Multiscale structural similarity for image quality assessment
The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index; none of the 14 sources its claims cite is checked for support yet.
Abstract
The structural similarity image quality paradigm is based on the assumption that the human visual system is highly adapted for extracting structural information from the scene, and therefore a measure of structural similarity can provide a good approximation to perceived image quality. This paper proposes a multiscale structural similarity method, which supplies more flexibility than previous single-scale methods in incorporating the variations of viewing conditions. We develop an image synthesis method to calibrate the parameters that define the relative importance of different scales. Experimental comparisons demonstrate the effectiveness of the proposed method.
Overview
CoChat Summary
TL;DR
This paper proposes a multi-scale structural similarity (MS-SSIM) image quality method that combines contrast and structure comparisons across scales and luminance comparison at the coarsest scale. On the LIVE JPEG/JPEG2000 database, MS-SSIM outperformed PSNR, the Sarnoff model, and single-scale SSIM variants on all reported evaluation criteria.
Key findings
- MS-SSIM achieved the best reported performance: CC 0.969, ROCC 0.966, MAE 3.86, RMS 4.91, and OR 1.16%, compared with the best single-scale SSIM at M=2 (CC 0.963, ROCC 0.959, MAE 4.21, RMS 5.38, OR 2.62%) and PSNR (CC 0.905, ROCC 0.901, MAE 6.53, RMS 8.45, OR 15.7%).
- Single-scale SSIM performance varied with scale; the best single-scale result was at M=2, and single-scale SSIM tended to give higher scores as scale increased.
- The calibrated cross-scale exponents were β1=γ1=0.0448, β2=γ2=0.2856, β3=γ3=0.3001, β4=γ4=0.2363, and α5=β5=γ5=0.1333.
- The authors conclude that multi-scale SSIM provides a meaningful balance between scales and outperforms the best single-scale SSIM and state-of-the-art image quality metrics.
Methods
- The MS-SSIM system iteratively applies a low-pass filter and downsamples by 2; at each scale it computes contrast and structure comparisons, computes luminance comparison only at Scale M, and combines them using exponents normalized so that ∑γj=1.
- Cross-scale parameters were calibrated by image synthesis: 5 scales and 12 distortion levels (range 2^3 to 2^14) were used with 10 original 64×64 images, producing 600 distorted images; 8 subjects selected images of equal perceived quality across scales at a viewing distance of 32 pixels per degree of visual angle.
- Evaluation used the LIVE database with 344 JPEG and JPEG2000 compressed images (bit rate 0.028 to 3.150 bits/pixel), MOS from 13–25 subjective scores, and compared PSNR, Sarnoff JNDmetrix 8.0, single-scale SSIM for M=1–5, and MS-SSIM using CC, MAE, RMS, OR, and ROCC after logistic mapping.
Limitations
- Stated: the image-synthesis calibration approach is “still rather crude,” and the authors are working on a more systematic approach.
- Stated: CSF cannot be directly used to derive the parameters because it is typically measured at visibility threshold with simplified stimuli, whereas the goal is comparing complex structured images at visible distortion levels.
- Evident from the text: calibration used only 10 original 64×64 grayscale images and 8 subjects, including one author, and evaluation was limited to the LIVE JPEG/JPEG2000 database.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 14 Not checked · 18 Own results and interpretations
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It can also be observed that the single-scale model tends to supply higher scores with the increase of scales.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Finally, for every one of the objective evaluation criteria, multi-scale SSIM model outperforms all the other models, including the best single-scale SSIM model, suggesting a meaningful balance between scales.”
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
“Experiments show that with an appropriate parameter settings, the multi-scale method outperforms the best single-scale SSIM model as well as state-of-the-art image quality metrics.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 5. DISCUSSIONS · p. 4 · hedged
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In general, the subjects agreed with each other on each image more than they agreed with themselves across different images.”
plainstates an associationn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The resulting parameters we obtained are β 1 = γ 1 = 0.0448, β 2 = γ 2 = 0.2856, β 3 = γ 3 = 0.3001, β 4 = γ 4 = 0.2363, and α 5 = β 5 = γ 5 = 0.1333, respectively.”
plainn = 8- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“However, the SSIM index algorithm introduced in [5] is a single-scale approach.”
What the sources say
- Not matchedcites [5] [5] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, 'Image quality assessment: From error measurement to structural similarity,' IEEE Trans. Image Processing , vol. 13, Jan. 2004. -
More than one paper fits this citation.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not matched
10 more claims
Create a free account to see all 33 claims and every source verdict.
- It may state two separate claims as one. 8
- Probably not a substantial claim. 13
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Further, as research in quality assessment has demonstrated, incorporating multiscale information when assessing quality produces QA algorithms that perform better in terms of correlation with human perception [13], [29].”
