Autostereoscopic video display with motion parallax

Abstract— A multi-viewer autostereoscopic display that provides motion parallax without encumbering the observer with special equipment is described. The display is built around a single liquid-crystal panel, from which multiple images are projected to a screen where they form the 3-D image. The display can be produced economically because it uses a single display panel and conventional optics. The primary advantage of this technique is its simplicity. Computer-generated images are supplied to the liquid-crystal panel with a single video board. 3-D television can be broadcast by a single unmodified television station (NTSC, PAL, SECAM, HDTV, etc.), and recorded and replayed in 3-D with a VCR. Applications include 3-D video arcade games, avionics, engineering workstations, scientific visualization, video phones, and 3-D television.     

A common approach to characterizing autostereoscopic and polarization‐based 3‐D displays

Abstract— Autostereoscopic and polarization‐based stereoscopic 3‐D displays recreate 3‐D images by providing different images in the two eyes of an observer. This aim is achieved differently for these two families of 3‐D displays. It is shown that viewing‐angle measurements can be applied to characterize both types of displays. Viewing‐angle luminance measurements are made at different locations on the display surface for each view emitted by the display. For autostereoscopic displays, a Fourier‐optics instrument with an ultra‐high‐angular‐resolution VCMaster3D is used. For polarization‐based displays, a standard Fourier‐optics instrument with additional glass filters is used. Then, what will be seen by an observer in front of the display is computed. Monocular and binocular quality criteria (left‐ and right‐eye contrast, 3‐D contrast) was used to quantify the ability to perceive depth for any observer position. Qualified monocular and binocular viewing spaces (QMVS and QBVS) are deduced. Precise 3‐D characteristics are derived such as maximum 3‐D contrast, optical viewing freedom in each direction, color shifts, and standard contrast. A quantitative comparison between displays of all types becomes possible.     

A 23‐in. full‐panel‐resolution autostereoscopic LCD with a novel directional backlight system

Abstract— An autostereoscopic display that shows stereoscopic images with full‐panel resolution has been developed,¹ but it has a problem in terms of unit size. To resolve this problem, a new directional backlight system was developed, and it was applied to a prototype autostereoscopic LCD. The backlight system has two light sources — one for the right eye and the another for the left eye — and an elliptically shaped mirror that controls the direction of light from the light sources. The LCD uses a field‐sequential method which re‐writes an image for one eye and one for the other eye at a frame rate of 120 Hz, and the light sources alternately blink in synchronization with each frame so that the LCD shows full‐panel‐resolution stereoscopic images without flicker. In this paper, the new backlight system is described. The backlight system is effective for large screen such as 23 in. on the diagonal. By using this backlight system, the prototype LCD achieved practible unit size, brightness over the entire screen, and cross‐talk.     

基于方向自适应采样搜索的快速运动估计算法

为了提高运动估计的搜索精度,提出了一种方向自适应采样搜索的运动估计算法。先利用搜索模板中块匹配误差的信息,自适应地采用两种三角搜索模板交替搜索,再针对大运动矢量的搜索,引入局部采样搜索策略。搜索过程中引入提前终止策略,提高了搜索速度。实验结果表明,在保证搜索速度的前提下,其峰值信噪比(PSNR)比全搜索算法平均降低了0.28dB,提高了搜索精度。     

A 6:1 image‐compression method using directional prediction for LCD overdrive

Abstract— An overdrive technology was developed and is widely used to diminish motion blur in LCDs. To store a previous frame in the overdrive operation in a limited‐sized memory, simple image‐compression techniques are required. By considering the strong correlation of nearby pixels in natural images, a new 6:1 color‐image‐compression method based on directional prediction is proposed. Different from the directional prediction of intra‐coding in H.264/AVC, the predictable direction is determined beforehand to minimize the computation complexity. A simple content‐adaptive quantization and bit‐streaming method, which preserves image details and is free from blocking artifacts, is also proposed. The experimental results show that our proposed method outperforms the vector quantization block truncation coding method with an average 3‐dB peak signal‐to‐noise ratio (PSNR) as well as the subjective quality in terms of blocking artifacts.     

Relationship between phoria and visual fatigue in autostereoscopic 3D displays

Many people complain about visual fatigue arising from viewing three‐dimensional (3D) displays. This paper investigates relationship between visual fatigue and viewers' phoria for viewing autostereoscopic 3D displays. Visual fatigue is evaluated through subjective symptoms with a questionnaire and optometric indicators comprising fusion range as well as accommodation convergence/accommodation (AC/A) ratio to measure the variation in visual functions. A screening test is adopted to divide the subjects into two groups based on whether they suffer from phoria. Then a 2 × 2 × 2 mixed design experiment is conducted with display type, viewing stage, and visual state as factors to examine visual fatigue during viewing session. The results show that phoria subjects obtain more severe visual fatigue than normal on subjective evaluation. The normal subjects reveal a more marked difference with phoria in fusion range and AC/A ratio after viewing 3D video clip. Fusion range can significantly distinguish between the two‐dimensional (2D) and 3D condition as well as between the pre‐ and post‐viewing stages. The sensitivity and specificity of fusion range is higher than AC/A ratio with respect to viewing of 3D contents, so it is more appropriate as an optometric indicator of visual fatigue for autostereoscopic 3D displays.     

