The influence of visual motion on fast reaching movements to a stationary object

One of the most important functions of vision is to direct actions to objects. However, every time that vision is used to guide an action, retinal motion signals are produced by the movement of the eye and head as the person looks at the object or by the motion of other objects in the scene. To reach for the object accurately, the visuomotor system must separate information about the position of the stationary target from background retinal motion signals-a long-standing problem that is poorly understood. Here we show that the visuomotor system does not distinguish between these two information sources: when observers made fast reaching movements to a briefly presented stationary target, their hand shifted in a direction consistent with the motion of a distant and unrelated stimulus, a result contrary to most other findings. This can be seen early in the hand's trajectory (approximately 120 ms) and occurs continuously from programming of the movement through to its execution. The visuomotor system might make use of the motion signals arising from eye and head movements to update the positions of targets rapidly and redirect the hand to compensate for body movements.     

Direct visuomotor transformations for reaching

The posterior parietal cortex (PPC) is thought to have a function in the sensorimotor transformations that underlie visually guided reaching, as damage to the PPC can result in difficulty reaching to visual targets in the absence of specific visual or motor deficits. This function is supported by findings that PPC neurons in monkeys are modulated by the direction of hand movement, as well as by visual, eye position and limb position signals. The PPC could transform visual target locations from retinal coordinates to hand-centred coordinates by combining sensory signals in a serial manner to yield a body-centred representation of target location, and then subtracting the body-centred location of the hand. We report here that in dorsal area 5 of the PPC, remembered target locations are coded with respect to both the eye and hand. This suggests that the PPC transforms target locations directly between these two reference frames. Data obtained in the adjacent parietal reach region (PRR) indicate that this transformation may be achieved by vectorially subtracting hand location from target location, with both locations represented in eye-centred coordinates.     

A Fast Moving Target Detection Tracking and Trajectory Prediction System for Binocular Vision

Aiming at achieving complex moving targets perception, this paper designs a novel fusion-information detection and tracking system for moving targets. The system includes three modules: target detection, target tracking and trajectory prediction. The target detection module can detect fast moving objects which can be used as a template. The target tracking module tracks the object and records its historical position. Finally, the trajectory prediction module can predict the trajectory of the moving object based on the historical position. After verification by the table tennis detection tracking and prediction binocular vision system, the system has high accuracy and speed in detection and tracking, and can basically meet the actual application needs.     

Integration of target and body-part information in the premotor cortex when planning action

To plan an action, we must first select an object to act on and the body part (or parts) to use to accomplish our intention. To plan the motor task of reaching, we specify both the target to reach for and the arm to use. In the process of planning and preparing a motor task, information about the motor target and the arm to use must be integrated before a motor program can be formulated to generate the appropriate limb movement. One of the structures in the brain that is probably involved in integrating these two sets of information is the premotor area in the cerebral cortex of primates. The lateral sector of the dorsal premotor cortex is known to receive both visual and somatosensory input, and we show here that neurons in this area gather information about both the target and the body part, while subsequent activity specifies the planned action.     

平面视觉跟踪系统的研究

介绍了一套自主研制的平面视觉跟踪系统的原理和实现．设计了基于目标速度反馈的PID控制算法,将系统的控制空间由非线性空间转换到线性空间来实现,解决了普通图象直接反馈控制方法由于非线性原因而产生控制效果差的现象．并通过静态目标锁定实验和动态目标跟踪对比实验,检验了该视觉跟踪系统和速度反馈PID控制算法的性能．实验中系统静态目标锁定过程时间约为0．48s,动态跟踪响应时间约为0．5s．     

监控场景下视频中全局移动对象的异常行为自动识别

传统异常行为识别方法容易受到外界环境的干扰,导致识别精度低下;且开销较高。为此,提出一种新的监控场景下视频中全局移动对象的异常行为自动识别方法。对异常行为特征进行提取,选用的特征为目标运动轨迹特征和外接矩形框宽高比特征,通过位置动态对监控场景下视频中全局移动对象运动轨迹特征进行提取,通过最小外接矩形框对移动对象进行标记。依据特征提取结果,通过异常测量函数实现监控场景下视频中全局移动对象的异常行为自动识别。实验结果表明,采用所提方法对全局移动对象的异常行为进行自动识别,不仅识别精度高,且开销较低。     

