多靶搜索中的扫视眼动研究

本文研究在二维多靶视觉搜索实验中的眼动。通过对成功扫视幅度的计算和分析，发现视力角（有用周边视网膜区域）在一定条件下随着靶干扰物的密度增大而减小，随着搜索场景对比度的增大而增大。熟练的搜索者比不熟练者呈现较大视力角；在显示屏作平面图形搜索场景，头部固定的试验条件下，绝大多数扫视幅度不大于８度。     

视、听、体感目标探测的EEG频谱和空间特异性分析

利用行为学和EEG频谱能量分析方法,考察大脑对视、听和体感单通道靶刺激探测的频段和空间半球分布.采集14位健康受试者的64导联脑电数据,实验材料分为视觉、听觉和体感觉的靶刺激和非靶刺激.分析了3个感觉通道下所有受试者靶刺激EEG Power显著强于非靶刺激的频谱和脑区分布,还比较了3个感觉通道靶刺激的行为学数据.结果表明,体感觉靶刺激探测的难度显著大于听觉和视觉靶刺激探测.delta和theta频段在三感觉通道靶刺激探测过程中都发挥主要作用,印证ERPs成份中的P300在三感觉通道靶刺激探测过程中发挥主要作用.大脑对近点靶刺激的探测具有明显的左半球优势.三感觉通道靶刺激探测的行为学结果(错误率)与EEGPower结果(显著性差异电极统计)趋于一致,表明EEG Power分析可作为视、听、体三通道认知任务难度分类的潜在电生理学参数.     

基于自适应T型和菱形搜索算法的CT图像的MAP重建

图像的超分辨率重建就是从一个低分辨率序列中重建出一幅高分辨率的图像.而运动估计则直接影响超分辨率重建的效果.我们提出一种基于物体边缘的自适应运动估计方法,对物体的边缘部分采用菱形搜索的方法,保证运动估计的精确度.对物体内部和背景部分采用自适应T型搜索和菱形搜索相结合的方法.保证运动估计的快速性.将这种运动估计的结果应用到最大后验概率(MAP)重建的过程中,可以保证图像重建算法的可靠性.通过对CT投影数据的分析并进行MAP重建,可以获得良好的高分辨率图像.     

体感电刺激诱发事件相关电位的研究

目的对视觉、听觉、体感3种不同模态下靶刺激诱发的事件相关电位(ERP)进行比较研究,探讨体感电刺激作为脑机接口(BCI)一种新的信号诱发模式的可能性,为基于体感ERP的BCI研究提供理论依据。方法选择17例视力或矫正视力正常、听力正常、躯体感觉正常且无任何大脑病史的被试者,其中男性8例,女性9例;年龄20~26岁,平均年龄22.6岁;均为右利手。分别记录17例健康的被试者在视觉、听觉、体感单通道靶刺激下诱发的脑电图;对3类靶刺激下ERP的时域参数(幅值、潜伏期)、行为学数据(反应时间、错误率)、脑源定位进行比较分析。结果 3类靶刺激模式下的ERP波形具有相似性,体感电刺激诱发的ERP幅值与视觉、听觉靶刺激相比无显著性差异;体感电刺激诱发ERP的峰值潜伏期显著长于视觉靶刺激;体感电刺激的反应时间显著长于视觉靶刺激,错误率也高于视觉、听觉靶刺激;体感电刺激诱发ERP的脑内源与视觉靶刺激相比具有相似性。结论相比于视觉、听觉靶刺激,大脑对于体感电刺激的探测难度高,敏感程度低;但从ERP的波形和幅值上看,体感电刺激可以诱发出稳定的、可被检测到的ERP波形,完全有可能应用于BCI系统作为一种新的ERP诱发模式。     

基于图像显著性的人工视觉图像处理策略

目的:在假体设备中的视觉信息处理模块引入适当的图像处理策略,优化低分辨率下的人工视觉信息,是解决假体植入者获取的视觉信息有限问题的一种可行方法。方法:基于全局亮度对比度特征的图像显著性检测算法（LC）,结合颜色空间变换和视觉注意力仿真处理模型,提出一种面向视网膜假体人工视觉信息处理策略。通过两个标准图像测试数据集对图像处理结果进行评估,同时对流行病学调查结果选取的盲人常用物体图片进行仿真。结果:评估结果验证了在提取图像前景方面与原始LC算法对比的优越性,仿真结果验证了在人工视觉条件下应用的可行性。结论:本文提出的策略有助于视网膜假体植入者在日常生活场景中更好地完成物体识别等基本视觉任务。     

