No-Reference Quality Evaluation of Light Field Content Based on Structural Representation of The Epipolar Plane Image

Abstract

As an emerging technology, Light Field (LF) has gained everincreasing importance in the domain. In order to provide guidance for the development of perceptually accurate Light Field (LF) processing tools and supervise the entire streaming system, robust perceptual quality assessment metrics are required. Especially, No-Reference (NR) metrics are preferable to compare LF with different angular resolutions. Some metrics have been developed by extending commonly used 2D image quality metrics to the 4D LF domain with angular consistency terms. Nonetheless, although these models consistently show slightly improved performance, most of them are limited in evaluating the quality of LF using the sub-aperture views with additional terms on the angular domain. There is an evident lack of reliable quality metrics that are tailored to LF content. To remedy this lack, we propose a NR quality metric for LF contents based on representing EPI with structural descriptors, including the Histogram of Gradients and the Convolutional Sparse Coding based descriptors. The primary motivation resides in our observations that (1) LF related distortions on the angular domain are highly noticeable on the Epipolar Plane Image representations (EPI); (2) most of the distortions in EPI are structurerelated. Extensive experiments on the MPI-LFA [1] LF image quality dataset demonstrate that our method provides competitive performance with the state-of-the-art NR image quality metrics.

Blind Perceptual Quality Assessment of LFI Based on Angular-Spatial Effect Modeling

“…The traditional handcrafted feature-based blind LFIQA metrics [33], [34], [35], [39], [54], [55], [56], [57], [58], [59], [60], [61] generally extract angular and spatial NSS features, and then utilize non-linear regression models [62] to produce the quality score. For example, Shi et al [54] design a Blind quality Evaluator of Light Field image (BELIF), in which the principal component of cyclopean image array is firstly generated, then the naturalness and structural similarity index are extracted to assess the spatial and angular quality degradation, respectively.…”

PVBLiF: A Pseudo Video-Based Blind Quality Assessment Metric for Light Field Image

“…Among NSS-based NR LF-IQA metrics, some metrics [42], [43], [44], [45], [46] extract NSS features from the original 2D representations of LFI, e.g., SAI, RI, EPI, and MLI. Luo et al [42] utilize the information entropy of SAIs and naturalness distribution of MLI to measure the spatial quality and angular consistency of LFI, respectively.…”

“…Shi et al [43] combine the naturalness distribution features of the light field cyclopean image array and the global and local features of EPIs to measure the LFI quality. Ak et al [44] extract the structural features in EPIs using convolutional sparse coding and histogram of oriented gradients. In VBLFI [45], the NSS and energy features are extracted from the mean difference image and SAIs using curvelet transform.…”

Light Field Image Quality Assessment with Dense Atrous Convolutions

“…In this scenario, accurate light field image quality assessment (LF-IQA) methods are important tools that play a vital role in the design of these algorithms. In recent years, several LF-IQA methods have been developed, most of them relying on hand-crafted features extracted from EPIs [2,3,4]. These features are mapped onto the corresponding subjective mean opinion scores (MOS) using machine learning (ML) based regression algorithms.…”