From No-Reference Image Quality Assessment in the Spatial Domain (2012) · cites this paper as [13]
Citations
Cites
Image quality assessment: from error visibility to structural similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · IEEE Transactions on Image Processing · doi:10.1109/tip.2003.819861 · Cited by 57,892
Self-citationA universal image quality index
Wang, Z., & Bovik, A. C. (2002) · IEEE Signal Processing Letters · doi:10.1109/97.995823 · Cited by 5,757
Self-citationImage quality measures and their performance
Eskicioglu, A. M., & Fisher, P. S. (1995) · IEEE Transactions on Communications · doi:10.1109/26.477498 · Cited by 2,117 · cited 1×
“Traditional perceptual image quality assessment methods are based on a bottom-up approach which attempts to simulate the functionality of the relevant early human visual system (HVS) components.”
BackgroundA Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients
Portilla, J., & Simoncelli, E. P. (2000) · International Journal of Computer Vision · doi:10.1023/a:1026553619983 · Cited by 1,806 · cited 2×
“These synthesis methods provide a powerful and efficient means of testing a model, and have the added benefit that the resulting images suggest improvements that might be made to the model [11].”
MethodSelf-citation(1995) · Medical Entomology and Zoology · Cited by 1,212
Pyramid-based texture analysis/synthesis
Heeger, D. J., & Bergen, J. R. (1995) · Proceedings of the 22nd annual conference on Computer graphics and interactive techniques - SIGGRAPH '95 · doi:10.1145/218380.218446 · Cited by 1,095
Video quality assessment based on structural distortion measurement
Wang, Z., Lu, L., & Bovik, A. C. (2004) · Signal Processing: Image Communication · doi:10.1016/s0923-5965(03)00076-6 · Cited by 1,024
Self-citationImage Quality Assessment: From Error Measurement to Structural Similarity
Wang, Z., Bovik, A. C., Sheikh, H. R., & Simoncelli, E. P. (2004) · Cited by 919
Self-citationTeo, P. C., & Heeger, D. J. (2002) · Proceedings of 1st International Conference on Image Processing · doi:10.1109/icip.1994.413502 · Cited by 471
Objective video quality assessment
(2003) · Handbook of Video Databases · doi:10.1201/9780203489864-45 · Cited by 359
Research works printed but not linked: 2
- 0, JNDmetrix Technology, Sarnoff Corporation
- 2000, Final report from the video quality experts group on the validation of objective models of video quality assessment, VQEG
Cited by
Items 5,901–6,000 of 6,047
Chai, Z., Zhang, X., Jiang, L., et al. (2026) · Research Square · doi:10.21203/rs.3.rs-9399727/v1
Yan, W., Xu, X., Zhang, Y., Xia, X., & Li, Y. (2026) · Electronic Research Archive · doi:10.3934/era.2026216
CondFoodGen: A Conditional Two-Stream Network for Controllable Food Image Generation
Zhao, M., Xiong, H., Min, W., Hou, S., Zhang, M., & Jiang, S. (2026) · IEEE Transactions on Image Processing · doi:10.1109/tip.2026.3699097
“Multi-Scale Structural SIMilarity (MS-SSIM) [44] evaluates structural similarity between real and generated images based on multiple scales.”
Micó, D. G., García, M., & Montagud, M. (2026) · ACM International Conference on Interactive Media Experiences · doi:10.1145/3788851.3815044
Temporally Coherent Dynamic Surface Reconstruction With Planar Gaussian Splatting
Cai, W., Ye, W., Ye, P., et al. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3702802
“For rendering evaluation, we report PSNR [22], MS-SSIM [23], and LPIPS [24].”
Liu, P., Wei, A., Zhang, C., et al. (2026) · IEEE Transactions on Image Processing · doi:10.1109/tip.2026.3700927
Me and ChatGPT: Making an Image Processing Class Demo Suite
Bovik, A. (2026) · IEEE Southwest Symposium on Image Analysis and Interpretation (SSIAI) · doi:10.1109/ssiai68148.2026.11553746
Self-citationEvaluation of AI-Generated Image Quality for Selected Artistic Styles
Kopczyński, M., & Ogar, J. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-27927-9_20
CGNet: A Lightweight Multimodal Framework for Congestion Prediction in VLSI Physical Design
Li, Y., Xia, Y., Zhong, G., et al. (2026) · ACM Transactions on Design Automation of Electronic Systems · doi:10.1145/3819824
Self-attention and Unet-Based Lossy Image Compression
Zeng, J., & Lei, J. (2026) · Lecture notes in electrical engineering · doi:10.1007/978-981-95-4897-2_36
Toward Generation-Centric Coding: Compressing Latents representation for TI2V Synthesis
Liu, J., Ji, W., Meng, X., Zhang, W., & Wang, Y. (2026) · International Conference on Multimedia Retrieval · doi:10.1145/3805622.3810600
“Specifically, we report PSNR and MS-SSIM [31] for low-level fidelity, alongside LPIPS [38] to quantify perceptual degradation in deep feature space.”
