Visual Depth Research Articles

Image Memorability Prediction Using Depth and Motion Cues

Memorability is one of the intrinsic image properties, which enables one to quantify to what extent images are memorable to the human cognitive system. Many works have shed light on various visual factors that influence the image memorability. Recently, one such study showed the influence of image depth and motion information on image memorability [3] . Based on these findings, this article proposes a deep learning-based prediction model, which utilizes depth and motion cues to predict the image memorability scores. The proposed model contains three deep CNN networks. Each of these three networks is individually trained to utilize one of the three visual factors: 1) visual depth information; 2) optical flow information; and 3) fine-tuned scene- and object-related features. In the end, all three networks are ensembled to predict the final memorability scores for the given image. An extensive set of experiments are conducted on large scale image memorability data set LaMem. From the experimental results, it is observed that the proposed model performs better than the current state-of-the-art model [3] by 2.44%.

Longitudinal Changes in Optical Quality, Spatial Vision, and Depth Vision after Laser Refractive Surgery for Myopia.

Laser refractive surgery procedures are efficacious at correcting myopia, but they result in long-term deterioration of optical quality that affects monocular and binocular visual performance. The study shows that the optical quality of the two eyes needs to be similar to optimize binocular visual performance after surgery. This study aimed to systematically analyze longitudinal changes in optical quality, high- and low-contrast logMAR acuity and random-dot stereoacuity before and after three refractive surgery procedures for correcting myopia. A total of 106 subjects (laser-assisted in situ keratomileusis, 40; photorefractive keratectomy, 26; small-incision lenticule extraction, 40) with myopia and astigmatism ≤1.5 D participated in this prospective cohort-based study. All aforementioned outcome variables were measured in both eyes pre-operatively and 1 week and 1, 3, and 6 months post-operatively. Pre-operative myopic spherical equivalent of refraction (median [25th to 75th interquartile range], -6.4 D [-8.0 to -4.2 D] for laser-assisted in situ keratomileusis, -4.3 D [-5.0 to -3.5 D] for photorefractive keratectomy, -5.5 D [-6.5 to -4.3 D] for small-incision lenticule extraction) was corrected to within ±0.75 D of emmetropia in all cohorts up to 6 months post-operatively (P < .001). Higher-order wavefront aberrations, uncorrected high- and low-contrast logMAR acuity, and stereoacuity all worsened and remained so up to 6 months post-operatively, relative to pre-operative values (P < .001). Stereoacuity worsened with both interocular average and difference in the magnitude of higher-order aberrations, whereas logMAR acuities worsened only with interocular average of these aberrations (r ≥ 0.40; P < .01 for all). Although the refractive surgery procedures tested here correct myopia, they result in a sustained (up to 6 months) loss of optical quality and spatial and depth-related visual functions post-operatively. Both interocular average and difference in the eye's optical quality seem to impair binocular visual functions after refractive surgery for myopia.

Visual-Haptic Size Estimation in Peripersonal Space

In perceptual psychology, estimations of visual depth and size under different spatial layouts have been extensively studied. However, research evidence in virtual environments (VE) is relatively lacking. The emergence of human-computer interaction (HCI) and virtual reality (VR) has raised the question of how human operators perform actions based on the estimation of visual properties in VR, especially when the sensory cues associated with the same object are conflicting. We report on an experiment in which participants compared the size of a visual sphere to a haptic sphere, belonging to the same object in a VE. The sizes from the visual and haptic modalities were either identical or conflicting (with visual size being larger than haptic size, or vice versa). We used three standard haptic references (small, medium, and large sizes) and asked participants to compare the visual sizes with the given reference, by method of constant stimuli. Results show a dominant functional priority of the visual size perception. Moreover, observers demonstrated a central tendency effect: over-estimation for smaller haptic sizes but under-estimation for larger haptic sizes. The results are in-line with previous studies in real environments (RE). We discuss the current findings in the framework of adaptation level theory for haptic size reference. This work provides important implications for the optimal design of human-computer interactions when integrating 3D visual-haptic information in a VE.

Do visual performance and optical quality vary across different contact lens correction modalities in keratoconus?