Effect of number of views to the viewing experience with autostereoscopic 3‐D displays

Abstract— The observers' 3‐D viewing experience when the way the content is created and shown on an autostereoscopic 3‐D display alternate is evaluated. The observer's depth impression, and the perceived contour accuracy and image naturalness or peskiness of the content shown on a 3‐D display, has been investigated. In addition, the consequences of the way the content is created to the results from the optical characterization for the same display have been studied. The alternation of the content was realized in two different ways. Firstly, the number of views for creating the image was varied. Two, five, and 14 views were used; the main focus being on testing the same display and treating it as an ordinary two‐view and a 14‐view display with inter‐sub‐view crosstalk. Also, the intermediate condition where five views with non‐uniform view‐specific crosstalk were used has been investigated. Secondly, the way the content is created was varied by using images with computer‐generated content and photos. The effect of these parameters on viewing experience as such and especially the effect of 3‐D crosstalk on the viewing experience were studied.     

Human motion prediction for human-robot collaboration

In human-robot collaborative manufacturing, industrial robots would work alongside human workers who jointly perform the assigned tasks seamlessly. A human-robot collaborative manufacturing system is more customised and flexible than conventional manufacturing systems. In the area of assembly, a practical human-robot collaborative assembly system should be able to predict a human worker’s intention and assist human during assembly operations. In response to the requirement, this research proposes a new human-robot collaborative system design. The primary focus of the paper is to model product assembly tasks as a sequence of human motions. Existing human motion recognition techniques are applied to recognise the human motions. Hidden Markov model is used in the motion sequence to generate a motion transition probability matrix. Based on the result, human motion prediction becomes possible. The predicted human motions are evaluated and applied in task-level human-robot collaborative assembly.     

基于距离预测的快速自适应运动估计算法

已有的快速搜索算法中,绝大多数都不检查搜索区中所有候选项,所以,当视频序列中有快速或随机运动时,这些算法常导致搜索陷入局部最优。为了解决这一问题,提出了一种估算当前搜索点和最佳点之间的距离的度量方法,在此基础上提出了一种新的自适应的阈值方案,并结合层次搜索技术,既有效提高了具有快速或随机运动的视频的运动估计的搜索速度,也保证了算法较好的PSNR性能。     

A pattern-based prediction: An empirical approach to predict end-to-end network latency

Understanding latency in network-based applications has received considerable attention to provide consistent and acceptable levels of services. This paper presents an empirical approach, a pattern-based prediction method, to predict end-to-end network latency. The key idea of the approach is to utilize past history of latency and their variation patterns in latency predictions. After some preliminary study on simple numerical prediction models we examine the effectiveness of the proposed method with real latency data and various definitions of network stability. Our results show that the pattern-based method outperforms any single numerical model obtaining an overall prediction accuracy of 86.2%.     

Learning intentions for improved human motion prediction

For many tasks robots need to operate in human populated environments. Human motion prediction is gaining importance since this helps minimizing the hinder robots cause during the execution of these tasks. The concept of social forces defines virtual repelling and attracting forces from and to obstacles and points of interest. These social forces can be used to model typical human movements given an environment and a person’s intention. This work shows how such models can exploit typical motion patterns summarized by growing hidden Markov models (GHMMs) that can be learned from data online and without human intervention. An extensive series of experiments shows that exploiting a person’s intended position estimated using a GHMM within a social forces based motion model yields a significant performance gain in comparison with the standard constant velocity-based models.     

Method for predicting motion artifacts in matrix displays

Abstract— A method is proposed to measure and characterize motion artifacts in matrix displays. By using a fast, V(λ)‐corrected photodiode and a data‐acquisition system, accurate measurements of the temporal luminance behavior (step response) are recorded. The motion artifacts of LCDs and PDP displays are predicted from these measurements using the properties of the human‐visual system. The method is validated with perceptual evaluation experiments, for which a new evaluation protocol is established. In the end, new measures are proposed to quantify the motion‐rendering performance of these matrix displays.     

The effects of latency and motion blur on touch screen user experience

Touch interfaces are designed to be metaphors for physically manipulating virtual objects. However, this direct manipulation illusion is broken by the nonidealities of the device. We decouple and show the impact of touch‐to‐display latency and motion blur on the user experience of a touch system. Using a custom device, we achieve latencies down to ~1 ms, with frame rates up to 1000 Hz. Motion blur is reduced independently of frame rate using synchronized display modulation. Optimizing these metrics yields a dramatic improvement in the user experience, demonstrating the opportunity for higher performance in touch devices.     

