2017
DOI: 10.1016/j.image.2016.12.001
|View full text |Cite
|
Sign up to set email alerts
|

Predictive no-reference assessment of video quality

Abstract: Among the various means to evaluate the quality of video streams, light-weight No-Reference (NR) methods have low computation and may be executed on thin clients. Thus, these methods would be perfect candidates in cases of real-time quality assessment, automated quality control and in adaptive mobile streaming. Yet, existing real-time, NR approaches are not typically designed to tackle network distorted streams, thus performing poorly when compared to Full-Reference (FR) algorithms. In this work, we present a … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
5
0

Year Published

2018
2018
2022
2022

Publication Types

Select...
7
1

Relationship

1
7

Authors

Journals

citations
Cited by 28 publications
(5 citation statements)
references
References 60 publications
(100 reference statements)
0
5
0
Order By: Relevance
“…In [37], the author uses deep convolutional neural networks (CNN) approach to measure the quality by integrating the feature learning and regression into one optimization process. Similarly, the author uses machine learning approach in [38] to combine a simple NR metrics approach to derive a predictive NR assessment metric. The algorithm obtained a correlation of over 97% correlation, but the algorithm has been tested up to the packet loss rate of 10% only.…”
Section: Metrics Based On Neural Network Approachmentioning
confidence: 99%
“…In [37], the author uses deep convolutional neural networks (CNN) approach to measure the quality by integrating the feature learning and regression into one optimization process. Similarly, the author uses machine learning approach in [38] to combine a simple NR metrics approach to derive a predictive NR assessment metric. The algorithm obtained a correlation of over 97% correlation, but the algorithm has been tested up to the packet loss rate of 10% only.…”
Section: Metrics Based On Neural Network Approachmentioning
confidence: 99%
“…It is challenging to distinguish features representing impairments from features that are part of the source content, and this is why NR metrics in general are not capable of predicting subjective video quality as accurately as state-of-the-art FR metrics. It has been shown that different distortion specific metrics (e.g., blur, blockiness and jerkiness) do not work well alone for NR assessment of global video quality, but they can be combined into a more accurate generic metric by machine learning [40]- [42]. Unfortunately, video quality metrics based on machine learning tend to be prone to irreproducibility, due to overfitting and the fact that the commonly used machine learning algorithms do not include a mechanism to ensure that different features are combined in a consistent manner [42].…”
Section: Objective Assessment Of Packet Loss Artifactsmentioning
confidence: 99%
“…To provide a solution to the conflicting requirements of accuracy and computational complexity, either subjective scores or objective FR metrics are typically predicted from both encoding and/or network-related NR/RR metrics in order to make real-time quality estimations [5,[10][11][12]. Most of these approaches are applied to the case of natural videos, i.e.…”
Section: Introductionmentioning
confidence: 99%
“…[13], often resulting in rather complex and computationally expensive relationships between the input features, i.e. the Quality of Service (QoS), and QoE [10,11,14,15]. However, the expectations of passive GVS are intrinsically different (e.g., more attention to moving objects, different perception of synthetic content, higher sensibility to fluidity...).…”
Section: Introductionmentioning
confidence: 99%