PurposeWhile several advanced contact lens (CL) designs are commercially available for vision correction in keratoconus, their visual performance and optical quality, relative to each other and controls remains unclear. This prospective, crossover study tested the hypothesis that these CL designs would have a differential impact on visual performance and optical quality in subjects with advanced keratoconus, but not in early to moderate disease states. MethodsSpatial vision (logMAR acuity and contrast sensitivity), depth vision (stereoacuity) and optical quality (higher-order wavefront aberrations) were measured on 28 bilaterally mild to advanced keratoconics (age: 20–28yrs; 15 males), novice to CL wear, and in 10 age-matched controls using well-established psychophysical and aberrometry techniques. All data were collected on keratoconic subjects with their spectacles and with conventional RGP, Kerasoft®, Rose K2® and Scleral RGP® CL’s in randomized order, atleast a week apart from each other. ResultsAll outcome variables deteriorated with keratoconus severity and improved with CL wear, relative to spectacles (p < 0.05). This improvement was smaller for Kerasoft CL (p < 0.05) and higher but comparable for the other three CL designs (p = 0.3), across all disease severity. Visual functions and optical quality outcomes never reached control levels for any correction modality (p < 0.05). ConclusionVisual performance and optical quality in keratoconus does not appear to improve commensurately with the sophistication of CL design across disease severity. Non-visual factors like quality of CL fit, wearing comfort and cost may therefore drive the choice of CL dispensed in keratoconus more than the performance efficacy of these lenses.

A novel muscle-computer interface for hand gesture recognition using depth vision

Muscle computer Interface (muCI), one of the widespread human-computer interfaces, has been widely adopted for the identification of hand gestures by using the electrical activity of muscles. Altho ...

Effect of Shock and Vibration Loading on the Fracture Mechanisms of a VT23 Titanium Alloy

The structure nonuniformity a VT23 titanium alloy after shock and vibration loading is established to lead to larger plasticity of the material and changes in fracture mechanisms. Localization of deformation processes influences the sizes and quantitative distribution of tear dimples on the fracture surface. The approach was advanced to detect tear dimples on the fracture surface of the alloy, in particular after shock and vibration loading. The fractographic inspection was used to compute the dimple parameters, among them area, number, equivalent diameter, and visual depth. The above parameters for the conglomerates of detected dimples became the basis for statistical analysis and establishment of additional fracture mechanisms of the alloy at the microlevel. It provides the grounds the grounds for preliminary comparative evaluation of the cond of titanium alloys, modified by shock and vibration loading, using fractographic examination data.

Neurophysiological testing for assessing construction workers' task performance at virtual height

Falling from heights is the primary cause of death and injuries at construction sites. As loss of balance has a fundamental effect on falling, it is important to understand postural regulation behavior during construction tasks at heights, especially those that require precise focus in an upright standing position (therefore, a dual-task demand on focus). Previous studies examined body sway during a quiet stance and dual tasks to understand latent factors affecting postural balance. Despite the success of these studies in discovering underlying factors, they lack a comprehensive analysis of a task's simultaneous cognitive load, postural sway, and visual depth. To address this limitation, this paper aims to examine construction workers' postural stability and task performance during the execution of visual construction tasks while standing upright on elevated platforms. To that end, two non-intrusive neurophysiological tests, a hand-steadiness task (HST) and a pursuit task (PT), were developed for construction tasks in a virtual environment (VE) as performance-based means to assess the cognitive function of workers at height. Workers' postural stability was measured by recording the mapped position of the Center of Pressure (COP) of the body on a posturography force plate, and the postural sway metrics subsequently calculated. A laboratory experiment was designed to collect postural and task performance data from 18 subjects performing the two batteries of tests in the virtual environment. The results demonstrated a significant decrease in the Root-Mean Square (RMS) of COP along the anterior-posterior axis during the Randomized Pursuit Task (RPT) and maximum body sway of the center of pressure (COP) in the mediolateral direction during both tests. Also, subjects exposed to high elevation predominately exhibit higher accuracy for RPT (P-value = 0.02) and lower accuracy for HST (P-value = 0.05). The results show that the combination of elevation-related visual depth and low-complexity dual tasks impairs task performance due to the elevation-induced visual perturbations and anxiety-driven motor responses. On the other hand, in the absence of visual depth at height, high task complexity surprisingly improves the pursuit tracking performance. As expected, during both tasks, alterations in postural control were manifested in the form of a body sway decrement as a compensatory postural strategy for accomplishing tasks at high elevation.