Uncertainty-Gated Generative Compression for Structure-Preserving Multimedia Retrieval
Ding, Y., Chen, Z. Q., & Wei, J. (2026) · International Conference on Multimedia Retrieval · doi:10.1145/3805622.3810699
“Classical codecs from JPEG [36] to VVC/VTM [11] and learning-based codecs [5, 6, 28] optimize for pixel-level fidelity (PSNR, MS-SSIM [38]), but at low bitrates this paradigm produces blurry reconstructions that lack natural image characteristics [10].”
Kou, S., Zhang, F., Nazarenus, J., Koch, R., Wang, C., & Dodgson, N. A. (2026) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2026.3703852 · Cited by 1
DiRIC: Diffusion Prior Refinement for Efficient Low-Rate Image Compression
Xia, Y., Zhou, Y., Wang, J., Hong, M., Wang, H., Chen, B., & Wang, Y. (2026) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2026.3704972 · Cited by 1
“(1) Distortion-based metrics: PSNR and MS-SSIM [53].”
Guo, Q., Bao, Z., Wang, S., Dong, C., Jiang, S., Zhu, L., Sun, Y., & Xu, X. (2026) · IEEE Transactions on Network Science and Engineering · doi:10.1109/tnse.2026.3704861
A Novel HARQ Scheme for Semantic Communication Systems
Han, Y., Bao, Z., Dong, C., et al. (2026) · IEEE Wireless Communications and Networking Conference Workshops (WCNCW) · doi:10.1109/wcncw67598.2026.11555505
“For the image quality evaluation, Peak Signal-to-Noise Ratio (PSNR) and Multi-Scale Structural Similarity(MS-SSIM) [16] are employed.”
Patch Size-Considered Incremental Learning for Video Compression Artifact Reduction
Feng, Z., & Jung, C. (2026) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas66217.2026.11562507
“We use a weighted combination of L 1 loss, L 2 loss [30], and multi-scale structural similarity (MS-SSIM) loss [31] to train the HOP network.”
M-SSIM-3DIFWI: 3-D Implicit Full Waveform Inversion Based on the Multiscale Structural Similarity
He, L., Song, C., Alkhalifah, T. A., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3704938
An Efficient Hardware Accelerator for JPEG-AI Image Compression on FPGA
Ma, W., Gao, S., Leng, S., et al. (2026) · IEEE International Symposium on Circuits and Systems proceedings · doi:10.1109/iscas66217.2026.11561920
Chen, H., Chang, Y., Guo, Y., et al. (2026) · Communications in Computational Physics · doi:10.4208/cicp.oa-2026-0002
A Novel Iterative Framework for Visible-Infrared Image Fusion in Foggy Weather Object Detection
Luo, L., Zhang, Z., Huang, K., et al. (2026) · Chinese Control and Decision Conference (CCDC) · doi:10.1109/ccdc69976.2026.11560432
“To further emphasize structural preservation at multiple resolutions, we incorporate the multi-scale structural similarity (MS-SSIM) loss [27], which evaluates luminance, contrast, and structure between I f and I g .”
Vishwakarma, A. K., Ranjan, R., Priyadarshi, R., & Kollie, D. S. (2026) · Discover Computing · doi:10.1007/s10791-026-10248-0 · Cited by 1
“Early extensions of image quality assessment models such as SSIM [60], MS-SSIM [61], VIF [59], and GMSD [58] adopted frame-wise evaluation followed by temporal pooling.”
InfluentialYao, B., Li, J. F., Ma, L., & Tian, Z. (2026) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2026.110805
Computational Attention-based Modeling of Individually Perceived Quality of Compressed Images
Tiotsop, L. F., Geissler, M., & Barkowsky, M. (2026) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3822181
“Early objective approaches relied on handcrafted metrics such as PSNR [32], SSIM [53], MS-SSIM [54], and VIF [44].”
pVAR: An Ultra Low Complexity Image Quality Assessment via Spatial Error Variance
Dziembowski, A. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3708417
“The proposed metric is benchmarked against the classic pixel-domain metric (PSNR) and widely adopted structural metrics (SSIM [6] and multi-scale SSIM: MS-SSIM [34]), including their three-component (YUV) and Immersive Video (IV) variants [21], [22].”
