Perceptual quality measurement of 3D images based on binocular vision

Abstract

Three-dimensional (3D) technology has become immensely popular in recent years and widely adopted in various applications. Hence, perceptual quality measurement of symmetrically and asymmetrically distorted 3D images has become an important, fundamental, and challenging issue in 3D imaging research. In this paper, we propose a binocular-vision-based 3D image-quality measurement (IQM) metric. Consideration of the 3D perceptual properties of the primary visual cortex (V1) and the higher visual areas (V2) for 3D-IQM is the major technical contribution to this research. To be more specific, first, the metric simulates the receptive fields of complex cells (V1) using binocular energy response and binocular rivalry response and the higher visual areas (V2) using local binary patterns features. Then, three similarity scores of 3D perceptual properties between the reference and distorted 3D images are measured. Finally, by using support vector regression, three similarity scores are integrated into an overall 3D quality score. Experimental results for two public benchmark databases demonstrate that, in comparison with most current 2D and 3D metrics, the proposed metric achieves significantly higher consistency in alignment with subjective fidelity ratings.

Stereoscopic image quality assessment by analysing visual hierarchical structures and binocular effects

“…2 in a simplified manner. As pointed out by several previous works [17, 18], V1 and V2 are mostly important for the design of IQA methods among them, not only owing to their huge sizes, but also because the quality‐related properties of V3, V4 and MT do not have fundamental differences compared to V1 and V2 [19]. Thus, the proposed method is designed largely relying on the interests of V1 and V2.…”

“…Serious visual discomfort inevitably leads to unsatisfactory visual experience and affects subjective perception upon image quality. The impact of visual discomfort is mentioned in several previous works in SIQA [4, 6, 17], but very rarely taken into quantitative computation [13]. In order to construct a comprehensive representation of stereoscopic image quality, influences by binocular rivalry and visual discomfort are both taken into consideration in this study.…”

An improved measurement model of binocular vision using geometrical approximation

“…It is widely used in industry, such as precision measurement of large scale, aviation and aerospace manufacturing, 3D surface topography measurement and so on [6]. One of the most important problems surrounding the system, including binocular stereo vision measurement, is how to improve its accuracy [7]. Many researchers have done a lot work in this field and many achievements have been obtained, which constitutes an active research area in vision metrology.…”

Quality of experience measurement for light field 3D displays on multilayer LCDs

“…A huge amount of data would be involved to provide a true 3D image with both correct focus and parallax cues, and this prevents these technologies from being widely accepted for daily use. In addition to these pure optical or physics‐based techniques, a computational 3D display technology known as compressive light field display has been investigated . In these multilayer light field displays, light emitted from a back display layer passes through one or more transparent display layers.…”

“…Different from some other works of the 3D content‐based image quality measurement, this paper only focuses on light field 3D display measurement and takes three‐layer light field 3D display as the example . In order to help validate the proposed measurement techniques, we built a uniform multi‐layer light field display (MLLFD) measurement platform with two different configurations: polarization‐based MLLFDs and attenuation‐based MLLFDs .…”