Depth vision guided hand gesture recognition using electromyographic signals

Hand gesture recognition has been applied to many research fields and has shown its prominent advantages in increasing the practicality of Human-Robot Interaction (HRI). The development of advanced techniques in data science, such as big data and machine learning, facilitate the accurate classification of the hand gestures using electromyography (EMG) signals. However, the processing of the collection and label of the large data set imposes a high work burden and results in time-consuming implementations. Therefore, a novel method is proposed to combine the benefits of depth vision learning and EMG based hand gesture recognition. It is capable of labeling the class of the collected EMG data under the guidance of depth vision automatically, without consideration of the hand motion sequence. Finally, we demonstrated the proposed method for recognizing the ten hand gestures using a Myo armband. The experiment is set in a supervised learning way to evaluate the performance of the designed Hk-means algorithm. It shows that the proposed method can succeed in hand gesture recognition without labeling the data in advance.

Autonomous Reinforcement Control of Underwater Vehicles based on Monocular Depth Vision

Abstract In this paper, a monocular depth prediction based end-to-end reinforcement control framework is proposed for autonomous control of underwater vehicles in the unknown environment. In the control framework, with the input of camera sensor RGB videos, a monocular depth prediction network is proposed to generate underwater depth images and a sequential reinforcement learning controller is also developed for autonomous obstacle-avoiding navigation and movement control. Simulated and experimental results demonstrate that the proposed control scheme can achieve remarkable performance on collision-avoidance navigation and autonomous control in the unknown environment.

세계주의 모더니즘을 적용한 현대 패션디자인 연구 I : 바우하우스의 기초교육과정, 직물공방, 건축공방을 중심으로

The purpose of this study is to identify the characteristics of cosmopolitan modernism found in contemporary fashion design by analyzing how modernist and Bauhaus principles have been applied to fashion. The research methods used for this study are a literary review and content analysis. The results suggest that cosmopolitan modernism is a narrow interpretation of modernism along with ideational modernism, abstract modernism, functional modernism, pre-conscious modernism, symbolic modernism and idealistic modernism. It is a complex design style that captures reason and emotion in harmony by symbolizing the individual’s ideas through an inquiry of forms, colors, and material attributes. The basic design style of cosmopolitan modernism is based on Bauhaus’s preliminary course, a weaving workshop, and an architecture workshop. It has the following aesthetic characteristics: conceptual universality, perpetual simplicity, functional artistry, visual relativity, and integrative autonomy. After the middle of the 20th century fashion design has the following attributes of cosmopolitan modernism: ‘geometrical module’, ‘subversion and reconstruction of modernistic color’, and ‘conceptual material’—by applying the study of the forms, colors, and materials of basic design style. The design characteristics of cosmopolitan modernism identified through analyzing the work of representative designers—Paco Rabanne, Sonia Rykiel, Issey Miyake, Missoni, Emilio Pucci, and Pierre Cardin—are the following: (1) planar playfulness created by the planar composition of different color and material modules, (2) relative dynamics established through overlay and penetration of planar and three-dimensional modules, (3) convergent artistry realized by the harmony of planar and three-dimensional materials, (4) three-dimensional visual depth built by the repetition of shapes and color contrasts, and (5) conceptual functionality realized by the functional application of geometric modules. In conclusion, cosmopolitan modernism offers a framework for a design practice that reflects the nature and emotions of the individual.

Occluded Pedestrian Detection Based on Depth Vision Significance in Biomimetic Binocular

Pedestrian detection and tracking has become an important field in the field of computer vision research. However, the existing pedestrian detection algorithms have some problems, such as low accuracy and poor stability due to the similar background and overlapped occlusion interference. Therefore, an occluded pedestrian detection method based on binocular vision is proposed in this paper. We simulate the recognition of human brain and use the deep learning network MobileNet to detect and locate the initial pedestrians. Then, binocular depth is introduced as visual salience prior information, which solves the problem of identifying pedestrians with similar background and occlusion. The experimental results show that our pedestrian detection framework greatly improves the pedestrian error detection under similar background and occlusion conditions.