Duan, Y.-K., Chen, Q.-Y., Dong, X., et al. (2026) · 2026 6th International Symposium on Computer Technology and Information Science (ISCTIS) · doi:10.1109/isctis70043.2026.11572355
“Multi-index monitoring identified epoch 120 [20] as the optimal balance point, where MS-SSIM [21], LPIPS [4], FID [22], and FG-FID [2] concurrently reached their ideal states.”
DeLux: Cross-Modal Local Artifact Restoration in Video Using Neuromorphic Data
Stachowiak, B., & Brzeziński, D. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37038-9_16
From Magnetic to Optical: NFC-to-Camera Side-Channel Communication via Standard-Compliant Barcodes
Peng, C., Dai, D., & Yang, L. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3709665
“We then evaluated the two images using five full-reference image quality assessment metrics [34], [35], [36], [37].”
Forget Superresolution, Sample Adaptively (when Path Tracing)
Bálint, M., Salaün, C., Seidel, H., & Myszkowski, K. (2026) · ACM Transactions on Graphics · doi:10.1145/3811377 · Cited by 1
“This task-informed training places MILO closer to perception-based error visibility prioritization, in contrast to SSIM/MS-SSIM [Wang et al. 2004, 2003], which emphasizes structural fidelity, and LPIPS [Zhang et al. 2018], which focuses on visual similarity.”
GDE: A Generative Diffusion Equalizer for ICSI-Free Semantic Communications in SAGINs
Shi, H., Zhou, L., Tian, W., et al. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3709876
“Consequently, methods that perform well under ideal or accurately estimated channel conditions may su ff er substantial degradation under the channel uncertainty typical of non-terrestrial links [22], [23], [24].”
Cao, Z., Yin, H., Wang, H., et al. (2026) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2026.3710288
“4) Evaluation Metrics: We evaluate restoration fidelity using PSNR, SSIM [47], and MS-SSIM [48], computed on where e MAC = 4 .”
Lee, H.-G., Shin, J., Han, W.-T., et al. (2026) · Agronomy · doi:10.3390/agronomy16131292
“The MS-SSIM is a multi-scale extension of the structural similarity index that was developed to overcome the limitations of single-scale SSIM analysis [34].”
InfluentialChannel Knowledge Base Enhanced Cross-Modal Semantic Communications
Wang, Y., Gao, H., Sun, M., et al. (2025) · IEEE Globecom Workshops (GC Wkshps) · doi:10.1109/gcwkshps68340.2025.11590930
“Based on the above premises, we define the reward function R t as where the peak signal-to-noise ratio (PSNR) and the multi-scale structural similarity index (MS-SSIM) [16] are used as metrics, representing image quality and similarity across multiple scales, respectively.”
Intelligent GAN-Based Framework for High-Resolution Satellite Imagery Enhancement
Sineglazov, V., & Holovachov, A. (2026) · International Workshop of IT-professionals on Artificial Intelligence · doi:10.55056/ceur-ws.org/vol-4164/paper3.pdf
“Wang et al. [25] proposed the LPIPS metric, which has good correlation with human judgment and was employed as an evaluation metric and training objective.”
Land-Then-Transport: A Flow Matching-Based Generative Decoder for Wireless Image Transmission
Fu, J., Xiao, M., Skoglund, M., & Kim, D. I. (2026) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2026.3710439 · Cited by 4
Li, T., Cao, Q., Sun, K., Wan, Z., Li, S., & Li, Z. (2026) · Asia-Pacific International Symposium on Electromagnetic Compatibility (APEMC) · doi:10.1109/apemc65388.2026.11593686
“…effectiveness of the proposed TriGaze360, we compare it against ten representative IQA models, including two 2D-IQA methods: SSIM [16], and MS-SSIM [17]; two SIQA methods: Chen [18], and PAD-Net [19]; two OIQA methods: WS-PSNR [10], and VGCN [11]; and four SOIQA methods: SOIQE [5], VP-BSOIQA [6],…”
Lv, T., Liu, Y., & Jiang, H. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116521
Off-Chip Super-Resolution AI Model to Support Embedded Memory Diagnosis
Anedda, S., Bernardi, P., Coppetta, M., et al. (2026) · Proceedings/Proceedings - European Test Workshop · doi:10.1109/ets69887.2026.11591702
SACT-FNet: Structure-Aware Context Transformer Fusion Network
Wang, Q., & Wei, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-2862-1_25
Liang, J., Xu, W., Liu, L., et al. (2026) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2026.3709956
VQ-DSC-R: Robust Vector Quantized-Enabled Digital Semantic Communication With OFDM Transmission
Chen, J., Ma, N., Xu, X., Zhu, T., Song, H., Dong, C., Liu, W., Meng, R., & Zhang, P. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3712380 · Cited by 1
“The batch size is set to 16, and the model is trained for [45] for pixel-level accuracy and MS-SSIM [46] for perceptual quality.”
