Blind Stereoscopic Omnidirectional Image Quality Assessment Using Predictive Coding Hierarchy

Wei Zhou; André Kaup

cs.CVAug 28, 2026

Blind Stereoscopic Omnidirectional Image Quality Assessment Using Predictive Coding Hierarchy

Authors: Wei Zhou, André Kaup

Organizations: School of Computational and Mathematical Sciences, Cardiff University, CF24 4AG Cardiff, United Kingdom · Chair of Multimedia Communications and Signal Processing, Friedrich-Alexander-Universität Erlangen-Nürnberg, 91058 Erlangen, Germany

Abstract

Stereoscopic omnidirectional images (SOIs) have provided users with newly immersive quality of experience in virtual reality environments. However, developing efficient and accurate perceptual quality assessment metrics for SOIs remains challenging due to many factors such as freely changeable field of views and binocular vision. In this paper, based on the characteristics of the human visual system (HVS), we propose a Predictive Coding Hierarchy-inspired metric (PCH) for blind/no-reference stereoscopic omnidirectional image quality assessment. Motivated by the viewing process of SOIs, the proposed PCH includes a local cyclopean perception module, a global predictive perception module, and a visual quality regressor. First, observers browse different spherical sceneries from viewports, and aggregate the local visual information to infer the perceptual quality of SOIs. Therefore, we extract various viewports, followed by cyclopean conversion and saliency detection to approach the perception and attention of the human brain. After the local aggregation, viewers then infer the global scene in their minds. Based on the binocular mechanism, we fuse left and right views to perform predictive coding hierarchy modelling. Finally, the visual quality regressor is exploited to obtain the ultimate quality score related to both local and global perceptual cues. Extensive experiments demonstrate that the proposed PCH achieves competitive and consistently improved performance compared with state-of-the-art quality assessment methods.

Figures & tables

Fig. 1: Illustration of perceiving a stereoscopic omnidirectional image with predictive coding hierarchy. We adopt luminance fusion since distortions are more sensitive in luminance.

Fig. 2: The pipeline of our proposed PCH for blind stereoscopic omnidirectional image quality assessment.

Category	Methods	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
FR IQA	PSNR	0.847	0.862	0.399
	SSIM [ 14 ]	0.866	0.850	0.408
	FSIM [ 49 ]	0.883	0.845	0.420
	GMSD [ 50 ]	0.899	0.894	0.346
	MJ3DQA [ 25 ]	0.877	0.847	0.410
FR OIQA	S-PSNR [ 51 ]	0.856	0.870	0.389

TABLE I: Performance comparison with existing methods on SOLID database, where IQA includes methods for both planar and stereoscopic content.

Category	Methods	Symmetric Distortion		Asymmetric Distortion		JPEG Compression		BPG Compression
Category	Methods	SRCC $\uparrow$	PLCC $\uparrow$	SRCC $\uparrow$	PLCC $\uparrow$	SRCC $\uparrow$	PLCC $\uparrow$	SRCC $\uparrow$	PLCC $\uparrow$
FR IQA	PSNR	0.807	0.902	0.747	0.779	0.831	0.861	0.881	0.898
	SSIM [ 14 ]	0.823	0.889	0.778	0.823	0.898	0.921	0.901	0.920
	FSIM [ 49 ]	0.837	0.914	0.807	0.840	0.910	0.919	0.911	0.931
	GMSD [ 50 ]	0.851	0.927	0.830	0.844	0.908	0.937	0.917	0.931
	MJ3DQA [ 25 ]	0.803	0.940	0.783	0.822	0.870	0.912	0.875	0.925

TABLE II: Performance comparison under different distortion settings, including symmetric/asymmetric distortions and JPEG/BPG compression schemes. Note that IQA includes methods for both planar and stereoscopic content.

Methods	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
Local Cyclopean Perception Module	0.917	0.922	0.381
Global Predictive Perception Module	0.908	0.919	0.402
Proposed PCH	0.938	0.947	0.318

TABLE III: Ablation study on module contribution.

Methods	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
6 viewports (without saliency)	0.911	0.924	0.380
Proposed PCH (6 viewports)	0.938	0.947	0.318
20 viewports (without saliency)	0.922	0.929	0.374
Proposed PCH (20 viewports)	0.940	0.949	0.310

TABLE IV: Influence of viewport number and saliency modelling for the proposed method.

Methods	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
You [ 64 ]	0.814	0.830	7.746
Benoit [ 65 ]	0.878	0.881	7.061
Hewage [ 66 ]	0.899	0.902	9.139
MJ3DQA [ 25 ]	0.916	0.917	6.533
Chen [ 67 ]	0.891	0.895	7.247
Bensalma [ 68 ]	0.875	0.887	7.559

TABLE V: Performance comparison on LIVE Phase I.