Combining Visual Contrast Information with Sound Can Produce Faster Decisions

Pieron’s and Chocholle’s seminal psychophysical work predicts that human response time to information relative to visual contrast and/or sound frequency decreases when contrast intensity or sound frequency increases. The goal of this study is to bring to the forefront the ability of individuals to use visual contrast intensity and sound frequency in combination for faster perceptual decisions of relative depth (“nearer”) in planar (2D) object configurations based on physical variations in luminance contrast. Computer controlled images with two abstract patterns of varying contrast intensity, one on the left and one on the right, preceded or not by a pure tone of varying frequency, were shown to healthy young humans in controlled experimental sequences. Their task (two-alternative, forced-choice) was to decide as quickly as possible which of two patterns, the left or the right one, in a given image appeared to “stand out as if it were nearer” in terms of apparent (subjective) visual depth. The results showed that the combinations of varying relative visual contrast with sounds of varying frequency exploited here produced an additive effect on choice response times in terms of facilitation, where a stronger visual contrast combined with a higher sound frequency produced shorter forced-choice response times. This new effect is predicted by audio-visual probability summation.

Visual Servo Tracking of Wheeled Mobile Robots With Unknown Extrinsic Parameters

In most visual servoing methods for wheeled mobile robots, the camera is supposed at the center of the robot, while it will bring much convenience for practical use when putting the camera a little away from the robot center. In this paper, to handle the challenge that is incurred by the uncalibrated translational camera-to-robot parameters, a novel visual servoing scheme is designed for wheeled mobile robots to track a given trajectory, which is defined by a sequence of images recorded previously. The trajectory tracking errors are defined with respect to both current and desired poses of the robot, and the open-loop error system is obtained after analyzing kinematic models of the visual robot system. Subsequently, to compensate both unknown extrinsic parameters and the visual depth, an adaptive visual servo tracking controller is elaborately designed, with disposing the nonholonomic motion constraint of the mobile robot. The tracking errors are strictly proven converging to zero asymptotically by employing Lyapunov techniques and Barbalat's lemma. Both simulation and comparative experiments are conducted to prove that the proposed strategy can drive the robot track the desired trajectory efficiently, even translational camera-to-robot parameters are unknown.

Population rate-coding predicts correctly that human sound localization depends on sound intensity.

Human sound localization is an important computation performed by the brain. Models of sound localization commonly assume that sound lateralization from interaural time differences is level invariant. Here we observe that two prevalent theories of sound localization make opposing predictions. The labelled-line model encodes location through tuned representations of spatial location and predicts that perceived direction is level invariant. In contrast, the hemispheric-difference model encodes location through spike-rate and predicts that perceived direction becomes medially biased at low sound levels. Here, behavioral experiments find that softer sounds are perceived closer to midline than louder sounds, favoring rate-coding models of human sound localization. Analogously, visual depth perception, which is based on interocular disparity, depends on the contrast of the target. The similar results in hearing and vision suggest that the brain may use a canonical computation of location: encoding perceived location through population spike rate relative to baseline.

SGANVO: Unsupervised Deep Visual Odometry and Depth Estimation With Stacked Generative Adversarial Networks

Recently end-to-end unsupervised deep learning methods have achieved an effect beyond geometric methods for visual depth and ego-motion estimation tasks. These data-based learning methods perform more robustly and accurately in some of the challenging scenes. The encoder-decoder network has been widely used in the depth estimation and the RCNN has brought significant improvements in the ego-motion estimation. Furthermore, the latest use of Generative Adversarial Nets(GANs) in depth and ego-motion estimation has demonstrated that the estimation could be further improved by generating pictures in the game learning process. This paper proposes a novel unsupervised network system for visual depth and ego-motion estimation: Stacked Generative Adversarial Network(SGANVO). It consists of a stack of GAN layers, of which the lowest layer estimates the depth and ego-motion while the higher layers estimate the spatial features. It can also capture the temporal dynamic due to the use of a recurrent representation across the layers. See Fig.1 for details. We select the most commonly used KITTI [1] data set for evaluation. The evaluation results show that our proposed method can produce better or comparable results in depth and ego-motion estimation.

Eye-head-hand coordination during visually guided reaches in head-unrestrained macaques.

