the higher the better): Peak Signal-to-Noise ratio (PSNR), and MS-SSIM (Wang et al., 2003), over all RGB channels.”
MethodVisCode: Embedding Information in Visualization Images using Encoder-Decoder NetworkPreprint
Zhang, P., Li, C., & Wang, C. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.03817 · Cited by 1
“where Lbce represents the BCEWithLogits loss [15] and Lssim denotes the structural similarity index (SSIM) loss [51].”
BackgroundIntrospective Learning by Distilling Knowledge from Online Self-explanationPreprint
Gu, J., Wu, Z., & Tresp, V. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.09140 · Cited by 1
“The similarity can be measured with different metrics, such as cosine distance, multi-scale mean squared error [32], and Wasserstein distance [33].”
MethodConditional Generative Modeling via Learning the Latent SpacePreprint
Ramasinghe, S., Ranasinghe, K., Khan, S. H., Barnes, N. M., & Gould, S. J. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.03132 · Cited by 7
“To overcome this shortcoming, more perceptually motivated metrics have been proposed such as SSIM Wang et al. (2004), MSSIM Wang et al. (2003), and FSIM Zhang et al. (2011).”
MethodSaliency Driven Perceptual Image CompressionPreprint
Patel, Y., Appalaraju, S., & Manmatha, R. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.04988 · Cited by 1
“Thus the estimated entropy is usually jointly minimized with a reconstruction loss such as mean-squared-error (MSE) or multi-scale structured similarity index (MS-SSIM) [43].”
BackgroundDeep Multi-Scale Features Learning for Distorted Image Quality AssessmentPreprint
Zhou, W., & Chen, Z.-B. (2020) · International Symposium on Circuits and Systems · doi:10.48550/arxiv.2012.01980 · Cited by 26
“The proposed method is compared with ten IQA models, i.e., PSNR, SSIM [10], MS-SSIM [11], IW-SSIM [12], FSIMc [13], VIF [14], SPSIM [15], LLM [22], DeepSim [23], and DIQaM-FR [24].”
MethodInfluentialAttributes Aware Face Generation with Generative Adversarial NetworksPreprint
Yuan, Z., Zhang, J., Shan, S., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.01782 · Cited by 1
“Finally we employ common metrics (e.g. BRISQUE, IS, FID and MS-SSIM) and an extra trained facial attribute predictor to do quantitative evaluations.”
MethodInfluentialAnomaly detection in surveillance videos using deep learning
Lu, Y. (2020) · Mspace (University of Manitoba) · Cited by 3
Robustness of Facial Recognition to GAN-based Face-morphing AttacksPreprint
Marriott, R. T., Romdhani, S., Gentric, S., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.10548 · Cited by 1
“where V GG9 is the output of the ninth layer of the VGG classification network [28] used to extract discriminative features, Nv is the number of VGG features, and MSSSIM is the Tensorflow implementation of the MSSSIM metric described in [30].”
MethodQuantifying Visual Image Quality: A Bayesian ViewPreprint
Duanmu, Z., Liu, W., Wang, Z., & Wang, Z. (2021) · Annual Review of Vision Science · doi:10.48550/arxiv.2102.00195 · Cited by 41
“Most of them apply the SSIM index in the sub-band at different spatial locations [108], orientations [80, 135], and frequency content [97, 122, 129] to simulate the characteristics of primary visual cortex.”
MethodSelf-citationDiagnosing Vulnerability of Variational Auto-Encoders to Adversarial AttacksPreprint
Kuzina, A., Welling, M., & Tomczak, J. M. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.06701 · Cited by 9
“Moreover, for supervised attacks we will measure similarity between reconstructions of the target and the adversarial image, MSSSIM[ (cid:101) x t , (cid:101) x a ] .”
