Image Retargeting Quality Assessment: A Study of Subjective Scores and Objective Metrics

Ma, Lin; Wang, Lin; Deng, Chenwei; Ngan, King Ngi

doi:10.1109/jstsp.2012.2211996

Cited by 129 publications

(62 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In Rubinstein's study [19], the authors used some objective distance metrics, such as bi-directional similarity [20], color layout descriptor [21], BDW [17], edge histogram [22], SIFTflow [23], and Earth-Mover's Distance [24], then estimated how well the objective metrics agree with the users' subjective preferences. Ma et al [25] conducted similar experiments. These two studies showed that the performances of the objective quality metrics for retargeted images are still not good enough.…”

Section: Related Workmentioning

confidence: 83%

Content-Aware Image Retargeting Incorporated with Letterboxing

Mishiba

Oyamada

Kondo

2017

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

SUMMARYConventional image retargeting methods fail to avoid distortion in the case where visually important regions are distributed all over the image. To reduce distortions, this paper proposes a novel image retargeting method that incorporates letterboxing into an image warping framework. Letterboxing has the advantage of producing results without distortion or content loss although being unable to use the entire display area. Therefore, it is preferable to combine a retargeting method with a letterboxing operator when displaying images in full screen. Experimental results show that the proposed method is superior to conventional methods in terms of visual quality measured by an objective metric.

show abstract

Section: Related Workmentioning

confidence: 83%

Content-Aware Image Retargeting Incorporated with Letterboxing

Mishiba

Oyamada

Kondo

2017

IEICE Trans. Inf. & Syst.

View full text Add to dashboard Cite

show abstract

“…In this experiment, the accuracy of the full reference image quality metric by using the dataset RetargetMe dataset [8] database and NRID database [7] is evaluated. The sample images from RetargetMe database are shown in Fig 3. For this dataset we have compared all the eight retargeted schemes used in [8].…”

Section: Resultsmentioning

confidence: 99%

Full Reference Image Quality Assessment of Perceptual Distortion based on Image Retargeting

Shigwan¹,

Birajdar²

2017

Proceedings of the International Conference on Communication and Signal Processing 2016 (ICCASP 2016)

View full text Add to dashboard Cite

Image quality assessment is used for automatic evaluation of the image accurately. In this article, objective reference image quality assessment metric is used to evaluate the quality of retargeted images. The metric measures perceptual geometric distortion and information loss of retargeted images using deformable spatial pyramid (DSP) algorithm. NRID and RetargetMe database are employed to measure the performance of the metric. Experimental results shows that the performance of DSP algorithm is superior thanScale-invariant feature transform (SIFT) in terms of runtime and accuracy. Image retargeting, fast dense correspondence, Image quality assessment, deformable spatial pyramid, full reference image quality assessment IntroductionNowadays, there is an increase demand in mobile devices like cellular phones and personal digital assistants. The images have high resolutions and various aspect ratios than the small screens of the mobile devices. To fit random size images of various aspect ratios, the image retargeting technique reduces/expands the images by automatically removing/adding less-important image plots thereby keeping the important features intact. This property is useful to create images of different sizes, which is adapted for various display devices like mobile, TV and personal digital assistants, etc. that usually have widely different resolutions. The common result to image retargeting is to distinctively rescale the source image according to the output screen size. Image quality is a characteristic of an image that estimates the image degradation of test image with a reference image. During various processing stages, digital images are subject to various types of distortions which degrade the visual quality of image. In this paper, in order to demonstrate the perceptual quality of retargeted image, a full reference image quality metric is presented. The two major distortions, the geometric distortion and information loss are observed to evaluate quality of retargeted images. These distortions are evaluated by using DSP algorithm. The contributions are  Estimation of the perceptual distortion, information loss distortion metric and fusion of both distortion metric of retargeting images is demonstrated in [3] using SIFT algorithm. To overcome the limitations like low accuracy and high execution time, DSP algorithm is used in place of SIFT. We show that, the performance of DSP algorithm is more effective than the SIFT algorithm. We analyze the rank correlation and error values using DSP algorithm for various retargeted schemes using In Section 2, an overview of the related work is discussed. In section 3, the system model is presented. In 4,5 and 6 sections the perceptual geometric distortion and saliency loss ratio metrics and the final quality metric which consists of fusion of two metrics is explained. In section 7 experimental results are represented and Section 8 concludes an article. Related Work

show abstract

“…It is the first in-depth perceptual study with a large number of users for image retargeting quality assessment. A different subjective study was proposed in [16], in which the user evaluation was carried out by simultaneous double stimulus for continuous evaluation [1] that scored only one retargeted image each time rather than pairwise comparison. Castillo et al [3] developed an image retargeting survey using eye tracking technology.…”

Section: Related Workmentioning

confidence: 99%

“…Recently, an aspect ratio similarity (ARS) metric [33] was proposed, which characterizes how the source image is resized into the target image by geometric changes and provides an efficient solution based on a Markov random field. Noting that human judgment of retargeting quality often involves multiple factors, several state-of-the-art methods combine several metrics that characterize different factors of image retargeting quality [16,17,12].…”

Section: Related Workmentioning

confidence: 99%

Learning to Rank Retargeted Images

Yang

Liu

Lai

2017

2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

Image retargeting techniques that adjust images into different sizes have attracted much attention recently. Objective quality assessment (OQA) of image retargeting results is often desired to automatically select the best results. Existing OQA methods output an absolute score for each retargeted image and use these scores to compare different results. Observing that it is challenging even for human subjects to give consistent scores for retargeting results of different source images, in this paper we propose a learningbased OQA method that predicts the ranking of a set of retargeted images with the same source image. We show that this more manageable task helps achieve more consistent prediction to human preference and is sufficient for most application scenarios. To compute the ranking, we propose a simple yet efficient machine learning framework that uses a General Regression Neural Network (GRNN) to model a combination of seven elaborate OQA metrics. We then propose a simple scheme to transform the relative scores output from GRNN into a global ranking. We train our GRNN model using human preference data collected in the elaborate RetargetMe benchmark and evaluate our method based on the subjective study in RetargetMe. Moreover, we introduce a further subjective benchmark to evaluate the generalizability of different OQA methods. Experimental results demonstrate that our method outperforms eight representative OQA methods in ranking prediction and has better generalizability to different datasets.

show abstract

Image Retargeting Quality Assessment: A Study of Subjective Scores and Objective Metrics

Cited by 129 publications

References 39 publications

Content-Aware Image Retargeting Incorporated with Letterboxing

Content-Aware Image Retargeting Incorporated with Letterboxing

Full Reference Image Quality Assessment of Perceptual Distortion based on Image Retargeting

Learning to Rank Retargeted Images

Contact Info

Product

Resources

About