Nonhuman primates have been used extensively to study eye-head coordination and eye-hand coordination, but the combination-eye-head-hand coordination-has not been studied. Our goal was to determine whether reaching influences eye-head coordination (and vice versa) in rhesus macaques. Eye, head, and hand motion were recorded in two animals with search coil and touch screen technology, respectively. Animals were seated in a customized "chair" that allowed unencumbered head motion and reaching in depth. In the reach condition, animals were trained to touch a central LED at waist level while maintaining central gaze and were then rewarded if they touched a target appearing at 1 of 15 locations in a 40° × 20° (visual angle) array. In other variants, initial hand or gaze position was varied in the horizontal plane. In similar control tasks, animals were rewarded for gaze accuracy in the absence of reach. In the Reach task, animals made eye-head gaze shifts toward the target followed by reaches that were accompanied by prolonged head motion toward the target. This resulted in significantly higher head velocities and amplitudes (and lower eye-in-head ranges) compared with the gaze control condition. Gaze shifts had shorter latencies and higher velocities and were more precise, despite the lack of gaze reward. Initial hand position did not influence gaze, but initial gaze position influenced reach latency. These results suggest that eye-head coordination is optimized for visually guided reach, first by quickly and accurately placing gaze at the target to guide reach transport and then by centering the eyes in the head, likely to improve depth vision as the hand approaches the target.NEW & NOTEWORTHY Eye-head and eye-hand coordination have been studied in nonhuman primates but not the combination of all three effectors. Here we examined the timing and kinematics of eye-head-hand coordination in rhesus macaques during a simple reach-to-touch task. Our most novel finding was that (compared with hand-restrained gaze shifts) reaching produced prolonged, increased head rotation toward the target, tending to center the binocular field of view on the target/hand.

Asymmetric time perception across visual depth planes and degrees of spatial certainty

Asymmetric time perception across visual depth planes and degrees of spatial certainty

Implantation of the First IMT Macular Lens in Age-Related Macular Degeneration (Literature Review)

This review presents the characteristics and results of clinical studies of patients with age-related macular degeneration implanted with the developed for this purpose first macular IOL – intraocular macular telescope (IMT). This lens was designed specifically for patients with the most severe or terminal form of age-related macular degeneration and is designed for monocular implantation to provide central vision while maintaining peripheral vision of the paired eye, which is important for maintaining the balance and orientation of patients. This device allows patients to see both in dynamic and static situations in the near, intermediate and far ranges. The disadvantages of this lens are a decrease in the visual field and depth of focus (which excludes its bilateral implantation), the need for a large (10–12 mm) incision for implantation, which can cause an increase in corneal astigmatism and the risk of complications, in particular, the pupillary block with an increase in intraocular pressure. There are also difficulties in the study of the fundus after its implantation to assess the small changes in the macula or to identify possible postoperative complications after cataract surgery (macular oedema, etc.). Also after the implantation of this lens, there is need for special programs of visual rehabilitation. The patient’s commitment to the rehabilitation measures for the adaptation of the central visual field of the operated eye with the peripheral vision of the second eye is crucial for success of the IMT macular telescope implantation procedure.

“Seeing” or “Being Seen”: Research on the Sight Line Design in the Lion Grove Based on Visitor Temporal–Spatial Distribution and Space Syntax

Research on the sight line design of the Classical Chinese Garden (CCG) is an important issue of CCGs’ sustainable development. Taking the Lion Grove as a case, GPS data loggers and questionnaires were employed to collect visitor temporal–spatial data and visiting motivations. We then calculated the “Revisiting Proportion” and “Average Speed” values. Furthermore, we selected the “Visual Control” values analyzed by Depthmap as an indicator of visibility. The statistical analysis of the relationship among “Revisiting Proportion”, “Average Speed”, and “Visual Control” values of each space showed that the spatial visual characteristic affected the visitor temporal–spatial distribution. Scenery spots in and around the large water pool, within one-step visual depth of each other, occupying the visual advantage of both “seeing” and “being seen”, can facilitate the transformation of sight lines and form the visual effect of “one step, one scene”. This research also proved that the sight line design of the Lion Grove was more intentional than random.

Linking perception to decision point complexity for adaptive indoor wayfinding support

Linking perception to decision point complexity for adaptive indoor wayfinding support