基于深度学习的多目标运动轨迹预测算法

针对多目标运动轨迹预测过程中由于检测精度和实时性不足造成部分目标位置信息丢失和预测准确度不高问题,提出基于改进卡尔曼滤波的多目标轨迹运动轨迹预测(MMTP)算法。MMTP算法在目标检测阶段使用YOLOv4检测器提升目标检测的准确率和速度;在目标匹配阶段采用KM匹配算法将当前检测框的检测目标与上一时刻预测的预测框的目标进行数据关联,从而增强目标关联的准确性,避免目标遮挡、目标交错和漂移造成的目标丢失;在目标坐标预测阶段,提出改进卡尔曼滤波算法为每个运动目标预测下一帧位置坐标并画出预测框,提高非线性场景中目标坐标的预测精度,降低预测坐标的误差。使用MOT16与实际交通系统拍摄的视频序列数据集验证算法整体性能,仿真结果表明,MMTP在目标检测阶段具有较好的检测精度和速度,有效提升了算法整体的运行速度;在目标匹配阶段,MMTP算法能增强目标关联的准确性,减少目标丢失,比RMOT、POI、SORT、Deep-SORT和YVTP算法更优。     

低码率下残差值预测的快速分数运动估计

提出了一种在低码率下的快速分数运动估计算法。该算法利用相邻运动补偿块时域的相关性,预测当前最优分数匹配位置的残差值,继而判断是否终止当前分数点搜索。根据实验结果,所提出的方法在不损失任何率失真性能的情况下,对于常用的几个QIF和CIF标准测试序列,其平均的分数搜索点数可减少到6以下。与全分数运动估计方法（FFPS）相比,所耗费的搜索点数减少了近一倍;而与目前常用的快速算法相比,所耗费的搜索点数也减少了50%～80%。     

双差值局部方向模式的人脸识别

针对差值局部方向模式（DLDP）特征提取不够充分和对光照、噪声等比较敏感的问题,提出一种双差值局部方向模式（DDLDP）人脸识别方法。首先,分别将半径为1的3×3领域像素灰度值和半径为2的5×5领域像素灰度值与8个Kirsch模板算子卷积,得到两组对应8个灰度响应值。然后,将半径为1的灰度响应值,按照相邻前后作差的方式,得到8个灰度响应差值,再将半径为1和2得到的灰度响应值上下作差,也得到8个灰度响应差值。最后,将两组灰度响应差值取绝对值,其最大绝对值所对应下标位置构成DDLDP码。仿真实验结果表明,相比同类基于局部方向模式的单一人脸识别算法,该方法具有更好识别效果。DDLDP更加完整地提取了人脸特征,且表现出对光照和噪声更好的鲁棒性。     

Analysing observer preferences when presenting a product in a rendered scene: 2D vs. autostereoscopic 3D displays

ABSTRACT

This research compares the way the image of a product included within a rendered scene shown on an autostereoscopic 3D display is rated versus the same image shown in a 2D display. The purpose is to understand the observer's preferences and to determine the features that a composition should have to highlight the product and to make its presentation more attractive to observers, thereby helping designers and advertisers who use both displays to prepare images to make them more effective when visually presenting a product.

The results show that observers like the images on autostereoscopic 3D displays slightly more than those presented by means of 2D displays. On both displays the product is perceived more quickly when it is larger than the other elements and is shown with greater chromatic contrast, but a composition is seen as more attractive when the chromatic relationship between all the elements is more harmonious.     

基于运动分析的运动目标实时跟踪系统

机器人的视觉伺服是机器人领域重要的研究方向.着力于提高机器人视觉反馈系统的实时性,提出了基于运动分析的的运动目标实时跟踪方法.该方法采用形态学方法标记连通域,并基于最小二乘法拟合运动轨迹曲线,预测下一时刻位置,设定连通域搜索范围,从而提高搜索速度.仿真结果表明:该方法处理效果良好,能满足后续系统实时性要求.     

融合时空图卷积网络与非自回归模型的三维人体运动预测

当前人体运动预测的方法大多采用基于图卷积网络的自回归模型,没有充分考虑关节间的特有关系和自回归网络性能的限制,从而产生平均姿态和误差累积等问题。为解决以上问题,提出融合时空图卷积网络和非自回归的模型对人体运动进行预测。一方面利用时空图卷积的网络提取人体运动序列的局部特征,可以有效减少三维人体运动预测场景中的平均姿态问题和过度堆叠图卷积层引起的过平滑问题的发生;另一方面将非自回归模型与时空图卷积网络进行结合,减少误差累计问题的发生。利用Human3.6M的数据集进行80 ms、160 ms、320 ms和400 ms的人体运动预测实验。结果表明,NAS-GCN模型与现有方法相比,能预测出更精确的结果。     

基于时空预测向量相关性的运动估计算法

针对UMHexagonS算法体现出来的问题,利用时间预测向量和空间预测向量的位置映射关系,提出了一种新的运动估计算法--基于时空预测向量相关性的运动估计算法。该算法首先在小范围得到最优点后,继续利用预测矢量的时空方向相关性进行特定方向的扩展搜索,避免了提前落入局部最优点,并减少了搜索点数,从而提高了搜索质量。实验结果表明,与UMHexagonS算法相比,该算法在保持码率基本不变的情况下,能有效地减少运动估计时间,并且能一定程度地提高单帧的峰值信噪比。关键词：