LVT: A Learned Video Transcoding Framework
Fu, N., Yang, D., Lin, Z., et al. (2026) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3820890
“For the evaluation of transcoding performance, the peak signal-to-noise ratio (PSNR) and the multiscale structural similarity (MS-SSIM) [37] are used to measure the quality of the reconstructed frames in comparison to the original frames.”
Brain-inspired visual sensing–perception for open-world environments
Lin, Y., Chen, Y., Meng, Y., Chen, X., Wang, T., Zhao, R., & Shi, L. (2026) · Nature Sensors · doi:10.1038/s44460-026-00095-3 · Cited by 1
“Among the referenced metrics, peak signal-to-noise rate (PSNR), LPIPS VGG-16 53 and multiscale structural similarity index measure (ms-SSIM) 54 are employed.”
SRDiffC: Ultra-Low-Bitrate Visual Communication with Generative Reconstruction
Wang, S., & Shi, W. (2026) · IEEE International Conference on Communications workshops · doi:10.1109/iccworkshops63917.2026.11586467
“We also report distortion metrics PSNR and MS-SSIM [25] to measure reconstruction fidelity.”
Talking Traces: Audio Reconstruction Power Side-Channel Attack on Neural Network FPGA Accelerators
Reibold, J., Meyers, V., & Tahoori, M. B. (2026) · IEEE International Symposium on Hardware Oriented Security and Trust (HOST) · doi:10.1109/host68814.2026.11604627
“We use the MSE, with errors averaged over all pixels, and Mean Structural Similarity Index Measure (MSSIM) [40] between spectrograms.”
Guo, D., Jia, Q., Wang, X., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3714006
Multi-scale wavelet transformer network for remote sensing image dehazing
Hu, T., Chen, M., Sun, W., Lai, X., Zhang, X., & Zhang, S. Y. (2026) · Multimedia Systems · doi:10.1007/s00530-026-02517-1
“We set M = 5 scales and adopt the original MS-SSIM scale weights α M , β j as in [45 ].”
Cascaded dual-branch deep neural network for simultaneous OCT and OCTA reconstruction
Chen, S., Chen, Y., Ma, D., Song, J., Šarunic, M. V., Ju, M. J., & Beg, M. F. (2026) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2026.111046
Multi-scale Perceptual Supervision Based Progressive Document Image Dewarping
Fang, C., & Peng, J. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3429-5_27
Zhu, X., Cheng, J., Li, C., et al. (2026) · Communications in computer and information science · doi:10.1007/978-981-92-3538-4_1
MA-Mamba: Modality-Adaptive Selective State Space Models for Dual-Domain Medical Image Fusion
Chen, X., Lu, C., Huang, L., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3403-5_20
Multi-condition guided diffusion model for face sketch-to-photo synthesis
Que, Y., Cheng, X., Shi, S., Wan, W., & Jiang, N. (2026) · Neural Networks · doi:10.1016/j.neunet.2026.109406
WREN: Low Light Image Enhancement Using Retinex Theory-Based Double U-Net-Like Structures
Kaneko, R., Hara, J., Higashi, H., et al. (2026) · Proceedings - International Conference on Image Processing · doi:10.1109/icip61757.2026.11630177
“Its internal loss functions are represented as follows: With the standard MSE L MSE , we use the multi-scale structural similarity (MS-SSIM) [23] loss L MS-SSIM , the perceptual loss L perc [24] in which ψ ( · ) in (15) denotes the high-level feature extractor (VGG-16 pre-trained on ImageNet) [25],…”
JFDFusion: joint frequency decomposition and deep feature learning for infrared–visible image fusion
Yu, Q., Zhou, D., Xiao, L., et al. (2026) · Pattern Analysis and Applications · doi:10.1007/s10044-026-01734-8
“…fidelity [ 27]; SSIM and MS-SSIM evaluate image quality from the perspectives of structural similarity and multi-scale structural preservation [28, 29]; Qabf quantifies the amount of edge and gradient informa - tion transferred from the source images to the fused image [30]; and SCD measures the…”
The Devil Is in the Dark Pixels: Toward Brightness Bias-Robust Denoising
Cho, S., Chen, Z., & Li, X. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37314-4_25
Dual-Domain Adversarial Purification for Robust Remote Sensing Scene Classification
Su, Y., Mei, S., Ma, M., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3715638
“To further preserve structural consistency, a multiscale structural similarity loss based on MS-SSIM [38] is adopted.”
