Blind Quality Assessment of Multiply Distorted Images Using Deep Neural Networks

Wang, Zhongling; Athar, ShahRukh; Wang, Zhou

doi:10.1007/978-3-030-27202-9_8

Cited by 9 publications

(19 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on neural network, Kaur A et al proposed a novel no-reference IQA method using canny magnitude and achieved excellent results on LIVE and TID2008 datasets, proving the efficiency of ANN [13]. Besides, Wang [16] and Sebastian [17] et al adopted deep learning to IQA and achieved favorable results. In remote sensing, studies usually focus on IQA of pan-sharpening images and hyperspectral images.…”

Section: Related Work and Novelties And Necessity Of The Studymentioning

confidence: 99%

A Unified Framework for Anomaly Detection of Satellite Images Based on Well-Designed Features and an Artificial Neural Network

Wang

You

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Nowadays, as the number of remote sensing satellites launched and applied in China has been mounting, relevant institutions’ workload of processing raw satellite images to be distributed to users is also growing. However, due to factors such as extreme atmospheric conditions, diversification of on-board device status, data loss during transmission and algorithm issues of ground systems, defect of image quality is inevitable, including abnormal color, color cast, data missing, obvious stitching line between Charge-Coupled Devices (CCDs), and inconstant radiation values between CCDs. Product application has also been impeded. This study presents a unified framework based on well-designed features an Artificial Neural Network (ANN) to automatically identify defective images. Samples were collected to form the dataset for training and validation, systematic experiments designed to verify the effectiveness of the features, and the optimal network architecture of ANN determined. Moreover, an effective method was proposed to explain the inference of ANN based on local gradient approximation. The recall of our final model reached 81.18% and F1 score 80.13%, verifying the effectiveness of our method.

show abstract

Section: Related Work and Novelties And Necessity Of The Studymentioning

confidence: 99%

A Unified Framework for Anomaly Detection of Satellite Images Based on Well-Designed Features and an Artificial Neural Network

Wang

You

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…A variety of metrics are used across these works to evaluate accuracy and transferability. These include Area under the Receiver Operating Characteristics (ROC) and Precision Recall (PR) Curves 10,18 , F-Scores 16, 17 , Pearson's Linear Correlation Coefficient (PLCC) and Spearman's Rank Correlation Coefficient (SRCC) 2,10,11,14,15,[19][20][21][22] , Root Mean Square Error (RMSE) 12,20 , post-FQA qualitative assessment 13,20 , and Rand Index 17 . The most popular metrics of this list, including PLCC, SRCC, ROC, and PR, are therefore evaluated in this paper to clearly compare the network performance.…”

Section: Figurementioning

confidence: 99%

“…Recently, deep learning models based on convolutional neural networks (CNNs) have emerged as viable FQA methods 2,[10][11][12][13][14][15][16][17][18][19][20][21][22][23] . Open source platforms such as HistoQC 13 , CellProfiler 3.0 17 and ImageJ 24,25 also leverage deep learning models for FQA.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Transferability of Deep Learning Models for Focus Quality Assessment in Digital Pathology

Miles

Hosseini

Tang

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Out-of-focus sections of whole slide images are a significant source of false positives and other systematic errors in clinical diagnoses. As a result, focus quality assessment (FQA) methods must be able to quickly and accurately differentiate between focus levels in a scan. Recently, deep learning methods using convolutional neural networks (CNNs) have been adopted for FQA. However, the biggest obstacles impeding their wide usage in clinical workflows are their generalizability across different test conditions and their potentially high computational cost. In this study, we focus on the transferability and scalability of CNN-based FQA approaches. We carry out an investigation on ten architecturally diverse networks using five datasets with stain and tissue diversity. We evaluate the computational complexity of each network and scale this to realistic applications involving hundreds of whole slide images. We assess how well each full model transfers to a separate, unseen dataset without fine-tuning. We show that shallower networks transfer well when used on small input patch sizes, while deeper networks work more effectively on larger inputs. Furthermore, we introduce neural architecture search (NAS) to the field and learn an automatically designed low-complexity CNN architecture using differentiable architecture search which achieved competitive performance relative to established CNNs.

show abstract

“…Recent deep neural network (DNN) based approach also enables joint training of features and quality predictor in an end-to-end fashion [7]. OU methods may employ NSS [8], distortion artifact detection [9] or local binary pattern [10] features, and may also make use of FR-IQA methods to annotate sample images for training [11], [12]. A specific type of NR-IQA methods of high relevance are NR image sharpness assessment (NR-ISA) methods that are designed specifically for evaluating image blur or sharpness.…”

Section: Introductionmentioning

confidence: 99%

Real Versus Fake 4k - Authentic Resolution Assessment

Shah

Akundy

Wang

2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

In recent years, the native 4K/Ultra High Definition (UHD) resolution has been trending towards the new normal of video content creation and distribution, but the practical pipelines of video acquisition, production and delivery often involve downscaling stages where the spatial resolution drops below the 4K level. Even though the video may be upscaled back to 4K/UHD resolution later, the content has lost its authentic resolution. This work aims at authentic resolution assessment (ARA). We first construct a database of over 10,000 real and fake 4K/UHD images. We then develop a two-stage ARA (TSARA) approach that classifies a video frame to have real or fake 4K resolution, where the first stage classifies local patches using a convolutional neural network (CNN), and the second stage aggregates local assessment into a global image level decision using logistical regression. Experimental results show that the proposed approach achieves high accuracy at low computational cost, and outperforms state-of-the-art no-reference (NR) image quality assessment (IQA) and image sharpness assessment (ISA) models. The built database and the proposed method are made publicly available 1 .

show abstract

Blind Quality Assessment of Multiply Distorted Images Using Deep Neural Networks

Cited by 9 publications

References 44 publications

A Unified Framework for Anomaly Detection of Satellite Images Based on Well-Designed Features and an Artificial Neural Network

A Unified Framework for Anomaly Detection of Satellite Images Based on Well-Designed Features and an Artificial Neural Network

Transferability of Deep Learning Models for Focus Quality Assessment in Digital Pathology

Real Versus Fake 4k - Authentic Resolution Assessment

Contact Info

Product

Resources

About