首页 | 本学科首页   官方微博 | 高级检索  
相似文献
 共查询到20条相似文献,搜索用时 359 毫秒
1.
视觉是人与外界互动获取信息的主要手段,而双目视差信息是人脑估计外界环境深度结构的重要视觉线索之一.因此,研究人脑处理双目视差的神经机制对了解人类的视觉系统意义重大.功能磁共振成像(f MRI)技术为双目视差研究提供了有效手段.目前在f MRI研究中,虽然已经有很多研究利用f MRI技术深入探究了人脑处理双目视差信息的神经机制,但是利用该技术采集的人脑信号如何分辨包含双目视差信息的立体图像依然有待研究.针对这一问题,设计了一种基于f MRI的实验,该实验选用随机点图生成人造立体视图像作为实验刺激;相较于自然立体图像,该种立体图像可以更加方便地提取出立体图像中包含的图像特征.结合实验特性提出了一种基于lasso回归算法的体素编码模型,该模型利用了视觉感受野的稀疏特性,可以较好地借助立体图像中的二维特征并对fMRI数据进行编码分析和解码分析.其编码分析结果表明利用体素编码模型可以较好预测人脑接收立体图像的脑信号的体素广泛分布在人脑的各个视觉区中,并且大部分体素分布在初级视区V1、V2d和V3d中.解码分析结果表明,初级视觉区V1可以利用立体图像中的二维特征实现立体图像的识别,并且背侧视觉区V3d、V7和h MT+/V5可以与V1协同工作进行立体图像的识别.  相似文献   

2.
针对现有的图像质量评价方法多从特征提取角度考虑图像的失真,忽视初级视皮层是视觉信息处理和认知推理的前提的问题,受人类视觉系统特性的启发,提出一种基于初级视皮层视觉特性的图像质量评价算法;该方法基于对初级视皮层视觉特性的学习,利用初级视皮层中感受野对视觉感知信息的稀疏编码特性,提取模拟初级视皮层感受野特性的基函数,结合独立成分分析和结构相似度算法构建一种初级视觉相似性测度法,并在LIVE图像数据库中进行实验。结果表明,该模型的预测结果与主观质量评价有很好的一致性,并优于已有的结构相似度算法。  相似文献   

3.
通过构建神经元发放模型实现对视觉系统刺激编码方式的解析是计算神经生理学领域的研究热点。在传统线性-非线性-泊松(LNP)模型的基础上,采用稀疏编码模型训练的基函数作为模型的刺激滤波器,进一步利用小世界网络优化神经元集群的连接结构,构建了一种新型初级视皮层(V1区)神经元发放模型,用于预测神经元在特定刺激模式下的响应活动。利用在LE大鼠V1区采集的多通道发放数据拟合模型参数,进一步验证模型的有效性。实验结果表明,与传统未选择基函数作为刺激滤波器以及未经过小世界网络优化的对照模型相比,该模型能更准确地预测大鼠V1区神经元在不同朝向光栅刺激下的响应。该研究表明,经小世界网络优化后,模型中神经元的连接结构具有更强的生物相似性,能更真实地反映初级视觉皮层神经元群的响应机制。  相似文献   

4.
针对图像表示,提出了一种基于改进稀疏编码模型的图像分类算法.首先,提取表示图像视觉局部特征的SIFT (Scale Invariant Feature Transform) 描述子;然后,利用稀疏编码方法生成基于SIFT描述子的视觉词汇库,将SIFT描述子编成稀疏向量;通过有效稀疏向量的区域融合和空间结合而获取整体的稀疏向量并用于图像表示;最后,采用随机森林多分类器对稀疏向量进行训练和测试.结果表明,与现有的算法相比,该算法的性能更佳,可以有效表示图像的特性并提高其分类的准确率.  相似文献   