基于单目运动摄像机的三维物体运动参数估计

针对单目视觉监测系统摄像机和目标物体同时运动时,如何有效测量物体三维运动参数的难题,提出一种基于单目序列图像对插入虚拟视点的算法确定运动物体位姿及运动参数的方法。动摄像机连续采集运动物体的图像序列,根据摄影几何原理对摄像机和运动物体进行分析,结合本质矩阵的特性,估算空间物体运动参数。实验结果表明,该方法能够在为摄像机与目标物体同时运动的情况下,有效地估测出物体三维运动参数,且该方法简单有效,具有较高的精确度和准确性。     

生成式与判别式视觉目标跟踪算法综述

视觉目标跟踪是对目标位置、速度、运动轨迹等信息检测与预测技术。该技术融合了计算机视觉、图像处理、深度学习等众多领域技术。本文将对目标跟踪算法发展情况以及研究现状进行梳理。首先介绍目前常用的基准数据集;其次指出生成式算法与判别式算法差异;再对传统的生成式算法进行简单的分析总结;随后围绕算法框架分别介绍相关滤波框架、深度学习框架、孪生网络框架、Transformer框架的判别式算法并分析不同算法的优缺点;最后分析目前动态目标跟踪存在的问题并展望。     

红外图像序列运动小目标的空间定位

提出了一种利用透视投影变换和卡尔曼滤波对低信噪比红外图像序列运动小目标进行三维空间定位的新方法,从而达到对红外运动小目标(直至几个像素)进行空间定位的目的.仿真实验结果证明了该方法的可行性和有效性.     

基于视觉目标跟踪的侦察机器人导航方法

针对未知环境下侦察机器人的自主导航问题,提出了一种基于视觉目标跟踪的侦察机器人导航方法.首先利用二进制鲁棒独立元素特征(BRIEF)提取方法来检测和描述待跟踪视觉目标的局部不变特征点,在快速的特征匹配计算基础上提出由粗到精的目标定位两步法实现机器人导航过程中视觉目标的实时准确跟踪.其次对基于视觉目标跟踪的自主导航任务进行行为分解和实现,在行为中集成视觉目标跟踪算法.最后利用基于宏行为的机器人事务执行机制实现移向视觉目标的自主导航控制.实验结果表明,提出的方法能够使侦察机器人实时准确地跟踪视觉引导目标,在复杂障碍物环境下可靠地完成移向目标的自主导航任务.     

一种基于改进视觉注意模型和局部自相似性的目标自动检测算法研究

针对基于视觉注意模型的检测算法只能检测到图像中的感兴趣区域,无法准确地提供目标的轮廓和位置的不足,提出了一种基于改进视觉注意模型和图像局部自相似性的目标自动检测算法。通过增加运动速度和运动方向特征改进了经典的Itti视觉注意模型。利用改进的视觉注意模型提取感兴趣区域,提高了视觉注意模型的检测能力。再利用图像在边缘处具有良好的局部自相似性,实现了基于图像局部自相似性的目标检测算法。实验表明,算法能快速检测到图像中的目标感兴趣区域,并对其进行精确分割和定位。     

交通车辆轮廓跟踪算法研究及其工程应用

针对模式识别在智能交通领域的实际工程应用,提出了一种提取运动车辆轮廓线的精细跟踪算法．首先,通过冗余离散小波变换法提取运动区域,检测出相邻两帧图像内的运动变化从而确定运动对象的存在及其初始位置;其次,以当前帧运动区域为参考,通过改进的mean-shift算法在后续帧中跟踪运动对象的中心位置;最后,以mean—shift跟踪窗口作为目标初始轮廓线,采用自适应水平集法得到目标轮廓,从而精确定位运动对象位置．实验结果表明本文算法能够以轮廓线的方式以较高精确度跟踪运动车辆目标,目前已被市交通局科研单位采纳,具有一定的工程应用前景．     

基于双目视觉的移动机器人动态目标识别与定位

提出了一种双目移动机器人实时动态目标识别与定位方法。该算法首先采用SIFT(Scale Invariant Features Transforms)算法提取目标特征,并结合双目视差特征进行目标匹配;然后通过区域增长方法进行目标区域的提取;最后结合双目视觉标定的模型对目标进行定位。实验结果表明:该方法在摄像机运动-目标运动情况下,能对局部特征未知或特征不明显的动态目标进行有效的识别与定位。     

运动对颜色信息整合的影响源于眼动

运用斜向运动及变速率运动实验模式,考察了运动对颜色信息整合的影响.实验发现:原先在眼动跟踪斜向运动刺激物和固视匀速运动刺激物时,都能发生的颜色信息整合现象,在分别变为固视斜向运动和变速运动刺激物时均不再发生.这表明,颜色整合起源于高级视觉皮层的流行观念是错误的.     

