采用分裂Bregman的遥感影像亮度不均变分校正

目的 遥感影像成像过程由于受到传感器自身以及其他一些外部环境因素的影响,往往会呈现出整体的亮度不均,导致遥感影像解译和制图精度的降低。因此,需要对遥感影像进行亮度不均匀校正,提高影像的质量。方法 感知驱动的亮度不均变分校正方法,是一种新型的单幅遥感影像亮度不均校正方法,它受人眼视觉系统特性的启发,能够在有效校正影像整体亮度的同时增强局部对比度。本文用分裂Bregman迭代实现了对感知驱动亮度不均变分校正模型的最优化求解,在实现对影像整体亮度不均校正的同时,大幅提高了计算效率。结果 模拟实验和真实实验结果均表明,采用分裂Bregman的亮度不均变分校正模型需要较少的计算时间,从效率上比采用最速下降法的校正模型提高了约67倍。结论 分裂Bregman方法能够有效求解感知驱动亮度不均变分模型,在保证校正结果整体亮度均匀,局部对比度增强的前提下,大大提高运算效率。     

结合图像融合与分割的快速去雾

目的 针对机器视觉系统在雾天条件下采集的图像存在对比度低、细节模糊的问题,提出一种结合图像融合与分割的场景复原方法。方法 基于光学反射成像的物理特性以及形态学运算分别获取雾气浓度的近似估计,计算图像的局部方差并利用加权融合的方法得出准确的大气耗散函数,通过分割雾气最浓区域或者天空区域求得精确的大气光值,最后由大气散射模型计算复原图像并进行亮度和色调的调整。结果 该方法可以有效避免光晕效应和天空颜色失真等不足,能快速复原场景的对比度和颜色。结论 实验结果表明,该方法的场景适应能力较强,复原效果和计算速度相比于前人的方法均有不同程度的提高。     

亮度—梯度联合约束的车牌图像超分辨率重建

目的 受成像距离、光照条件、动态模糊等因素影响,监控系统拍摄的车牌图像往往并不具备较高的可辨识度。为改善成像质量,提升对车牌的识别能力,提出一种基于亮度与梯度联合约束的车牌图像超分辨率重建方法。方法 首先充分结合亮度约束和梯度约束的优势,实现对运动位移和模糊函数的精确估计;为抑制重建图像中的噪声与伪影,基于车牌图像的文字化特征,进一步确定了亮度与梯度联合约束的图像先验模型。结果 为验证该方法的有效性,利用监控系统获得4组车牌图像,分别进行模拟和真实的超分辨率重建实验。在模拟实验中将联合约束图像先验重建结果与拉普拉斯、Huber-Markov（HMRF）以及总变分（TV）先验的处理结果进行对比,联合约束先验对车牌纹理信息的恢复效果优于其他3种常见图像先验;同时,在模拟和真实实验中,将本文算法与双三次插值、传统最大后验概率、非线性扩散正则化和自适应范数正则化方法的超分辨率重建结果进行比较,模拟实验的结果表明,在不添加噪声情况下,该算法峰值信噪比（PSNR）和结构相似性（SSIM）指标分别为35.326 dB和0.958,优于其他4种算法;该算法在真实实验中,能够有效增强车牌图像纹理信息,获得较优的视觉效果,通过对重建车牌图像的字符识别精度比较,本文算法重建结果的识别精度远高于其他3种算法,平均字符差距为1.3。结论 模拟和真实图像序列的实验结果证明,基于亮度—梯度联合约束的超分辨率重建方法,能够降低运动和模糊等参数的估计误差,有效减少图像中存在的模糊和噪声,提高车牌的识别精度。该算法广泛适用于因光照变化、相对运动等因素影响下的低质量车牌图像超分辨率重建。     