5.
针对目前零样本图像分类均采用图像底层视觉特征训练属性分类器而导致分类精度较低的问题,提出一种基于稀疏编码空间金字塔模型的零样本学习方法,给出系统结构流程图.首先从原始视觉图像中提取SIFT特征,并进行SIFT特征点提取;然后构建空间金字塔最大池化模型,对已提取的SIFT中间特征进行稀疏编码;最后建立间接属性预测模型.给出基于稀疏编码的空间金字塔最大池化模型的零样本学习算法步骤,完成对目标图像的属性预测,从而达到零样本图像分类的目的.在Shoes数据集与OSR数据集上进行了对比试验.结果表明:试验证实了文中算法的有效性;相对于传统算法,试验耗时减少,图像属性预测精度增加,图像分类识别率提高.  相似文献   

6.
针对在压缩传感中独立使用全局或局部稀疏字典所分别导致的图像细节或整体图像结构信息的丢失,提出了一种联合利用局部和全局稀疏约束来捕捉磁共振图像细节和整体结构信息的磁共振图像重建算法。该算法首先从特定的磁共振图像中训练出稀疏字典,然后利用该字典进行局部稀疏编码。其次,利用预定义的全局字典来加强磁共振图像的全局稀疏性。最后,在局部和全局稀疏的共同约束下,利用非线性共轭梯度算法来对重建模型进行求解。整个重建过程可以重复迭代以逐步改善重建质量。实验结果表明:当下采样因子达到10时,相比于字典学习算法(dictionary learning MRI,DLMRI),提出的算法在重建质量上可以提高1-6dB。  相似文献   

7.
传统的基于稀疏编码的图像分类算法没有考虑不同特征之间的关系。针对这一不足,该文提出了一种新的词典学习算法。该算法考虑特征所处的多个流形空间上的拓扑结构,并显式地对其进行建模,然后将该模型嵌入到稀疏编码算法中构造多流形上的词典优化目标函数。为求解该优化问题,还提出了使用坐标下降的方法,同时给出了收敛性分析。在图像分类3个基准数据集上的实验结果表明,提出的算法分类性能超过了基于传统稀疏编码的算法。这也进一步证明了该算法的有效性。  相似文献   

8.
为了提高重建图像的分辨率,提出一种改进的稀疏表示超分重建算法.在稀疏编码阶段,引入非局部相似正则化以改进稀疏编码目标函数,并通过非局部相似正则化获得图像非局部冗余,以保持图像边缘信息.为了进一步恢复图像的边缘细节信息,提出一种基于改进双边滤波的全局误差补偿模型,以实现重建图像的误差补偿.实验结果表明:与Bicubic,L1SR,SISR,ANR,NE+LS,NE+NNLS,NE+LLE和A+(16 atoms)等算法相比,无论在主观视觉效果,还是在峰值信噪比和结构相似性指标上,所提算法都有显著的提高.  相似文献   

9.
稀疏表示模型是通过将字典中的原子进行组合得到期望的结果.为了解决传统字典学习中所有图像块重建均使用同一个字典,从而忽略了最佳稀疏域的问题,提出来一种基于多字典和稀疏噪声编码的图像超分辨率重建算法.在字典训练时,利用图像的特征将它们合理地划分成若干个簇,每个聚类训练生成子字典对,利用最佳字典对进行重建.在求解稀疏系数阶段,引入稀疏编码噪声去除噪声的影响,利用图像非局部自相似性来获得原始图像稀疏编码系数的良好估计,然后将观测图像的稀疏编码系数集中到这些估计当中.实验表明,与ASDS算法和SSIM算法相比较,该算法有更好的重建结果,获得了更丰富的图像细节和更清晰的边缘.  相似文献   

10.
提出一种基于组稀疏卡尔曼滤波的机动轨迹多步预测方法。首先引入组稀疏编码,通过一次学习建立简单的多步线性回归预测模型,克服了传统方法未能充分利用历史数据而导致预测精度降低的问题;再利用最小角回归算法来计算该模型的稀疏系数,进一步改善模型系数估计的准确性;然后改进了卡尔曼滤波算法,并结合上述组稀疏编码算法,来确保预测结果的精确性;最后通过与传统BP、长短时记忆网络和组稀疏编码方法的仿真比较,验证了所提方法的有效性。  相似文献   

