Effect of ambient illumination level on perceived autostereoscopic display quality and depth perception

Abstract

Twenty participants viewed real-world videos and still images with a small hand-held autostereoscopic device in three ambient illuminations with three display luminance levels. Perceived depth, naturalness, overall image quality, and viewing discomfort were evaluated. The results show that illumination and luminance levels influence perceived quality parameters, but perceived depth is less sensitive to changes in ambient illumination than other viewing quality parameters associated with 3D viewing. Viewing 3D scenes lasting up to 40 min may cause some mild eyestrain symptoms, but symptoms of visually induced motion sickness are rare. The use of autostereoscopic hand-held devices therefore seems feasible in different contexts with different display luminance levels, and the overall viewing experience is relatively comfortable for most of the adult viewers.

Introduction

As stereoscopic displays become more common, the issues related to various possible uses become more relevant. For example, if a mobile device has an autostereoscopic display, the display should be usable in various illuminance conditions ranging from a dim living room in the evening to bright sunlight in daytime. However, it is well known that a large difference between the display luminance and the illumination level of the environment can impair the visibility of the display, leading to sensations of discomfort or transient adaptation effects from fixating back and forth between two luminance levels (e.g., [1], [2], [49]).

Chakrabarti et al. [3] observed that room illuminance affected the intensity ratio of maximum to minimum monitor luminance. According to these authors, the consequent loss in the contrast ratio causes image contrast degradation. Even though a high contrast ratio is achieved when a very low monitor surround luminance of 0–5% of the maximum is used in low ambient light of five lux or less, an increase in room illuminance can drastically reduce image contrast. Sheedy et al. [1] tested the effects of the luminance surrounding a computer display and showed that performance decreased at lower levels of surround luminance whereas higher surround luminance levels had no effect on transient adaptation. Optimal performance was attained for surround luminance levels that were equal to or higher than the display luminance. When participants were able to adjust the surround luminance level, they selected a luminance that was slightly less than the central luminance at 91 cd/m², but the preferred levels varied widely across the subject sample.

Näsänen and Ojanpää [4] studied performance with user interface icons and found that one’s perception of icons is quite resistant to small or moderate decreases in contrast and sharpness, but hypothesized that in performance, high illuminance outdoor conditions might have a greater effect on icon perception. Boydstun et al. [16] showed that stereoscopic depth perception survives significant interocular differences in luminance levels, even up to differences of 60%, provided that display luminance was approximately 0.63 cd/m² or higher. The authors concluded that developers of stereo displays can use a fairly large range of interocular luminance levels and still achieve good stereo depth perception. Glabe and Baumann [14] tested the effect of luminance and contrast on stereoscopic acuity and found that moderate contrast (C = 0.5) produced significantly lower thresholds of binocular depth perception than did low (C = 0.05) and high (C = 0.95) contrast with a fixed luminance level (250 cd/m²), whereas different luminance levels (50, 250, and 1600 cd/m²) had no effect on the depth perception threshold. In addition, inadequate brightness and contrast may affect the perceptual cue integration process, which can lead to misinterpretation of perceived size, depth, and distance within a scene presented (e.g., [45], [44]).

Because most of the results on viewing experiences have used either large-size 2D displays, binocular displays or other than autostereoscopic displays (e.g., 3D TV, 3D cinema), we can assume that similar to those results, viewing a small-size autostereoscopic display in different environments with various display luminance levels may have some influence on perceived image quality and other user experiences [8], [47], [48].