VIVID: Vision-and-Text Integrated Vectorization for Image Documents
Yuan, Z., & Wu, X. H. (2026) · doi:10.1145/3810987.3815538
“Specifically, 𝑆 𝑚𝑠 − 𝑠𝑠𝑖𝑚 ( 𝑐 ) measures global visual similarity [34], 𝑆 𝑜𝑟𝑏 ( 𝑐 ) measures local structural consistency via ORB-based matching [10, 24], 𝑆 𝑡𝑒𝑥𝑡 − 𝐹 1 ( 𝑐 ) measures character-level textual fidelity, 𝑆 𝑟𝑒𝑛𝑑 ( 𝑐 ) penalizes invalid or non-renderable…”
FaRJSCC: Frequency-Aware Rate-Adaptive Joint Source-Channel Coding for Semantic Communications
Zhu, C., Wang, B., & Li, X. (2026) · doi:10.1109/acc6g70208.2026.11605519
Ran, Y., Sang, H., Chen, S., et al. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3716183
“To evaluate the video reconstruction quality, we employ Peak Signal-to-Noise Ratio (PSNR) [36], Multi-Scale Structural Similarity (MS-SSIM) [37], and Bjøntegaard Delta Bit Rate (BD-BR) [36].”
Low Displacement Rank Dictionary Learning
Dantas, C. F., Lima, V. C., & Lopes, R. (2026) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2026.3716412
A Privacy-preserving Edge-cloud Video Analytics System via Policy-based Frame Transformation
Lu, R., Wu, W., Shi, S., Hu, C., & Wang, D. (2026) · ACM Transactions on Internet Technology · doi:10.1145/3833875 · Cited by 2
“The deviation is measured by MSSIM [58].”
InfluentialMennaoui, A. E., Meehan, J., Hemrit, G., & Dugelay, J. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31441-3_17
Pixel‐Wise Uncertainty Quantification of Accelerated MRI Reconstruction
Giannakopoulos, I. I., Muthukumar, L. B. G., Lui, Y. W., & Lattanzi, R. (2026) · Magnetic Resonance in Medicine · doi:10.1002/mrm.70526 · Cited by 2
“Our loss function was the structural similarity index measure (SSIM) [34].”
MambaStyle: Efficient StyleGAN Inversion for Real Image Editing with State-Space ModelsPreprint
Lopez, J., Hinojosa, C. A., Argüello, H., et al. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7194395
Tsuchiya, R., Usui, K., Sakamoto, H., et al. (2026) · Radiological Physics and Technology · doi:10.1007/s12194-026-01108-z
From pixels to strokes: A survey on generative handwriting
Joshi, P., Patil, Y. M., Saha, S., & Doermann, D. (2026) · International Journal on Document Analysis and Recognition (IJDAR) · doi:10.1007/s10032-026-00609-2
Integrating “Where” and “How” for Expertise-Aware 8K Video Quality Prediction
Sugito, Y., & Tanaka, T. (2026) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex69967.2026.11618322
“MSBlock: For video-based quality features, we focused on MS-SSIM [12], which reflects perceptual structural similarity.”
Comparative Study of Training Strategies for Projection-Domain Denoising in Multislice Helical CT
Balogh, Z., Hassan, M. N., Shahid, M., et al. (2026) · Journal of Imaging Informatics in Medicine · doi:10.1007/s10278-026-02145-1
Perceptually-Weighted Video Quality Metric for Asymmetric Encoded Sports Videos
Meyer, A. P., Janzen, J., Reddy, D., et al. (2026) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex69967.2026.11618346
“The metrics used for comparison are PSNR-Y, SSIM [8], MS-SSIM [11], Funque [3], LPIPS [4], and different configurations of VMAF.”
A Mamba-Based Perceptual Loss Function for Learning-Based UGC Transcoding
Qi, Z., Feng, C., Zhang, F., Xu, X., Liu, S., & Bull, D. (2026) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex69967.2026.11618348
“Simple metrics like PSNR and SSIM [14, 15] provide straightforward fidelity measurements, while perceptually inspired metrics, such as VMAF [16], incorporate human visual system characteristics to better align with subjective judgments.”
Scale-Agnostic Image Quality Assessment via Magnitude-Aware Ranking
Zizien, A., & Fliegel, K. (2026) · International Conference on Quality of Multimedia Experience (QoMEX) · doi:10.1109/qomex69967.2026.11618328
“Significant milestones such as the Structural Similarity Index (SSIM) [1] and its multi-scale variant (MS-SSIM) [2] sought to address this by explicitly modeling the HVS’s sensitivity to structural information and luminance.”
