Natural Image Quality Assessment Based on Visual Biological Cognitive Mechanism

Natural Image Quality Assessment Based on Visual Biological Cognitive Mechanism

Run Zhang (Communication University of China, Beijing, China) and Yongbin Wang (Communication University of China, Beijing, China)
Copyright: © 2019 |Pages: 26
DOI: 10.4018/IJSI.2019010101

Abstract

With the focus of the main problems in no-reference natural image quality assessment (NR-IQA), the researchers propose a more universal, efficient and integrated resolution based on visual biological cognitive mechanism. First, the authors bring up an inspiring visual cognitive computing model (IVCCM) on the basis of visual heuristic principles. Second, the authors put forward an asymmetric generalized gaussian mixture distribution model (AGGMD), and the model can describe the probability distribution density of the images more precisely. Third, the authors extract the quality-aware multiscale local invariant features (QAMLIF) statistic and perceptive from natural images and form quality-aware uniform features descriptors (QAUFD) based on clustering and encoding the visual quality features. Fourth, the authors build topic semantic model and realize the resolution with Bayesian inference with IVCCM, AGGDM and QAUFD to implement NR-IQA. Theoretical research and experimental results show that the proposed resolution perform better with biological cognitive mechanism.
Article Preview
Top

1. Introduction

The authors propose a more general-purpose, effective and integrated resolution to implement NR-IQA based on visual biologically cognitive mechanism, including a inspiring visual cognitive computing model (IVCCM) on the basis of visual heuristic principles, an asymmetric generalized Gaussian mixture distribution model (AGGMD), the quality-aware multiscale local invariant features (QAMLIF) and quality-aware uniform features descriptors (QAUFD), topic semantic model, and then the authors realize the resolution with Bayesian inference with IVCCM, AGGDM and QAUFD to implement NR-IQA, which perform better with theoretical research and experimental results.

1.1. Background

Images are visual basis that people can apperceive the real world. It is an effective way in which human beings can achieve, express and communicate information each other. With the development of images processing technology and visual cognitive scientific theory, much more progress has been made in NR-IQA applications. (Venkatanath,2015) put forward a novel no-reference perception-based images quality evaluator (PIQUE) algorithm on the basis of extracting local features just in spatial domain, imitated human behavior, and employed multivariate Gaussian model (MGM) as probability distribution function (PDF) of natural images features. (Zhao, 2015) proposed an efficient general-purpose blind/no reference image quality assessment (NR-IQA) algorithm with frequency domain features of phase congruency values and local spectral entropy values. In the inspired by inspired by the sparse representation of visual scenes in the primary visual cortex of the human visual system (HVS), (Priya,2016) presented a no reference image quality assessment (NR-IQA) method. (Plataniotis, 2016) put forward parametric models in which described the universal features of chromatic data from natural images, and computed the correlation of chromatic characteristics like hue, saturation, opponent angle and spherical angle between spatially adjacent pixels with the help of color invariance descriptors, and so on. All the algorithm played an advantageous role in images quality assessment with no reference.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 8: 4 Issues (2020): 1 Released, 3 Forthcoming
Volume 7: 4 Issues (2019)
Volume 6: 4 Issues (2018)
Volume 5: 4 Issues (2017)
Volume 4: 4 Issues (2016)
Volume 3: 4 Issues (2015)
Volume 2: 4 Issues (2014)
Volume 1: 4 Issues (2013)
View Complete Journal Contents Listing