Visual Importance Pooling for Image Quality Assessment

By Anush K. Moorthy, A. Bovik. IEEE Journal on Selected Topics in Signal Processing, 2009

367 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/1b934b9618eb6d106e0e84c8e9b4dc1deed9f3a8

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. A. Bovik
2. Xiaokang Yang
3. Zhou Wang
4. Guangtao Zhai
5. Lei Zhang
6. Jongyoo Kim
7. Sanghoon Lee
8. Kede Ma

In Media Technology

1. A. Bovik
2. Zhou Wang
3. Guangtao Zhai
4. Xiaokang Yang
5. Ke Gu
6. Jongyoo Kim
7. Sanghoon Lee
8. Kede Ma

Citations in the corpus

• UGC-VQA: Benchmarking Blind Video Quality Assessment for User Generated Content (2020, methodology): “A total of eleven pooling methods were tested: three Pythagorean means (arithmetic, geometric, and harmonic mean), median, Minkowski (p = 2) mean, percentile pooling (20%) [84], VQPooling [82], primacy and recency pooling [85], hysteresis pooling [60], and our previously proposed”
• Blind Image Quality Assessment: From Natural Scene Statistics to Perceptual Quality (2011, background): “The set of features is the lowest 5% [50], [51] of the structural correlation values so obtained for each pair, leading to a total of fea-tures— .”
• Seven Challenges in Image Quality Assessment: Past, Present, and Future Research (2013, background): “In [166], Moorthy and Bovik incorporated both visual-fixation-based weighting and quality-based weighting into an IQA algorithm.”
• Learning without Human Scores for Blind Image Quality Assessment (2013, background): “On the other hand, it has been found that in an image, the predicted quality of the worst local areas has a good linearity to human perception [12, 24].”
• Video Quality Assessment on Mobile Devices: Subjective, Behavioral and Objective Studies (2012, methodology): “We evaluate three different methods of temporal pooling: (1) Mean, (2) Percentile pooling [11], [33], [34], and (3) Memoryeffect based pooling.”
• Using Free Energy Principle For Blind Image Quality Assessment (2015, methodology): “Also, SROCC has been widely used to search for the suitable parameters in some existing IQA approaches [6], [14], [61].”
• Fully Deep Blind Image Quality Predictor (2017, background): “It is well-known that severe distortions in an image heavily influence subjective impression [41].”
• Blind Image Quality Assessment: A Natural Scene Statistics Approach in the DCT Domain (2012, background): “Unlike much of the prior work on image/video quality assessment (QA) [1], [2], [4], [23], [24], we make little direct use of specific perceptual models such as area V1 cortical Fig.”
• Motion Tuned Spatio-Temporal Quality Assessment of Natural Videos (2010, background): “Important topics in these directions include scalability of VQA, utilizing models of visual attention and human eye movements in VQA [6], [78]‐[80], exploration of advanced spatial and temporal pooling strategies for VQA [ 80 ], reduced reference VQA, and no reference VQA.”
• From Patches to Pictures (PaQ-2-PiQ): Mapping the Perceptual Space of Picture Quality (2019, background): “Local picture quality is deeply related to global quality, although this relationship is not well understood [21], [22].”
• Multi-level Wavelet-CNN for Image Restoration (2018, neutral): “, Classic5 and LIVE1 [39].”
• Unsupervised feature learning framework for no-reference image quality assessment (2012, background): “In many image quality assessment algorithms [12, 18], “percentile pooling” has been adopted and it is motivated by the observation that humans tend to perceive “poor” regions in an image and these “poor” regions heavily affect the subjective impression.”
• Image Quality Assessment Based on Gradient Similarity (2012, neutral): “More details, comparison, and discussion of these performance measures can be found in [34] and [35].”
• Using labeled data to evaluate change detectors in a multivariate streaming environment (2009, neutral): “Although these complex networks have improved the accuracy, their segmentation speeds are very slow.”