听力障碍与听力正常儿童视觉注意技能比较

目的:考察听力障碍和听力正常儿童的视觉注意技能。方法:采用Toni智力测验筛选出16名听力障碍儿童和18名听力正常儿童为被试,采用DMDX系统呈现实验材料和记录被试完成视觉搜索任务的反应时和准确率。结果:听力障碍儿童的平均反应时长于听力正常儿童(1272.4±60.3/977.6±51.3,P=0.001);搜索项目大小与被试听力的交互作用及被试听力、搜索项目大小和搜索特征的交互作用均显著(F(2,29)=5.863,P=0.005;F(4,29)=3.14,P=0.017)。结论:听力障碍儿童搜索目标的速度慢于听力正常儿童;两组被试有相似的加工方式,即对单一的颜色特征的搜索进行了平行加工,而对方向和颜色的混合特征进行了系列加工。     

基于实例分割和显著性计算的人工视觉多目标优化处理

通过实例分割Swin-Transformer提取分割所有前景对象,融合亮度、大小和位置图像显著性特征,提出模拟人类视觉注意机制的多特征融合注意力层级计算模型,为不同级别的前景物体采用适合的光幻视分辨率和亮度表达,实现不同的刺激编码策略进行层级优化处理。通过人工假体视觉的仿真试验表明,在所提出的多目标层级优化表达策略下,试验被试完成多目标识别的准确率、识别时间表现具有一定的显著提升。利用深度学习实例分割技术,层级化光幻视编码以仿生人类视觉选择性注意,达到增强假体植入者在复杂场景下的多物体感知,为视觉假体图像信息编码和优化处理研究的发展与应用提供参考。     

情绪信息对不同知觉负载下视觉搜索绩效的影响

目的：探讨情绪信息对不同知觉负载下视觉搜索绩效的影响。方法：采用字母搜索任务,调节目标字母的灰度来操纵知觉负载,进行3（情绪面孔：积极、中性和消极）×9（知觉负载：目标字母灰度RGB 0,0,0～RGB 80,80,80）的被试内实验设计,记录被试在不同条件下的字母搜索正确率和反应时。结果：（1）相对于高知觉负载,低知觉负载条件下的视觉搜索正确率更高,反应更快;（2）情绪效价与知觉负载的交互作用显著。在低知觉负载水平下,积极面孔条件下的视觉搜索正确率更低;高知觉负载水平下没有表现出情绪效价的影响作用。结论：知觉负载和情绪信息会共同影响视觉搜索绩效。低知觉负载条件下积极情绪分心物表现出干扰优势。     

基于小波变换的听觉脑机接口技术研究

基于听觉的脑机接口系统为视力减退或者无法控制眼球运动的闭锁综合症患者提供了一种新的交流通道,使其可以与外界环境进行简单的交流.为了实现在线二分类听觉脑机接口系统,采用包含两种靶刺激、一种非靶刺激的三音oddball范式作为听觉刺激,通过一维离散小波变换对所得脑电数据进行单次样本特征提取,选取低频的特征向量,并采用支持向量机(SVM)进行目标与非目标的识别.结果表明,利用小波变换可以有效地单次提取特征向量P300,相当于20次靶刺激响应的叠加结果,最终目标识别正确率可高达80％以上,达到与视觉诱发的BCI模式可比的效果,可以用于二分类的脑机接口系统.     