The image quality of three-dimensional images is fundamentally more complex than that of two-dimensional images, as the depth dimension creates a much richer and more engaging experience for the participants, and multiple parameters, such as depth reproduction quality and 3D-specific impairments, may affect the final quality of viewing experiences [5], [6], [7], [8]. According to Seuntiëns [9], image quality, depth, naturalness, and visual comfort best describe the overall 3D visual experience (see also [10], [11]). Naturalness incorporates more image quality than does a viewing experience in which only the perceived image quality varies rather than the degree of depth. In cases where depth varies, naturalness takes better account of the added value of depth than does viewing experience. The difference between naturalness and quality as a subjective evaluation concept lies in the fact that naturalness refers to what observers perceive as a truthful representation of reality (i.e., perceptual realism), whereas perceived quality refers to a subjective preference scale. For example, IJsselsteijn et al. [12] reported that the use of disparity information not only increases the ratings of perceived depth, but also affects the naturalness of depth. Seuntiëns et al. [13] investigated the concepts of naturalness, viewing experience, and presence in relation to image quality, depth, and Ambilight technology in the context of 3D television. According to their results, the concept of viewing experience takes into account the quality level of the video as well as enhancements such as 3D and Ambilight. Moreover, depth and dynamic Ambilight provide the viewer with more sensory information, which results in an enhanced sensation of presence.

Several papers have connected stereoscopic viewing and large screen sizes to the enhanced sense of presence (e.g., [17], [18], [19], [21], [22], [23]) (a subjective experience of being in one place or environment, even when one is physically situated in another [20]), but only a few papers on presence-related experiences with small-size autostereoscopic displays have been published. Shibata et al. [8] showed that when viewing experiences of small-size 2D and 3D displays were compared, there was a clear increase in the sense of presence experienced when 3D display was used. Comparison of the small and large 3D screens showed that images on a small screen were described as clearly more realistic. Seuntiëns [9; see also [12], [13]] as well as Shibata et al. [8] have concluded that realism is one of the main factors that determine good stereoscopic image quality.

Thus, in addition to depth impression, overall image quality and other experiences, we used image naturalness in our studies to refer to the more realistic presentation associated with an increase in the sense of presence [43].

Many tasks in everyday life, such as doing close work, reading small print, or long-term computer-related work, can contribute to visual fatigue and induce symptoms such as eyestrain, watery or dry eyes, a feeling of pressure when the eyes are open, hot eyes, difficulty focusing or blurred vision, and headaches [24], [25], [26], [52]. In the context of 3D applications and devices, several research results have reported eyestrain and changes in visual functioning as a result of viewing stereoscopic presentations (e.g., [17], [27], [28], [29], [30]; for an overview, see [24], [31]). Yano et al. [27], for example, found that people experienced more eyestrain with dynamic 3D screens than with static ones. Kooi and Toet [6] found that even a limited vertical disparity, crosstalk, and blur could cause noticeable viewing discomfort while viewing stereoscopic static images. Among other recommendations, the authors suggested that lenticular screens should only be used to display stereo images with limited disparity to avoid viewing discomfort resulting from the combination of crosstalk and luster. Moreover, the results of Häkkinen et al. [32], who utilized a mobile phone with a parallax barrier autostereoscopic display, indicate that utilizing small disparities is an effective way to avoid visual strain.

In addition to eyestrain, immersion in a 3D VE could induce motion sickness symptoms, such as nausea and discomfort (see [32], [33], [34]). Häkkinen and colleagues [53] reported significantly higher sickness scores as a result of wearing a 3D head-mounted display than from a bi-ocular HMD or CRT display. Moreover, the nature of the task used to study subjective experiences clearly influences the outcome. For example, motion scenes can evoke illusory feelings of self motion (i.e. vection) and as a result, the user might experience queasiness or nausea [35], [36], [37]. On the other hand, motion scenes could enhance the presence experience and thus make the experience more realistic [38].

In summary, added depth information influences perceived image quality, but it also makes the images look more realistic or natural, which in turn may contribute to the sense of presence. Additionally, depth magnitude, binocular image imperfections, and the features of the tasks used may induce symptoms of eyestrain and visually induced motion sickness. Thus, it is important to ensure that new system and application features such as stereoscopic mode meet users’ expectations and cause no serious discomfort. Our main goal was to investigate different subjective experiences when a small-size autostereosopic display was used to view natural videos and still images. On a more detailed level, perception of depth impression, overall image quality, image naturalness, and discomfort were evaluated with different surrounding illumination and display luminance conditions.