11.
To simulate the brain functions, a quantum associative memory combined with information preprocessing by a sparse coding model is presented. The sparse coding scheme is used to simulate the information transformation from retina up to primary visual cortex (V1) along the visual path and the quantum associative memory is used to simulate the pattern processing functions of the brain such as the pattern storing, forgetting and retrieving. Experimental results show that the model exhibits good associative ability on face recognition. Considering the huge storage capacity, mass parailel-distributed processing ability and oscillatory phenomena of the quantum system, this model might he a biological plausible implementation.  相似文献   

12.
 采用基本ICA模拟视觉感知机制对自然图像分解得到的图像基函数在空间排列上是混乱的,这与视觉生理机制相互矛盾.模拟视皮层感受野间的信息整合机制,建立了新的计算模型.针对基于内容的图像故障区域检测问题,提出了相应的高效率少样本检测算法. 首先,以列车正常和故障图像序列作为训练数据,利用拓扑ICA方法学习图像基函数,由此得到的独立分量系数作为神经元响应,然后模拟同步振荡机制选择响应强烈的神经元,输出其对应的内容,最后通过自动对比实现图像故障区域的快速定位.实验结果表明,与传统方法相比较,引入视觉信息整合机制的新模型及其算法能够提高故障检测率.  相似文献   

13.
Machine vision is an active branch of artificial intelligence. An important problem in this area is the trade-off among efficiency, accuracy and computation complexity. The human visual system can keep watchfulness to the perimeter of a viewing field while at the same time focus on the center of the field for fine information processing. This mechanism of appropriate assignment of computing resources can reduce the demand for huge and complex hardware structure. Therefore, the design of a computer model based on the biological visual mechanism is an effective approach to resolve problems in machine vision. In this paper, a multi-layer neural model is developed based on the features of receptive field of ganglion in retina to simulate multi-scale perceptive fields of ganglion cell. The neural model can maintain alert on the outer area of the image while capturing and processing more important information in the central part. It may provide valuable inspiration for the implementation of real-time processing and avoidance of huge computation in machine vision.  相似文献   

14.
S Shipp  S Zeki 《Nature》1985,315(6017):322-325
V5 and V4 are areas of macaque monkey prestriate visual cortex that are specialized for involvement in different aspects of visual perception, namely motion for V5 (refs 1-4) and colour vision, with other possible functions, for V4 (refs 2, 5-9). Thus, it is unlikely that they should be fed the same information for further processing, yet both receive a strong input from patches of the upper layers of V2 (refs 10, 11), the area immediately adjoining the primary visual cortex, V1. V2, however, seems to comprise functionally distinct subregions, which can be revealed by staining the tissue for the mitochondrial enzyme cytochrome oxidase. Here we report that V4 and V5 are connected with separate cytochrome oxidase-defined subregions of V2, suggesting that cortical pathways dealing with motion and colour perception are segregated in their passage through V2, and reinforcing evidence for functional specialization in the visual cortex.  相似文献   

15.
针对传统算法在抗光照变化影响、大位移光流和异质点滤除等方面的不足,从人类视觉认知机理出发,提出了一种基于机器学习和生物模型的运动自适应V1-MT(motion-adaptive V1-MT,MAV1MT)序列图像光流估计算法.首先,引入基于ROF模型的结构纹理分解(structure-texture decomposition,STD)技术,有效解决了光照和色彩变化的影响.其次,利用多V1细胞加权组合及非线性正则化模拟MT细胞模型,并结合岭回归训练学习得到运动自适应的权重,解决对目标的运动速度感知问题.最后,引入由粗到精的增强方法和图像金字塔局部运动估计采样,将V1-MT运动估计模型应用于实际大位移视频序列.理论分析和实验结果表明,新方法能更加拟合人眼视觉信息处理特性,对视频序列具有普适、有效、鲁棒的运动感知性能.  相似文献   

16.
云阴影的存在严重影响了高分辨率航空遥感影像的视觉效果和影像质量.为去除云阴影,提出了一种无需人工参与的自动处理方法.首先采用光谱特征阈值法初步检测云阴影,再结合形态学方法进行后处理,提取完整的云阴影区域;阴影去除时,设计了一种利用非阴影区域信息、阴影整体区域信息、阴影像素局部窗口及像素本身信息的综合补偿模型,所需参数均可自动获取,进而实现阴影的自动去除.实验表明,区域级、窗口级和像素级多层次信息的引入,使模型的细节敏感度增强,提升了对比度,能灵活根据复杂的云阴影遮挡情况对阴影进行合理提升,有效解决了部分补偿方法存在的补偿效果不均匀的问题,阴影中地物的色彩与纹理等信息均得到更真实的再现.  相似文献   