MethodInfluentialBai, Y., Liu, X., Zuo, W., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.17015 · Cited by 1
“Most recent methods for lossy image compression adopt variational auto-encoder (VAE) architecture [17, 18] based on transform coding [11], where the rate is modeled by the entropy of quantized latent variables and the reconstruction distortion is measured by PSNR or MS-SSIM.”
MethodConditional Coding and Variable Bitrate for Practical Learned Video CodingPreprint
Ladune, T., Philippe, P., Hamidouche, W., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.09103 · Cited by 1
“Following the CLIC21 video test conditions, D is based on the MS-SSIM [10]: D(x̂t,xt) = 1−MS-SSIM(x̂t,xt).”
MethodNTIRE 2021 Challenge on Quality Enhancement of Compressed Video: Dataset and StudyPreprint
Yang, R., & Timofte, R. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.10782 · Cited by 1
“This may indicate that when optimizing the enhancement networks for PSNR, the MS-SSIM performance will be normally consistent with PSNR.”
MethodInfluentialFashion-Guided Adversarial Attack on Person SegmentationPreprint
Treu, M., Le, T.-N., Nguyen, H. H., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.08422 · Cited by 1
“where Lsim is computed as the multi-scale structural similarity index measure (MS-SSIM) between X̃ and X [43], and Ltv stands for the total variation loss, which is helpful for reducing noise in the adversarial texture [29].”
Region-Adaptive Deformable Network for Image Quality AssessmentPreprint
Shi, S., Bai, Q., Cao, M., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.11599 · Cited by 1
“FR-IQA methods commonly include SSIM [30], MS-SSIM [31] and PSNR, etc .”
MethodInfluentialImage Super-Resolution Quality Assessment: Structural Fidelity Versus Statistical NaturalnessPreprint
Zhou, W., Wang, Z., & Chen, Z. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.07139 · Cited by 1
“Given the original HR image x , and the SISR generated test image y , inspired by the success of MS-SSIM [17], we define subband patch level structural fidelity measure as: where x and y denote the patches extracted from the k -th subband from x and y , respectively, σ x and σ y are their standard…”
InfluentialSelf-citationDeep Graphics Encoder for Real-Time Video Makeup Synthesis from ExamplePreprint
Kips, R., Jiang, R., Ba, S., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.06407 · Cited by 1
Task-driven Semantic Coding via Reinforcement LearningPreprint
Li, X. S., Shi, J., & Chen, Z. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2106.03511 · Cited by 1
“Therefore, many perceptual distortion metrics, such as structural similarity index (SSIM) [22] and multi scale structural similarity index (MS-SSIM) [23], have been proposed.”
BackgroundDeep Image Synthesis from Intuitive User Input: A Review and PerspectivesPreprint
Xue, Y., Guo, Y.-C., Zhang, H., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.04240 · Cited by 1
Improving Generalizability in Limited-Angle CT Reconstruction with Sinogram ExtrapolationPreprint
Wang, C., Zhang, H., Li, Q., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.05255 · Cited by 1
“5) [23] and peak signal-to-noise ratio (PSNR) [22].”
Substitutional Neural Image CompressionPreprint
Wang, X., Jiang, W., Wang, W., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.07512 · Cited by 1
Edge-competing Pathological Liver Vessel Segmentation with Limited LabelsPreprint
Feng, Z., Wang, Z., Wang, X., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2108.00384 · Cited by 1
“Qin et al. (2019) adopted the patch-level SSIM loss (Wang, Simoncelli, and Bovik 2003) to assign higher weights to the edges.”
MethodSelf-citationVTAMIQ: Transformers for Attention Modulated Image Quality AssessmentPreprint
Chubarau, A., & Clark, J. J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.01655 · Cited by 18
Structure-aware scale-adaptive networks for cancer segmentation in whole-slide imagesPreprint
Sun, Y., Lopez, G., Wang, Y., Huang, X., Zhou, H., & Zhang, Q. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2109.12617 · Cited by 1
NeRFReN: Neural Radiance Fields with ReflectionsPreprint
Guo, Y.-C., Kang, D., Bao, L., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2111.15234 · Cited by 1
“We compare NeRFReN with NeRF and NeRF+D on our RFFR test set using three metrics: PSNR, SSIM [26] and LPIPS [32] (see Tab.”