多尺度构图先验的显著目标检测

目的 针对基于对比度的显著检测方法,因忽略了特征的空间分布而导致准确性不高的问题,启发于边界先验关于图像空间布局的思想,提出构图先验的显著检测方法。方法 假定目标分布于三分构图线周围,根据相关性比较计算显著值。首先,对图像进行多尺度超像素分割并构造闭环图;其次,提取构图线区域超像素特征并使用Manifold Ranking算法计算显著目标与背景的分布;然后,从目标和背景两个角度对显著值进行细化并利用像素区别性对像素点的显著值进行矫正;最后,融合多尺度显著值得到最终显著图。结果 在公开的MSRA-1000、CSSD、ECSSD数据集上验证本文方法并与其他算法进行对比。本文方法在各数据集上准确率最高,分别为92.6%,89.2%,76.6%。且处理单幅图像平均时间为0.692 s,和其他算法相比也有一定优势。结论 人眼视觉倾向于在构图线周围寻找显著目标,构图先验是根据人眼注意机制研究显著性,具有合理性,且构图先验的方法提高了显著目标检测的准确性。     

背投式多投影系统中各向异性问题的研究

目的 针对背投式多投影系统中存在的各向异性问题,提出一套行之有效的解决方案.方法 首先,为了解决以静态观众为主体的多投影系统各向异性问题,基于反馈机制提出了以特定位置为中心的“最佳观测域”,并在该区域内实现了真正的视觉无缝画面校正;然后,基于视觉人体跟踪技术进一步提出了以特定交互者为中心的“动态最佳观测域”框架,并结合模糊预测控制思想从人体重心检测、跟踪和投影仪亮度实时补偿等3个方面对该框架内的各个组成模块的关键技术进行了阐述.结果 实验结果显示,利用交互者运动的时空连续性优化后的动态最佳观测域算法,具有较好的鲁棒性和实时性.同时,与其他类似方法相比获得了更好的视觉无缝校正效果,其中观测距离为1 m、3 m和5 m时,对应的亮度样本标准差分别为17.11、13.17和9.2.结论 提出的各向异性解决方案,将实时反馈信号引入校正过程中,有效提高了输出画面质量.通过实验测试和性能评估,验证了文中方法的可行性.     

多先验特征与综合对比度的图像显著性检测

目的 图像的显著性检测在计算机视觉中应用非常广泛,现有的方法通常在复杂背景区域下表现不佳,由于显著性检测的低层特征并不可靠,同时单一的特征也很难得到高质量的显著图。提出了一种通过增加特征的多样性来实现显著性检测的方法。方法 在高层先验知识的基础上,对背景先验特征和中心先验特征重新进行了定义,并考虑人眼视觉一般会对暖色调更为关注,从而加入颜色先验。另外在图像低层特征上使用目前较为流行的全局对比度和局部对比度特征,在特征融合时针对不同情况分别采取线性和非线性的一种新的融合策略,得到高质量的显著图。结果 在MSRA-1000和DUT-OMRON两个公开数据库进行对比验证,实验结果表明,基于多先验特征与综合对比度的图像显著性检测算法具有较高的查准率、召回率和F-measure值,相较于RBD算法均提高了1.5%以上,综合性能均优于目前的10种主流算法。结论 相较于基于低层特征和单一先验特征的算法,本文算法充分利用了图像信息,能在突出全局对比度的同时也保留较多的局部信息,达到均匀突出显著性区域的效果,有效地抑制复杂的背景区域,得到更加符合视觉感知的显著图。     

图像亮度线索下的单目深度信息提取

目的 深度信息的获取是3维重建、虚拟现实等应用的关键技术,基于单目视觉的深度信息获取是非接触式3维测量技术中成本最低、也是技术难度最大的手段。传统的单目方法多基于线性透视、纹理梯度、运动视差、聚焦散焦等深度线索来对深度信息进行求取,计算量大,对相机精度要求高,应用场景受限,本文基于固定光强的点光源在场景中的移动所带来的物体表面亮度的变化,提出一种简单快捷的单目深度提取方法。方法 首先根据体表面反射模型,得到光源照射下的物体表面的辐亮度,然后结合光度立体学推导物体表面辐亮度与摄像机图像亮度之间的关系,在得到此关系式后,设计实验,依据点光源移动所带来的图像亮度的变化对深度信息进行求解。结果 该算法在简单场景和一些日常场景下均取得了较好的恢复效果,深度估计值与实际深度值之间的误差小于10%。结论 本文方法通过光源移动带来的图像亮度变化估计深度信息,避免了复杂的相机标定过程,计算复杂度小,是一种全新的场景深度信息获取方法。     