Balancing Visual Fidelity and Detection Reliability: Scalable Remote Sensing Image Compression
Tang, R., Zhou, P., Jia, J., Wang, K., Lei, Z., Shi, C., Ma, H., & Xiong, H. (2026) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2026.3719025
“3) Metrics and Comparative Methods: In order to evaluate the performance of our model in image quality, we use the metrics including bits per pixel (bpp), Peak Signal-to-Noise Ratio (PSNR) [55], Multi-Scale Structural Similarity Index Measure (MS-SSIM)[56].”
Can You Find the Difference? Visually Identical Image Detection
Jin, H. H., Feng, Y., Zhao, J., et al. (2026) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw68408.2026.00135
Under-Canopy Terrain Reconstruction in Dense Forests Using RGB Imaging and Neural 3D Reconstruction
Sheffer, R., Pinchover, C., Zisman, H., et al. (2026) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw68408.2026.00148
Wang, Z., Li, G., Pan, F., Fan, Z., Cai, X., & Wang, D. (2026) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1016/j.cacaie.2026.100170
Self-citationEUMD: Event-Based Unrolled Motion Deblurring Via Learnable Prior Optimization
Zhang, H., Fu, D., Miller, J., Choudhury, A., Lee, C., & Monga, V. (2026) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw68408.2026.00074 · Cited by 1
VAOT: Vessel-Aware Optimal Transport for Retinal Fundus Enhancement
Dong, X., Zhu, W., Xiong, Y., Chen, X., Wang, H., Li, X., Cheng, J., Wang, Z., Tang, S., Dumitrascu, O. M., et al. (2026) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw68408.2026.00037
Self-citationBridging Restoration and Diagnosis: A Comprehensive Benchmark for Retinal Fundus Enhancement
Dong, X., Zhu, W., Chen, X., Wang, H., Li, X., Xiong, Y., Cheng, J., Wang, Z., Tang, S., Dumitrascu, O. M., et al. (2026) · IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw68408.2026.00035
“Building on OTT-GAN , OTE , and OTRE [39,40] incorporates the Multi-Scale Structural Similarity Index Measure (MS-SSIM) [2, 34] as a perceptual loss to improve the structural consistency of the translated images, alongside identity regularization to better preserve image content.”
Self-citationVFBGM: Very Fast Block-Wise Gradient Magnitude-Based Quality Metric for Infrared Images
Wu, Z., Shen, X., Chen, K., Zhang, Y., & Zhai, G. (2026) · Communications in computer and information science · doi:10.1007/978-981-95-9938-7_2 · Cited by 1
Wang, X., Li, K., Zhang, J., Gulomsafdarov, A., Xie, X. N., Zhang, K., Cai, Q., Liu, J., Wang, K., & Li, K. (2026) · IEEE Transactions on Communications · doi:10.1109/tcomm.2026.3721643
“…motion and adding the residual reconstruction ˜ x i, res : The overall loss function used to train our inverse model comprises the L 1 loss and the multi-scale structural similarity (MS-SSIM) loss [26], which is regarded as an effective combination for image restoration tasks as suggested in [27].”
A Neural Frame Streaming Framework with a Simultaneous Prior for End-cloud Collaborative Rendering
Fan, H., Yuan, Y., Ding, Y., Chen, Y., Bao, H., Huo, Y., & Wang, R. (2026) · ACM Transactions on Graphics · doi:10.1145/3832308
Pyramid Mamba with Multi-directional Scanning for Light Field Image Super-Resolution
Lyu, W., Ke, W., Sheng, H., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-9846-5_5
A Dual Cross-Attention-Based Dynamic Weighting Fusion Method for Cross-Pair Landsat and MODIS Images
Shuai, L., Jing, Y., She, X., et al. (2026) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2026.3722243
“The structural loss function is based on the multiscale structural similarity [58] and is formulated as where l M ( · ) denotes the luminance consistency at the M th scale; c j ( · ) and s j ( · ) represent the contrast similarity and structural similarity at the j th scale, respectively; and υ M ,…”
Álvarez-Narciandi, G., García-Fernández, M., Laviada, J., et al. (2026) · European Conference on Antennas and Propagation · doi:10.23919/eucap68105.2026.11612131
“The quality of these images has been also analyzed quantitatively using the multi-scale structural similarity (MS-SSIM) index [36], which was computed to compare them with a reference image of the target.”
Bench, C. A., Desai, V., Roozemond, C., van Engen, R. E., & Thomas, S. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-35393-1_17 · Cited by 2
“…FID are sensible by comparing them to changes in mean image-wise mean absolute error (MAE), the multi-scale structural similarity index (MS-SSIM) [21] between the original and augmented images (model-free metrics for content similarity, that encode information about structural preservation), and…”
García-Fernández, M., Álvarez-Narciandi, G., Álvarez, Y., et al. (2026) · European Conference on Antennas and Propagation · doi:10.23919/eucap68105.2026.11612640
“Besides the qualitative comparison among the retrieved radar images, the degradation caused by the aforementioned deviations and/or errors is quantitatively evaluated using a metric called structural similarity index measure (SSIM) [14], which measures the similarity between the input images.”