卡尔曼滤波在视觉伺服机器人控制中的应用

在基于视觉伺服的机器人控制中,准确跟踪动态目标的关键在于快速准确地完成连续图像中的目标辨识.本文在基于特征差异的彩色目标识别方法的基础上,应用卡尔曼滤波算法对运动目标位置进行预测,在预测的局部范围内搜索匹配目标,将全局搜索改变为局部搜索,大大减少了图像处理数据量,提高了机器人快速跟踪运动目标的实时性.     

Anticipation of moving stimuli by the retina

A flash of light evokes neural activity in the brain with a delay of 30-100 milliseconds, much of which is due to the slow process of visual transduction in photoreceptors. A moving object can cover a considerable distance in this time, and should therefore be seen noticeably behind its actual location. As this conflicts with everyday experience, it has been suggested that the visual cortex uses the delayed visual data from the eye to extrapolate the trajectory of a moving object, so that it is perceived at its actual location. Here we report that such anticipation of moving stimuli begins in the retina. A moving bar elicits a moving wave of spiking activity in the population of retinal ganglion cells. Rather than lagging behind the visual image, the population activity travels near the leading edge of the moving bar. This response is observed over a wide range of speeds and apparently compensates for the visual response latency. We show how this anticipation follows from known mechanisms of retinal processing.     

基于幂次变换和区域收缩算法的运动目标检测与定位

运动目标分割与跟踪是计算机视觉中的重要研究课题,而目标检测与定位是其中的必要步骤,对分割与跟踪效果影响很大.在此提出一种新的运动目标检测与定位方法,该方法在差分二值图像上,通过区域收缩定位到运动像素密度较大的区域,从而实现运动目标定位.在多目标情况下,先通过幂次变换突出不同位置的目标,然后再通过区域收缩实现目标定位,给出目标的特征矩形,便于进一步的跟踪与识别.该方法不需要任何关于目标数的先验知识,对噪声鲁棒性较强.文中给出的实验结果证明该算法的有效性.     

类人足球机器人彩色目标识别与头部视觉跟踪

为了在机器人足球比赛中能够快速准确地跟踪运动目标,提出了一种颜色识别与头部视觉跟踪相结合的方法.利用HSV色彩空间颜色阈值的判别和种子点区域生长的填充算法识别物体,计算物体质心并判断其位置,然后通过舵机控制机器人头部的转动跟踪目标.实验结果表明,该方法具有较高的跟踪速度与准确性.     

基于Kalman滤波器与肤色模型的手势跟踪方法

有效和鲁棒的手势跟踪是动态手势识别的前提,针对手势及其运动的特点,提出了结合Kalman滤波器和肤色模型的手势运动目标跟踪方法.首先通过背景差法和YCb’Cr’空间上的椭圆肤色模型检测出手部运动目标,通过目标区域的空间结构参数来设置Kalman滤波器的各项运动参数,然后计算空间结构特征的跟踪匹配函数对目标预测位置进行修正,获得运动手势目标区域及其运动轨迹.实验结果表明,所提方法能有效地跟踪手势,并能较好地适应手势在运动过程中的手形变化、轨迹转弯等情况,检测准确,鲁棒性高.     

Humans integrate visual and haptic information in a statistically optimal fashion.

When a person looks at an object while exploring it with their hand, vision and touch both provide information for estimating the properties of the object. Vision frequently dominates the integrated visual-haptic percept, for example when judging size, shape or position, but in some circumstances the percept is clearly affected by haptics. Here we propose that a general principle, which minimizes variance in the final estimate, determines the degree to which vision or haptics dominates. This principle is realized by using maximum-likelihood estimation to combine the inputs. To investigate cue combination quantitatively, we first measured the variances associated with visual and haptic estimation of height. We then used these measurements to construct a maximum-likelihood integrator. This model behaved very similarly to humans in a visual-haptic task. Thus, the nervous system seems to combine visual and haptic information in a fashion that is similar to a maximum-likelihood integrator. Visual dominance occurs when the variance associated with visual estimation is lower than that associated with haptic estimation.