视线追踪系统眼部特征检测及视线方向计算方法研究

基于单视觉主动红外光源系统,提出了一种视线检测方法．在眼部特征检测阶段,采用投影法定位人脸;根据人脸对称性和五官分布的先验知识,确定瞳孔潜在区域;最后进行人眼特征的精确分割．在视线方向建模阶段,首先在头部静止的情况下采用非线性多项式建立从平面视线参数到视线落点的映射模型;然后采用广义回归神经网络对不同头部位置造成的视线偏差进行补偿,使非线性映射函数扩展到任何头部位置．实验结果及在交互式图形界面系统中的应用验证了该方法的有效性．

     

使用密集弱注意力机制的图像显著性检测

目的 基于全卷积网络（FCN）模型的显著性检测（SOD）的研究认为,更大的解码网络能实现比小网络更好的检测效果,导致解码阶段参数量庞大。视觉注意力机制一定程度上缓解了模型过大的问题。本文将注意力机制分为强、弱注意力两种：强注意力能为解码提供更强的先验,但风险很大;相反,弱注意力机制风险更小,但提供的先验较弱;基于此提出并验证了采用弱注意力的小型网络架构也能达到大网络的检测精度这一观点。方法 本文设计了全局显著性预测和基于弱注意力机制的边缘优化两个阶段,其核心是提出的密集弱注意力模块。它弥补了弱注意力的缺点,仅需少量额外参数,就能提供不弱于强注意力的先验信息。结果 相同的实验环境下,提出的模型在5个数据集上取得了总体上更好的检测效果。同时,提出的方法将参数量控制在69.5 MB,检测速度达到了实时32帧/s。实验结果表明,与使用强注意力的检测方法相比,提出的密集弱注意力模块使得检测模型的泛化能力更好。结论 本文目标是使用弱注意力机制来提高检测效能,为此设计了兼顾效率和风险的弱注意力模块。弱注意力机制可以提高解码特征的效率,从而压缩模型大小和加快检测速度,并在现有测试集上体现出更好的泛化能力。     

融合GF-MSRCR和暗通道先验的图像去雾

目的 针对雾天图像高亮和雾浓区域中容易出现场景透射率值求取不准确,导致复原后的图像细节丢失、出现光晕现象、对比度和色彩难以满足人眼的视觉特性等问题,提出了一种融合引导滤波优化的色彩恢复多尺度视网膜算法（GF-MSRCR）和暗通道先验的图像去雾算法。方法 首先利用加权四叉树方法从最小通道图中快速搜索全局大气光值,再从图像增强角度应用GF-MSRCR算法初步估计场景透射率值,依据暗通道先验原理对最小通道图进行二次估测,根据两次求取结果按一定比例进行像素级图像融合,得到场景透射率估计值;利用变差函数修正估计值,经中值滤波进一步优化得到场景透射率的精确值,最后通过大气散射模型恢复雾天图像,调整对比度和恢复颜色后,得到了轮廓完整且细节清晰的无雾图像。结果 理论分析和实验结果表明,经本文算法去雾处理后的图像信息熵、对比度、平均梯度、结构相似性分别平均提升了7.87%、21.95%、47.73%和15.58%,同时运行时间缩短了53.22%,对近景、含小部分天空区域、含大片天空区域和含白色物体场景的多种类型雾天图像显示出较好的复原效果。结论 融合GF-MSRCR和暗通道先验的图像去雾算法能快速有效保留图像的细节信息、消除光晕,满足了人眼的视觉特性,具有一定的实用性以及普适性。     

基于单投影仪与柱面反射镜的沉浸感显示系统

在分析传统沉浸感显示系统优缺点的基础上,设计并实现了一种新型的沉浸感显示系统。该系统使用柱面反射镜对单投影仪投射光线进行反射,并在弧形背投幕上成像,能够获得无缝画面,实现了广角度虚拟场景的连贯显示。通过合理设计柱面反射镜形状,实现了投影画面在水平方向上的均匀放大。通过对图像进行预变形可以基本消除因投影幕曲率而产生的投影画面形变,同时采用背投技术使参与者在虚拟场景中活动更加自如。该系统克服了传统的多投影仪或多显示器沉浸感显示系统中存在的图像拼接问题,且易于构建,经实验验证能够获得良好的沉浸感显示效果。     

