Image Quality Assessment by Comparing CNN Features between Images

Amirshahi, Seyed Ali; Pedersen, Marius; Yu, Stella X.

doi:10.2352/issn.2470-1173.2017.12.iqsp-225

Cited by 33 publications

(43 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The naïve approaches do this by simply computing mathematical distances between the images based on norms such as L 2 or L 1 , but these are well-known to be perceptually inaccurate [52]. Others have proposed metrics that try to exploit known aspects of the human visual system (HVS) such as contrast sensitivity [22], high-level structural acuity [52], and masking [48,51], or use other statistics/features [3,4,14,15,38,[44][45][46]53,54,57]. However, such hand-coded models are fundamentally limited by the difficulty of accurately modeling the complexity of the HVS and therefore do not work well in practice.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

PieAPP: Perceptual Image-Error Assessment Through Pairwise Preference

Prashnani

Cai

Mostofi

et al. 2018

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

230

192

View full text Add to dashboard Cite

The ability to estimate the perceptual error between images is an important problem in computer vision with many applications. Although it has been studied extensively, however, no method currently exists that can robustly predict visual differences like humans. Some previous approaches used hand-coded models, but they fail to model the complexity of the human visual system. Others used machine learning to train models on human-labeled datasets, but creating large, high-quality datasets is difficult because people are unable to assign consistent error labels to distorted images. In this paper, we present a new learning-based method that is the first to predict perceptual image error like human observers. Since it is much easier for people to compare two given images and identify the one more similar to a reference than to assign quality scores to each, we propose a new, large-scale dataset labeled with the probability that humans will prefer one image over another. We then train a deep-learning model using a novel, pairwise-learning framework to predict the preference of one distorted image over the other. Our key observation is that our trained network can then be used separately with only one distorted image and a reference to predict its perceptual error, without ever being trained on explicit human perceptual-error labels. The perceptual error estimated by our new metric, PieAPP, is well-correlated with human opinion. Furthermore, it significantly outperforms existing algorithms, beating the state-of-the-art by almost 3× on our test set in terms of binary error rate, while also generalizing to new kinds of distortions, unlike previous learning-based methods.

show abstract

Section: Introductionmentioning

confidence: 99%

“…compares our proposed dataset with the four largest existing IQA datasets 3. Our dataset is substantially bigger than all these existing IQA datasets combined in…”

mentioning

confidence: 99%

PieAPP: Perceptual Image-Error Assessment Through Pairwise Preference

Prashnani

Cai

Mostofi

et al. 2018

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

230

192

View full text Add to dashboard Cite

show abstract

“…Other IQMs based on modelling the low-level vision have also been proposed, such the spatial CIELAB (S-CIELAB) [5]. In recent years the use of deep learning has attracted attention of many researchers [6,7,8,9,10,11,12].…”

Section: Introductionmentioning

confidence: 99%

On the Use of a Convolutional Neural Network to Predict Perceptual Quality of Images without Reference for Different Viewing Distances

Chetouani¹,

Pedersen²

2019

2019 IEEE International Conference on Image Processing (ICIP)

Self Cite

View full text Add to dashboard Cite

A plethora of image quality metrics have been proposed in the literature. These metrics aims to estimate the perceptual image quality automatically. One important key aspect that the perceived quality is dependent on is the viewing distance from the observer to the image. In this study, we propose to consider this information by estimating the quality of a given image without a reference image for different viewing distances. For that, a Convolutional Neural Network (CNN) model was used in this study. Relevant patches are first selected from the image and they are then used as inputs to the CNN. The selection is here based on saliency information. The used CNN is composed of two outputs that correspond to the predicted subjective scores for two viewing distances (50 cm and 100 cm). Our method was evaluated using the Colourlab Image Database: Image Quality (CID:IQ) that provides subjective scores at two different viewing distances. The obtained results show the efficiency of our method.

show abstract

“…22 naive human observers (10 men and 12 women, age [20][21][22][23][24][25][26][27], following the recommendation of minimum 15 observers by CIE [59] and ITU [60], were invited to give perceptual ratings to the different levels sharpness of images using a five force-choice based category judgment. The five categories, bad, poor, fair, good, and excellent, were represented by numbers from 1 to 5.…”

Section: Methodsmentioning

confidence: 99%

“…Gradient Magnitude Similarity Deviation (GMSD) [16] computes a local quality map by comparing the gradient magnitude maps of the reference and distorted image, and uses standard deviation to obtain the final IQ score. Amirshahi et al [26] proposed an IQ metric based on features extracted from Convolutional Neural Networks (CNNs), which produced good results on different databases. Zhao et al [27] evaluated IQ metrics for perceived sharpness of projection displays, where the images were blurred, and found that SSIM, FSIM and VIF produced good results.…”

Section: ) Full Reference Metricsmentioning

confidence: 99%

Evaluation of image quality metrics for sharpness enhancement

Cheng

Pedersen

Chen

2017

Proceedings of the 10th International Symposium on Image and Signal Processing and Analysis

Self Cite

View full text Add to dashboard Cite

Abstract-Image quality assessment has become a meaningful research field due to the explosive growth of image processing technologies in imaging industries. It is becoming more usual to quantify the quality of an image using image quality metrics, rather than carrying out time-consuming psychometric experiments. However, there is little research on the performance of image quality metrics on quality enhanced images. In this paper, we focus on images that have been enhanced by sharpening. A psychometric experiment was designed with observers giving scores to different images enhanced by sharpening on a display in a controlled dark environment. The results showed that full reference image quality metrics performed well when sharpening did not improve the visual image quality, while in images where sharpening increased the visual quality the performance was lower. No reference image quality metrics show better predictions than full reference image quality metrics in most cases.

show abstract

Image Quality Assessment by Comparing CNN Features between Images

Cited by 33 publications

References 22 publications

PieAPP: Perceptual Image-Error Assessment Through Pairwise Preference

PieAPP: Perceptual Image-Error Assessment Through Pairwise Preference

On the Use of a Convolutional Neural Network to Predict Perceptual Quality of Images without Reference for Different Viewing Distances

Evaluation of image quality metrics for sharpness enhancement

Contact Info

Product

Resources

About