MethodInfluentialLearning Transformer Features for Image Quality AssessmentPreprint
Zeng, C., & Kwong, S. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2112.00485 · Cited by 2
“It has shown nice alignment to human vision system and inspires many later pieces of research like FSIM[21] and MS-SSIM[26].”
MethodMorphology for Hexagonal Image ProcessingPreprint
Çevik, T., Nematzadeh, S., Çevik, N., et al. (2022) · Research Square · doi:10.21203/rs.3.rs-1537955/v1 · Cited by 1
Matched filtering determines human visual search in natural images
Toet, A. (2011) · SciVee · doi:10.4016/28456.01 · Cited by 1
Image quality and forgery detection copula-based algorithms
Hammad, R. (2016) · doi:10.24124/2016/bpgub1124 · Cited by 1
Local Aggregative Attack on SAR Image Classification ModelsPreprint
Du, M., Du, M., Bi, D., et al. (2022) · doi:10.22541/au.165633740.01163731/v1 · Cited by 1
Hair Color Digitization through Imaging and Deep Inverse GraphicsPreprint
Kips, R., Bokaris, P., Perrot, M., et al. (2022) · arXiv (Cornell University) · Cited by 1
Morphology for Hexagonal Image ProcessingPreprint
Çevik, T., Nematzadeh, S., Çevik, N., et al. (2022) · Research Square · doi:10.21203/rs.3.rs-1537955/v2 · Cited by 1
Walk Like Me: Video to Video Action TransferPreprint
Ghafoor, M., Javed, K., & Mahmood, A. (2022) · doi:10.36227/techrxiv.21070066 · Cited by 2
2C-Net: Integrate Image Compression and Classification via Deep Neural NetworkPreprint
Liu, L., Chen, T., Liu, H., et al. (2022) · Research Square · doi:10.21203/rs.3.rs-2049607/v1 · Cited by 1
Shi, W., Yu, M., Zhang, C., et al. (2022) · Research Square · doi:10.21203/rs.3.rs-2233402/v1 · Cited by 1
Luna, R., Zabaleta, I., & Bertalmı́o, M. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.12.22.521587 · Cited by 1
Kang, I., Wu, Z., Jiang, Y. H., et al. (2023) · Research Square · doi:10.21203/rs.3.rs-2459757/v1 · Cited by 1
Similarity Evaluation of Graphic Design Based on Deep Visual Saliency FeaturesPreprint
Liu, Z., An, J., Huang, C., et al. (2023) · Research Square · doi:10.21203/rs.3.rs-2537865/v1 · Cited by 1
Inverse modelling of core flood experiments for predictive models of sandstone and carbonate rocksPreprint
An, S., Wenck, N., Manoorkar, S., et al. (2023) · doi:10.22541/essoar.168298704.44298407/v1 · Cited by 1
Multi-Information Guided Camouflaged Object DetectionPreprint
Shi, C., Zhao, L., Wang, R., et al. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4500697 · Cited by 1
Artifact generation when using perceptual loss for image deblurringPreprint
Krawczyk, P., Gaertner, M., Jansche, A., Bernthaler, T., & Schneider, G. (2023) · doi:10.36227/techrxiv.23791962 · Cited by 2
Advancing Privacy in Deep Learning Through Data TransformationsPreprint
Ay, S., Ünal, C. B., Cardei, M., Rajendran, S., Zhang, W., & Topaloĝlu, Ü. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4525320 · Cited by 1
MSTCNN: A Transformer with multi-scale convolutional neural network for multi-focus image fusionPreprint
Zhai, H., Chen, Y., Zeng, Z., Luo, N., & Li, Q. (2023) · Research Square · doi:10.21203/rs.3.rs-3303519/v1 · Cited by 1
