This Research Topic is part of a series with:
Scene-Dependent Image Quality and Visual AssessmentImage quality is defined as the degree of excellence an image conveys and can be assessed using both visual psychophysics and predictive modelling. Currently, there are many unresolved challenges in the assessment of image and video quality. For example, engineering-type models that account for system performance, fail to address the action of advanced scene content-aware algorithms which are incorporated into commercial imaging systems. Also, computational quality models which are derived from images of individual natural scenes, fail to directly relate to component parameters in the system hardware and the signal image processes. Imaging is a rapidly growing and diverse area of technology, with new developments being made in systems which are tailored for observers and those incorporating computer vision. Understanding and predicting how advances in this technology impact final image quality is clearly of high significance.
Accordingly, we specify that the aim of this Research Topic is to provide a platform for addressing and discussing the importance of image quality models which combine system performance properties (e.g. cameras, displays, VR headsets, etc), the properties of the vision system (human or non-human) and scene content. When modelling image quality for an observer, functions such as basic spatial and colour sensitivities are usually incorporated into the mathematical framework. In the past, these functions have not specifically been related to scene content, although some advances in measuring spatial sensitivity from natural scenes have recently been made. There are also psychological aspects which are worthy of further consideration when quality is assessed by observers. For example, how do other factors, such as aesthetic appreciation, individual preferences and attention span impact quality? The link between visual system properties and image aesthetics is currently unknown in the predictive modelling of image quality for computer vision and deep learning purposes. Whether we consider human or computer vision applications, robust image quality models, which fully integrate all relevant aspects of the imaging chain are clearly important in the development of relevant technologies.
Contributions are invited that address factors relating to the issues described above. We welcome original research articles, reviews and brief research reports. Contributions based on interdisciplinary research which combine imaging science, vision science and computer vision are particularly welcome.
Specific areas of interest to the Research Topic include (but are not limited to) the following;
? Recent developments in image quality and/or visual modelling which account for scene content.
? Recent developments in video quality
? Scene content modelling (for image/video quality purposes)
? Aesthetic aspects, including cultural differences.
? Computational aesthetics relevant to image quality
? Image quality modelling for computer vision systems