Methods	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
You [ 64 ]	0.786	0.800	6.772
Benoit [ 65 ]	0.728	0.748	7.490
Hewage [ 66 ]	0.501	0.558	9.364
MJ3DQA [ 25 ]	0.889	0.900	4.987
Chen [ 67 ]	0.880	0.895	5.102
Bensalma [ 68 ]	0.751	0.770	7.204

TABLE VI: Performance comparison on LIVE Phase II.

Patch Sizes	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
64 × 64	0.913	0.933	5.842
128 × 128	0.918	0.934	5.901
256 × 256	0.905	0.923	6.313

TABLE VII: Influence of patch size in predictive coding modelling on LIVE Phase I.

Patch Sizes	SRCC $\uparrow$	PLCC $\uparrow$	RMSE $\downarrow$
64 × 64	0.896	0.910	4.656
128 × 128	0.913	0.922	4.405
256 × 256	0.907	0.920	4.425

TABLE VIII: Influence of patch size in predictive coding modelling on LIVE Phase II.

Explore similar work

Apr 27, 2026

Viewport-Unaware Blind Omnidirectional Image Quality Assessment: A Unified and Generalized Approach

Blind omnidirectional image quality assessment (BOIQA) presents a great challenge to the visual quality assessment community, due to different storage formats and diverse user viewing behaviors. The main paradigm of BOIQA models includes two steps, ie, viewport generation, and quality prediction, which brings an extra computational burden and is hard to generalize to other visual contents (eg, 2D planar image). Thus, in this paper, we make an attempt to solve these issues. First, we experimentally find that BOIQA can be formulated as a blind (2D planar) image quality assessment (BIQA) problem, ie, the first step - viewport generation - is no longer needed, which narrows the natural gap between BOIQA and BIQA. Then, we present a new BOIQA approach, which has three merits: ie, viewport-unaware - it accepts an omnidirectional image in the widely used equirectangular projection format as input without any transformation; unified - it can also be applied to BIQA; and generalized - it shows better generalizability against other competitors. Finally, we validate its promise by held-out test, cross-database validation, and the well-established gMAD competition.

Jiebin Yan, Kangcheng Wu, Jingwen Hou +3

School of Computing and Artificial Intelligence, Jiangxi University of Finance and Economics, Nanchang 330032, Jiangxi, China, and also with the Jiangxi AI Quality Testing and Inspection Center, Nanchang 330029, Jiangxi, China · School of Artificial Intelligence, Xidian University, Xi’an, China

Image Quality Assessment

Ask PDF

Mar 15, 2026

RL-ScanIQA: Reinforcement-Learned Scanpaths for Blind 360°Image Quality Assessment

Blind 360°image quality assessment (IQA) aims to predict perceptual quality for panoramic images without a pristine reference. Unlike conventional planar images, 360°content in immersive environments restricts viewers to a limited viewport at any moment, making viewing behaviors critical to quality perception. Although existing scanpath-based approaches have attempted to model viewing behaviors by approximating the human view-then-rate paradigm, they treat scanpath generation and quality assessment as separate steps, preventing end-to-end optimization and task-aligned exploration. To address this limitation, we propose RL-ScanIQA, a reinforcement-learned framework for blind 360°IQA. RL-ScanIQA optimize a PPO-trained scanpath policy and a quality assessor, where the policy receives quality-driven feedback to learn task-relevant viewing strategies. To improve training stability and prevent mode collapse, we design multi-level rewards, including scanpath diversity and equator-biased priors. We further boost cross-dataset robustness using distortion-space augmentation together with rank-consistent losses that preserve intra-image and inter-image quality orderings. Extensive experiments on three benchmarks show that RL-ScanIQA achieves superior in-dataset performance and cross-dataset generalization. Codes are available at https://github.com/wangyuji1/RLScanIQA.git.

Yujia Wang, Yuyan Li, Jiuming Liu +3

Victoria University of Wellington · The Hong Kong University of Science and Technology (Guangzhou) · University of Cambridge +1

Image Quality Assessment

Ask PDF

May 4, 2026

Pixel Perfect: Relational Image Quality Assessment with Spatially-Aware Distortions

Traditional image quality assessment (IQA) methods rely on mean opinion scores (MOS), which are resource-intensive to collect and fail to provide interpretable, localized feedback on specific image distortions. We overcome these limitations by shifting from absolute quality prediction to a relational and directional assessment. Our approach utilizes a self-supervised synthetic distortion engine to generate training data, eliminating the need for manual annotation. A distortion prediction network is trained with an anti-symmetric objective to produce spatially-aware, disentangled maps that identify the type, intensity, and direction of distortions relative to a reference image. Subsequently, a scoring network is trained via contrastive learning on ordinally ranked image sets to predict a relational quality score. Our method provides a more granular and interpretable approach to IQA for the targeted optimization of image processing algorithms without requiring any human-labeled quality scores.

Fadeel Sher Khan, Long N. Le, Abhinau K. Venkataramanan +2

The University of Texas at Austin Austin, TX, USA · Samsung Research America Plano, TX, USA

Image Quality Assessment

Ask PDF