One-to-Multiple Clean-Label Image Camouflage (Omclic) Based Backdoor Attack on Deep LearningPreprint
Wang, G., Ma, H., Gao, Y., Abuadbba, A., Zhang, Z., Kang, W., Al-Sarawi, S. F., Zhang, G., & Abbott, D. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4579073 · Cited by 1
TomoSuitePY: A Pythonic Workflow for Noisy and Sparse Angle TomographyPreprint
Judge, W., Andrade, V. D., Biçer, T., Carlo, F. D., Cabana, J., & Allen, E. M. (2023) · ChemRxiv · doi:10.26434/chemrxiv-2023-xvtmq · Cited by 1
Transformer-Based Hierarchical Dynamic Decoders for Salient Object DetectionPreprint
Zheng, Q., Zheng, L., Deng, J., et al. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4552874 · Cited by 1
Gummadi, D., Chan, P. H., Wang, H., et al. (2023) · doi:10.36227/techrxiv.24566371.v1 · Cited by 1
A New Index for Assessing Quality of Solar X-ray Imaging System and Image ReconstructionPreprint
Li, Z., Yu, W., Su, Y., et al. (2024) · Research Square · doi:10.21203/rs.3.rs-4016826/v1 · Cited by 1
DCT-based Perceptual Loss For Screen Content Image Super-ResolutionPreprint
Sekhavaty, H., Hosseinkhani, M., & Mansouri, A. (2024) · Research Square · doi:10.21203/rs.3.rs-3869544/v1 · Cited by 1
Higgins, G., Razavi‐Far, R., Zhang, X., David, A., Ghorbani, A. A., & Ge, T. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.5062204 · Cited by 1
Rai, N. S. H., & Ranjani, G. (2024) · doi:10.1109/csitss64042.2024.10817020 · Cited by 1
Direct Inversion for Bidimensional Distributions of Aerosol CharacteristicsPreprint
Sipkens, T. A., & Corbin, J. C. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5203085 · Cited by 1
Ghost 2.0: Generative High-Fidelity One Shot Transfer of HeadsPreprint
Groshev, A., Iashchenko, A., Paramonov, P., Dimitrov, D., & Кузнецов, А. В. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5221493 · Cited by 7
AI4CellFate: Interpretable Early Cell Fate Prediction with Generative AIPreprint
Cunha, I. M., Panconi, L., Bauer, S., Gestin, M., Latron, E., Sahai, E., Marois, A. L., & Griffié, J. (2025) · bioRxiv · doi:10.1101/2025.05.12.653464 · Cited by 2
Vogler, B. T. L., Gentsch, G. J., Carravilla, P., Helmerich, D. A., Heiby, J. C., Reglinski, K., Durso, W., Klein, T., Hildebrandt, F., Zahid, R., et al. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.16.676574 · Cited by 1
EBREnv: SVBRDF Estimation in Uncontrolled Environment Lighting via Exemplar-Based Representation
Wang, L., Zhao, J., Zhang, L., et al. (2025) · SIGGRAPH Asia Conference Papers · doi:10.1145/3757377.3763905 · Cited by 1
“Specifically, we utilize the MS-SSIM [Wang et al. 2003] algorithm to identify the regions where SGs fail to capture sufficient details, and then dynamically add new SGs to these regions, thereby enriching the representation.”
BAIT: Visual-illusion-inspired Privacy Preservation for Mobile Data Visualization
Cheng, S., Zhang, S., Ma, D., & Wang, Y. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791259 · Cited by 2
“For the similarity between 𝐼 and 𝐼 , we use the Multi-Scale Structural Similarity Index (MS-SSIM) [83] to measure the similarity between the decoy visualization and privacy-preserving visualization [1, 66].”