17.
Stevens CF 《Nature》2001,411(6834):193-195
A hallmark of mammalian brain evolution is the disproportionate increase in neocortical size as compared with subcortical structures. Because primary visual cortex (V1) is the most thoroughly understood cortical region, the visual system provides an excellent model in which to investigate the evolutionary expansion of neocortex. I have compared the numbers of neurons in the visual thalamus (lateral geniculate nucleus; LGN) and area V1 across primate species. Here I find that the number of V1 neurons increases as the 3/2 power of the number of LGN neurons. As a consequence of this scaling law, the human, for example, uses four times as many V1 neurons per LGN neuron (356) to process visual information as does a tarsier (87). I argue that the 3/2 power relationship is a natural consequence of the organization of V1, together with the requirement that spatial resolution in V1 should parallel the maximum resolution provided by the LGN. The additional observation that thalamus/neocortex follows the same evolutionary scaling law as LGN/V1 may suggest that neocortex generally conforms to the same organizational principle as V1.  相似文献   

18.
Gaze direction controls response gain in primary visual-cortex neurons   总被引:11,自引:0,他引:11  
Trotter Y  Celebrini S 《Nature》1999,398(6724):239-242
To localize objects in space, the brain needs to combine information about the position of the stimulus on the retinae with information about the location of the eyes in their orbits. Interaction between these two types of information occurs in several cortical areas, but the role of the primary visual cortex (area V1) in this process has remained unclear. Here we show that, for half the cells recorded in area V1 of behaving monkeys, the classically described visual responses are strongly modulated by gaze direction. Specifically, we find that selectivity for horizontal retinal disparity-the difference in the position of a stimulus on each retina which relates to relative object distance-and for stimulus orientation may be present at a given gaze direction, but be absent or poorly expressed at another direction. Shifts in preferred disparity also occurred in several neurons. These neural changes were most often present at the beginning of the visual response, suggesting a feedforward gain control by eye position signals. Cortical neural processes for encoding information about the three-dimensional position of a stimulus in space therefore start as early as area V1.  相似文献   

19.
CFD数值方法是目前海上浮式设施爆燃冲击研究中得到广泛运用的方法.文中以某FLNG工艺区发生爆燃事件为研究背景,基于通用CFD软件CFX对蒸气云爆炸进行数值分析,提出了基于能量释放方程模拟分析的冲击波安全评价方法,并对能量和能量释放时间两个变量对爆燃冲击强度的敏感性进行研究.该冲击波后果评价方法通过在爆心位置设立以能量大小和能量释放时间为变量的step函数,选定k-ε模型,模拟爆炸冲击获得超压数据,并结合超压伤害准则和TNO爆炸源能量和气云体积的转换关系对冲击波伤害后果进行评价.研究表明:基于CFD的能量释放方程能够很好的模拟爆炸冲击波并获得表征冲击强度的超压数据.方程中变量因素对冲击强度影响显著,特别是爆炸近场区域.提出的安全分析方法可以有效地对爆炸冲击后果进行评价.  相似文献   

20.
The cognitive model ABGP is a special model for agents , which consists of awareness , beliefs, goals and plans .The ABGP agents obtain the knowledge directly from the natural scenes only through some single preestablished rules like most agent architectures .Inspired by the biological vis-ual cortex ( V1 ) and the higher brain areas perceiving the visual feature , deep convolution neural networks ( CNN) are introduced as a visual pathway into ABGP to build a novel visual awareness module .Then a rat-robot maze search simulation platform is constructed to validate that CNN can be used for the awareness module of ABGP .According to the simulation results , the rat-robot imple-mented by the ABGP with the CNN awareness module reaches the excellent performance of recogniz-ing guideposts , which directly enhances the capability of the communication between the agent and the natural scenes and improves the ability to recognize the real world , which successfully demon-strates that an agent can independently plan its path in terms of the natural scenes .  相似文献   

设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号