美科学家制出“仿生眼”助盲人恢复视力

美国科学家说,将可在两年内提供“仿生眼睛”植入手术,帮助数百万盲人恢复视力。美国的研究人员已获准于两年内在五个治疗中心为50到70名病人安装这种“仿生眼睛”。以希腊神话中百眼巨人阿古斯(Agrus)命名的“阿古斯二型”系统利用一个安装在眼镜上的照相机,把视觉信号传送到眼睛里的电极。以前接受不够先进的人工视网膜移植手术的病人能够“看到”光线、影像和物体的运动。但图像不够清晰。一名失明者在1999年接受了这种手术,现在他上街时能够避开长的或较低的树枝,但看人时好像是看到一团黑影。不过美国加州大学的科学家说,他们研造的“…     

Efficiency of searchpatterns

Search theory deals with the efficiency of covering searchpatterns; a classical example is systematic row search with sweep spacing dependent upon window size. Human visual search is a function of contrast and size of the targets and of effective visual lobe size, this latter influenced by clutter and foveal load. Human searchpatterns often are optimal covers. Experimental results presented in this paper demonstrate that this is true for instrumental search with display windows controlled by human subjects and quantified with a formula for efficiency. Visual searchpatterns using free eye movements often develop complex irregular sequential searchpatterns.     

Differential influence of the visual framework on end point accuracy and trajectory specification of arm movements

In this study the influence of visual scene on both arm end point accuracy and spatial path kinematics was evaluated. Eight subjects, immersed in a virtual environment, were required to point to one of ten targets located at two distances and in five directions. Targets were presented in frameworks of different complexity. The simplest framework was constituted by a uniform background, the most complex framework was constituted by a perspective-arranged grid. In the other two conditions it consisted of lines having a direction parallel to either the subject's sagittal or frontal body axis. Movements were executed without vision of both target and framework. The results showed that pointing movements were hypometric in all conditions. No difference in end point localization was observed between movements executed after presentation of the simplest and the most complex scenes. However, hypometria significantly increased when the scene was formed by lines parallel to the subject's sagittal axis. Visual information on the scene was also used to specify hand path parameters. Trajectory curvature increased with decreasing complexity of the framework. Correspondingly, the pointing kinematics varied. Taken together, these results suggest that visual analysis of cues surrounding the target can influence both target localisation and hand path planning. However, scene complexity is directly related only to determining trajectory curvature. We conclude that planning an arm movement consists of at least two processes: target localisation and hand path specification. Environmental visual cues forming the scene are taken into account differently during the two processes.     

Visual search strategies are indexed by event-related lateralizations of the EEG

Two processes have been proposed for picking out information from a visual scene. A parallel process that detects salient features and a following serial process for higher order vision. However, this separation is still under dispute. The current study investigated whether event-related lateralizations (ERLs) of the electroencephalogram are a useful tool to examine these two processes. In a visual search for a colour- or form singleton or a conjunction target, reaction time (RT), P3 amplitudes and ERLs served as dependent variables. RT replicated earlier results for colour and conjunction targets. P3 amplitudes decreased and ERL latencies increased for these conditions. However, form singletons showed RT-, P3- and ERL-results comparable to conjunction targets. ERL-results differed in some conditions from RTs. The results suggest that target attributes alone cannot dissociate between different search strategies but showed that efficient as well as less efficient processes can be utilised for the same targets proposing that processing demands are determined by the inter-relation of target and the distractors.     

Electrophysiological revelations of trial history effects in a color oddball search task

In visual oddball search tasks, viewing a no‐target scene (i.e., no‐target selection trial) leads to the facilitation or delay of the search time for a target in a subsequent trial. Presumably, this selection failure leads to biasing attentional set and prioritizing stimulus features unseen in the no‐target scene. We observed attention‐related ERP components and tracked the course of attentional biasing as a function of trial history. Participants were instructed to identify color oddballs (i.e., targets) shown in varied trial sequences. The number of no‐target scenes preceding a target scene was increased from zero to two to reinforce attentional biasing, and colors presented in two successive no‐target scenes were repeated or changed to systematically bias attention to specific colors. For the no‐target scenes, the presentation of a second no‐target scene resulted in an early selection of, and sustained attention to, the changed colors (mirrored in the frontal selection positivity, the anterior N2, and the P3b). For the target scenes, the N2pc indicated an earlier allocation of attention to the targets with unseen or remotely seen colors. Inhibitory control of attention, shown in the anterior N2, was greatest when the target scene was followed by repeated no‐target scenes with repeated colors. Finally, search times and the P3b were influenced by both color previewing and its history. The current results demonstrate that attentional biasing can occur on a trial‐by‐trial basis and be influenced by both feature previewing and its history.     