Immersive front-projection analysis using a radiosity-based simulation method

Video projectors are designed to project onto flat white diffuse screens. Over the last few years, projector-based systems have been used, in virtual reality applications, to light non-specific environments such as the walls of a room. However, in these situations, the images seen by the user are affected by several radiometric disturbances, such as interreflection. Radiometric compensation methods have been proposed to reduce the disturbance caused by interreflection, but nothing has been proposed for evaluating the phenomenon itself and the effectiveness of compensation methods. In this paper, we propose a radiosity-based method to simulate light transfer in immersive environments, from a projector to a camera (the camera gives the image a user would see in a real room). This enables us to evaluate the disturbances resulting from interreflection. We also consider the effectiveness of interreflection compensation and study the influence of several parameters (projected image, projection onto a small or large part of the room, reflectivity of the walls). Our results show that radiometric compensation can reduce the influence of interreflection but is severely limited if we project onto a large part of the walls around the user, or if all the walls are bright.     

Multi‐primary image projector using programmable spectral light source

This paper presents a practical prototype of a multi‐primary image projector system in which light source spectra can be programmable for suiting any purpose. Our multi‐primary projection system is mainly configured with a light source component and an image projection component. The programmable light source can reproduce any spectral curve. Spatial images are then generated using a digital mirror device chip that quickly controls the intensity of the light source spectra in 2D image plane. The multi‐primary images in our projection system are reproduced by multiplexing the time‐sequential images with different primary colors. Our multi‐primary image projector realizes not only wide gamut projection but also spectral projection. To achieve this, we also show how light source spectra of four or six primary colors are designed.     

多投影机非平面自动几何校正

非平面投影以其高度的沉浸感、多样的表现形式和灵活的场景布局,近年来越来越多地应用在虚拟现实、数字娱乐、展览展示等领域。非平面投影与平面投影相比,由于涉及到投影机位姿标定以及失真图像校正等问题,实现难度较高。目前的投影校正方法虽然可以通过重建或者非重建方法实现非平面屏幕投影的几何校正,但是它们多少都存在成本高昂,步骤复杂,需要人工干预,适用面窄等问题。针对非平面屏幕中最常使用的柱幕面型,提出一种非重建的多投影机投影几何校正方法,此方法可自动实现柱幕环境的几何投影校正,并且可对柱幕在加工中引入的倾斜、表面起伏等外形偏差予以修正,实现了复杂柱幕环境下多投影机高精度、短时间投影几何校正。提出的方法不仅适用于柱幕,还可扩展应用于其他可参数化描述的面型,如球面、双曲面、抛物面等面型的投影环境,具有广泛的适用性。     

Immersive virtual reality in the age of the Metaverse: A hybrid-narrative review based on the technology affordance perspective

Immersive virtual reality (VR) that utilizes head-mounted displays (HMD) is one of the key emerging technologies of the 21st century and has drawn keen attention from consumers, practitioners, and scholars in various disciplines. Nevertheless, the information systems (IS) discipline has neglected immersive VR, given that only a handful of studies have been published in mainstream IS journals. However, the recent advancements in immersive VR technology provide new opportunities for organizations and IS researchers. In light of these points, we reviewed the immersive VR literature to provide a holistic view of opportunities and challenges for organizations and future research directions for the IS field. By examining the technical capabilities of immersive VR and the previous literature, we identified five affordances: embodiment, interactivity, navigability, sense-ability, and create-ability. Our review of the 151 studies from the IS and related fields synthesized how these affordances were utilized in various research domains. Guided by the affordance-actualization theory, we also identified the strategic opportunities and challenges that come with implementing VR. The actualization of immersive VR affordances in organizations is indeed a fruitful area for IS scholars as there are various venues to move the IS field as well as the VR research in the organizational context forward.     