When the Codec Hallucinates: User Perceptions of Miscompressed Images
Hofer, N., & Böhme, R. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3790293
Physics-informed line-of-sight learning for scalable deterministic channel modelingPreprint
Wang, X., Huang, J., Zhou, C., Shen, X., & Cheng, N. (2026) · Research Square · doi:10.21203/rs.3.rs-9261533/v1 · Cited by 1
Chen, Z., & Wang, H. (2026) · Displays · doi:10.1016/j.displa.2026.103566 · Cited by 1
Level of Detail for Trees Using Clustered Ellipsoids
Clasen, M., Paar, P., & Prohaska, S. (2011)
Application of Attention Models in Image Processing
Zhang, L., & Lin, W. (2013) · Wiley eBooks · doi:10.1002/9780470828144.ch8
Blind multichannel image deconvolution with regularization
Mseddi, W. S., Abed‐Meraim, K., & Beghdadi, A. (2005) · IEEE International Symposium on Signal Processing and Information Technology (ISSPIT) · doi:10.1109/isspit.2004.1433701
Which metric can predict coding gain of H.265/HEVC over H.264/AVC?
Xiao, J., Song, L., Luo, Z., et al. (2015) · IEEE International Symposium on Broadband Multimedia Systems and Broadcasting (BMSB) · doi:10.1109/bmsb.2015.7177228
“• Multiscale SSIM (MSSIM) [16] is designed based on SSIM but shows better consistency with subjective evaluation than PSNR, SSIM and so on.”
MethodCross-layer perceptual optimization for wireless video transmission
Khalek, A., & Nazih, A. (2013) · Texas ScholarWorks (Texas Digital Library)
Sarwar, S. M., & Hossain, I. (2013) · International Conference on Informatics, Electronics & Vision (ICIEV) · doi:10.1109/iciev.2013.6572683
“Multi-scale SSIM (MS-SSIM) [3], Percentile Pooling SSIM (P-SSIM) [3], Complex-Wavelet SSIM index (CW-SSIM) [4], Gradient-based Structural Similarity (G-SSIM) [5], and Three-Component Weighted SSIM [6].”
MethodVideo packet priority assignment based on spatio-temporal perceptual importance
Reddy, S. S., Manasa, K., & Channappayya, S. S. (2015) · National Conference on Communications · doi:10.1109/ncc.2015.7084822
A Method of Video Quality Assessment for CMMB Video Test Sequences
Zhu, S.-J., & Yang, J. J. (2012) · International Conference on Computer Distributed Control and Intelligent Environmental Monitoring · doi:10.1109/cdciem.2012.16
Minimum image quality assessment based on saliency maps: a human visual approach
Barreira, J., Bessa, M., & Magalhães, L. (2013) · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · doi:10.1117/12.2005380
Neural learning-based image quality metric without reference
Chetouani, A. (2014) · International Conference on Image Processing Theory, Tools and Applications (IPTA) · doi:10.1109/ipta.2014.7002004 · Cited by 3
Effective preprocessing stage in the fourier transform domain for image quality assessment
Liu, M., Zhai, G., Gu, K., & Yang, X. (2014) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex.2014.6982318
vPSNR: a visualization-aware image fidelity metric tailored for diagnostic imaging
Lundström, C. F. (2012) · International Journal of Computer Assisted Radiology and Surgery · doi:10.1007/s11548-012-0792-4 · Cited by 2
Capturing structural distortions in digital images and videos
Peng, P. (2012) · Summit (Simon Fraser University)
New Method for Spatial Scalable Video Quality Evaluation
Chen, C., Ji, W., Chen, B.-W., et al. (2015) · International Conference on Platform Technology and Service · doi:10.1109/platcon.2015.11
Image inpainting via variation of variances and linear weighted filling-in
Zhang, Q., & Lin, J. (2011) · Optical Engineering · doi:10.1117/1.3594789
Gadzicki, K., & Zetzsche, C. (2013) · Journal of Modern Optics · doi:10.1080/09500340.2013.809162
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.