A 7-criteria evaluation approach for the responsible use of synthetic medical data
Zamzmi, G., Subbaswamy, A., Deshpande, R., Sharma, D., Delfino, J. G., & Badano, A. G. (2026) · BJR|Artificial Intelligence · doi:10.1093/bjrai/ubag005
Toward Quality Assessment of 3D Gaussian Splatting Coding
Prazeres, J., Mahmoudpour, S., Perry, S., et al. (2026) · Proceedings - International Conference on Image Processing · doi:10.1109/icip61757.2026.11630520
Sanz-Rodríguez, S., & Frydensbjerg, J. (2026) · Proceedings - International Conference on Image Processing · doi:10.1109/icip61757.2026.11629873
Daroch, M., Saldanha, A., Jha, R. R., et al. (2026) · Proceedings - International Conference on Image Processing · doi:10.1109/icip61757.2026.11630198
Das, S., Roy, S. D., Bhowmik, M. K., & Memon, N. (2026) · Machine Vision and Applications · doi:10.1007/s00138-026-01889-8
“0.6111 ± 0.0571 0.5899 ± 0.0119 0.6002 ± 0.0315 0.6105 ± 0.0216 0.5851 ± 0.0222 0.5802 ± 0.0421 0.5952 ± 0.0136 Traditional techniques, including Error Level Analysis (ELA) [42], ADQ [40], DCT [41], and NOI1 [44], depend on pixel-level inconsistencies, compression artifacts, or noise-based features.”
InfluentialWhen Restoration Becomes The Reference: Reusing Full-Reference IQA in Blind Settings
Sekhri, A., Sendjasni, A., Amirshahi, S. A., & Larabi, M. (2026) · Proceedings - International Conference on Image Processing · doi:10.1109/icip61757.2026.11629900
Matulin, M., Mrvelj, Š., Periša, M., et al. (2026) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-032-32862-5_3
Nupur, Kumar, N., Suhane, S., et al. (2026) · International Journal of Online and Biomedical Engineering (iJOE) · doi:10.3991/ijoe.v22i08.62380
Jiao, Y., Peng, H., & Li, L. (2026) · Image and Vision Computing · doi:10.1016/j.imavis.2026.106157
Neural Texture 4D Gaussian Splatting: Enhancing Expressive Dynamic Scene Rendering
Xu, J., Guo, J., & Zheng, L. (2026) · International Conference on Computer Communication and Artificial Intelligence (CCAI) · doi:10.1109/ccai69603.2026.11642023
“Following the common practice in high-fidelity reconstruction [14], [15], we utilize the L 1 color loss to minimize the pixel-wise difference between the rendered image ˆ I and the ground truth I .”
Self-Supervised Vision Transformers for Spacecraft Pose Estimation Across Domain Shift
Bergia, S., Caracciolo, I., Stesina, F., & Fantino, M. (2026) · IEEE International Workshop on Metrology for AeroSpace (MetroAeroSpace) · doi:10.1109/metroaerospace69299.2026.11646723
“MS-SSIM [12] computes luminance, contrast, and structure comparisons over M dyadic scales via a weighted product: Y (cid:2) (cid”
InfluentialVDRS-VQA: Viewport-Domain Residual Sampling for No-Reference 360$^\circ$ Video Quality Assessment
Wu, X., Yan, X., Yang, C., Yu, X., An, P., & Lee, S. (2026) · IEEE Signal Processing Letters · doi:10.1109/lsp.2026.3725499
“Early full-reference metrics, such as S-PSNR [2], WS-PSNR [3], CPP-PSNR [4], and MS-SSIM [5], compensate for ERP sampling non-uniformity by spherical sampling, latitude weighting, or alternative projection domains.”
Iqbal, K., Xue, Y., Rafique, A., Camacho, D., Ilyas, M., & Aurangzeb, K. (2026) · Expert Systems · doi:10.1111/exsy.70397 · Cited by 2
“This makes it more sensitive to perceptual distortions than pixel-wise errors and allows a finer comparison be-tween real and synthesised image sets (Wang et al. 2003).”
PMNet: A Positional Mamba Network for SAR-Assisted Cloud Removal
Zhou, Y., Xia, Y., Tang, H., et al. (2026) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2026.3726690
“The multiscale structural similarity loss L MS - SSIM promotes structural consistency and texture and boundary recovery [18].”
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.