A VR Playground for Learning Abstract Mathematics Concepts

Immersive virtual reality (VR) has attracted the attention of many researchers and educators who predicted that VR would considerably affect how learning and teaching are conducted. The research presented here aims to investigate how an interactive immersive virtual learning environment affects conceptual learning, specifically learning of fractions in mathematics. A virtual environment (VE) designed to simulate a playground was created and evaluated through empirical studies with 60 primary school students between the ages of 8 and 12. Results suggest that children who fully interacted with the VE were able to problem-solve but that there was no strong evidence of the expected conceptual change. Rather, it was the passive VR environment, where a virtual robot guided activity, that seemed to support student reflection and recall, leading to indications of sustained conceptual change.     

Camera-based detection and removal of shadows from interactive multiprojector displays

Front-projection displays are a cost-effective and increasingly popular method for large format visualization and immersive rendering of virtual models. New approaches to projector tiling, automatic calibration, and color balancing have made multiprojector display systems feasible without undue infrastructure changes and maintenance. As a result, front-projection displays are being used to generate seamless, visually immersive worlds for virtual reality and visualization applications with reasonable cost and maintenance overhead. However, these systems suffer from a fundamental problem: Users and other objects in the environment can easily and inadvertently block projectors, creating shadows on the displayed image. Shadows occlude potentially important information and detract from the sense of presence an immersive display may have conveyed. We introduce a technique that detects and corrects shadows in a multiprojector display while it is in use. Cameras observe the display and compare observations with an expected image to detect shadowed regions. These regions are transformed to the appropriate projector frames, where corresponding pixel values are increased and/or attenuated. In display regions where more than one projector contributes to the image, shadow regions are eliminated.     

基于相机的多投影光度补偿系统

现有关于光度补偿的研究主要基于单投影,可在带有纹理和颜色的表面上进行投影显示。而在多投影系统中,由于多投影带来了重叠区域的光度不一致性,相机的每个像素覆盖的区域增大等问题,容易致使光度补偿失败。介绍一种基于相机的多投影光度补偿系统,虽然也是通过获取环境参数的手段,但提出一种方便快捷的迭代方式来提高投影表面反射率参数的获取精度,可以达到更好的效果,并仅需对投影区域捕捉两幅以上图像。同时,对多投影仪重叠区域的融合效果也做了改进,统一解决了光度补偿和投影拼接问题。最后,提供的实验结果表明有较好的改进效果。     

Immersive authoring of Tangible Augmented Reality content: A user study

Immersive authoring refers to the style of programming or developing content from within the targetexecutable environment. Immersive authoring is important for fields such as augmented reality (AR) in which interaction usability and user perception of the target content must be checked first hand, in situ. In addition, the interaction efficiency and usability of the authoring tools itself is equally important forease of authoring. In this paper, we propose design principles and describe an implementation of animmersive authoring system for AR. More importantly, we present a formal user study demonstrating its benefits and weaknesses. In particular, our results demonstrate that, compared to using the traditional 2D desktop development method, immersive authoring gained significant efficiency in specifying spatial arrangements and behavior tasks, a major component of AR content authoring. However, it was not so successful for abstract tasks such as logical programming. Based on this result, we suggest that a comprehensive AR authoring tool should include such immersive authoring functionality to help, particularly non-technical media artists, create effective contents based on the characteristics of the underlying media and interaction style.     

Distributed, multi-sensor tracking of multiple participants within immersive environments using a 2-cohort camera setup

Vision-based human motion tracking has gained significant interest in recent years, as the need for more intuitive human–computer interaction paradigms are sought after. Immersive environments, such as large-scale displays and virtual reality systems are particularly suitable for such interaction mechanisms, as they provide a controlled basis for experimentation and evaluation. However, despite the growing interest, human-motion tracking within immersive environments is still in its infancy. The lack of visible light and the presence of multiple people within such environments pose several challenges to the problem. In this research a robust framework for real-time, multi-camera human-motion tracking within an immersive environment is presented. The static nature of the environment is exploited to utilise a novel 2-cohort camera layout. This layout proves to be highly useful for tracking, providing several benefits over existing techniques, which treat all cameras as the same.