Attentional selection and identification of visual objects are reflected by distinct electrophysiological responses

Lateralised ERP responses were measured over posterior visual brain regions in response to visual search arrays that contained one colour singleton. In the localisation task, responses were determined by the visual hemifield where this singleton was presented. In the discrimination task, they were determined by the singletons’ shape. While an N2pc component was elicited in an identical fashion in both tasks, a subsequent sustained contralateral negativity was consistently present at posterior sites in the discrimination task only. This dissociation demonstrates that these two activations reflect distinct visual processing stages. We suggest that while the N2pc reflects the ability of the visual system both to identify and localise a relevant stimulus in the scene, the late sustained activity reflects the subsequent in-depth analysis and identification of these stimuli.     

Top‐down control of audiovisual search by bimodal search templates

To test whether the attentional selection of targets defined by a combination of visual and auditory features is guided in a modality‐specific fashion or by control processes that are integrated across modalities, we measured attentional capture by visual stimuli during unimodal visual and audiovisual search. Search arrays were preceded by spatially uninformative visual singleton cues that matched the current target‐defining visual feature. Participants searched for targets defined by a visual feature, or by a combination of visual and auditory features (e.g., red targets accompanied by high‐pitch tones). Spatial cueing effects indicative of attentional capture were reduced during audiovisual search, and cue‐triggered N2pc components were attenuated and delayed. This reduction of cue‐induced attentional capture effects during audiovisual search provides new evidence for the multimodal control of selective attention.     

Effects on visual search of lesions of the superior colliculus in infant or adult rats

Summary The superior colliculus was removed from rats at either one or five days of age or in maturity. Four months later they were tested on two versions of a visual search task. Experiment 1 required animals to retrieve food pellets concealed in a depression in the top of identical narrow pillars arranged in an arena. Rats with lesions of the superior colliculus, regardless of the age at operation, showed a large number of return errors compared with sham-operated controls. Return errors were defined as occasions on which the animal returned to pillars that had previously been visited on that trial, before every pillar had been visited at least once. Experiment 2 compared the ability of infantand adult-operated animals to detect and locate a single, baited white pillar in an array of black ones. There were no group differences in response latencies to targets presented in the rostral visual field (within 40° of the midline). However, animals operated on in adulthood or at 5 days of age were slower than both sham-operated animals and animals operated on at one day of age in their responses to more peripheral targets. The latter two groups were indistinguishable.     

儿童视觉搜索的速度-正确率平衡

目的:本研究采用新的统计方法处理速度-正确率平衡问题。方法:提出了"真实能力"和"使用策略"两个指标,而不采用传统的根据正确率的指标删除某些被试后再做统计的方法。结果:在儿童视觉搜索能力的发展中,对字母和汉字的真实搜索能力的发展存在这样的规律:搜索字母的真实能力从8到9岁有显著的提高,搜索汉字则从10到11岁有显著的提高,对数字的搜索则从7到12岁都有稳步的提高;而在搜索策略的使用方面:对于字母搜索的策略的显著提高出现在9到10岁的过渡,对数字搜索的策略在11到12岁有非常显著的提高。结论:新的统计方法支持了视觉搜索是以平行加工为主,部分存在序列加工的理论,而且其中的序列加工只和被试的策略有关,而与其真实能力无关。     

Searching for two feature singletons in the visual scene: the localized attentional interference effect

The localized attentional interference (LAI) effect was investigated in a visual search task requiring participants to simultaneously monitor two spatially separated features from the same or different dimensions. In Experiment 1, the search type was blocked and targets were defined by fixed feature values in two dimensions (e.g., a yellow item and a circular item). In contrast, in Experiment 2, participants had to look for a color and a form singleton, with the exact feature values varying randomly across trials. In both experiments, reaction times (RTs) were generally slower when two features were CLOSE to, rather than DISTANT from, each other. Moreover, RTs to CLOSE stimuli increased as the search set size increased, while RTs to DISTANT stimuli were unaffected by set size. Experiment 3 also used a singleton search task, but with the two singletons defined either in different dimensions or in the same dimension. A larger interference effect for CLOSE, as compared to DISTANT, stimuli was found for cross-dimension than for intra-dimension targets. These findings suggest that neighboring items, irrespective of whether these items are from the same or different dimensions, interfere with each other in attentional selection, and that searching for two cross-dimension targets may engage a process of dimension switching to effectively solve the ambiguity of each item, especially when these items are close to each other.