有指向性的视觉注意计算机模型

注意把有限的处理资源优先分配给那些需要精细加工的信息,能提高视觉信息加工中的检测能力和响应速度.基于生物视觉系统的生理结构特点,建立了模拟生物视觉注意系统的有指向性的视觉注意计算机模型.模型首先模拟生物视网膜的成像机制,将视场图像转化为视网膜图像;然后将最大梯度边缘检测和c-均值聚类等方法相结合,对视网膜图像中的目标进行编码,分别提取每个目标的颜色、中心以及边缘点集合等基本信息;最后用知识库中指向性目标的特征来指导注意焦点的转移.实验结果表明,利用此模型能较好地实现注意焦点的转移.     

基于视觉信息的PCNN参数自适应设定及模型改进

脉冲耦合神经网络(PCNN)参数决定该模型在数字图像处理领域的应用.现阶段网络参数自适应设定是依据图像统计信息或网络自身结构.基于此,提出基于生物视觉信息的PCNN参数自适应设置方法及模型改进.该方法通过对生物视觉感知理论与PCNN网络性质的分析,揭示了视觉感知理论与PCNN网络参数M、W和β的同源性,给出依据视觉感知模型自适应设定PCNN网络参数W、M和β的方法,并设计出具有生物视觉特征的PCNN改进模型.实验验证了该模型的几何不变性,在基于内容的图像检索领域取得了良好效果.     

视网膜假体视觉信息处理模型

视觉假体方案是神经工程领域用于部分或完全恢复视觉的主要手段,建立生物视网膜模型,并仿真其视觉信息处理功能,是视网膜假体研究中的一个重要组成部分.从已知的生理机制出发,提出两层结构的视网膜信息处理模型:外网状层(Out Plexiform Layer,OPL)信息提取和内网状层(Inner Plexiform Layer,IPL)信息编码,通过时空滤波、静态非线性调整和泊松峰电位产生,建立输入刺激图像和输出峰电位序列之间的直接关系,并在 MATLAB 平台上结合其图形模块进行仿真研究,得到了与内、外网状层处理结果对应的边缘轮廓图像和携带视觉信息的峰电位序列,为视网膜假体研究提供了一种理论上的可行性模型.     

基于纹理图像分析的生物视觉模型不变性评价

基于纹理图像,从计算机视觉角度对生物视觉模型——视皮层目标识别的标准模型进行定量分析与评价。对原始图像分别进行尺度、旋转及仿射等变化,利用标准模型提取变化后图像的生物视觉特征,再根据提取的生物视觉特征对纹理图像进行分类,采用图像分类结果的曲线下面积来定量分析和评价生物视觉模型是否具有不变性。大量与局部二元模式特征的对比实验表明,该模型提取的生物视觉特征对于纹理图像具备优良的尺度、旋转与仿射不变性。     

结合视觉感知与LBP傅里叶直方图的图像检索

为了有效地提取图像特征以提高图像检索性能,借鉴生物视觉信息处理过程中的提取图像特征,提出一种结合视觉感知与局部二值模式(LBP)傅里叶直方图的图像检索算法.首先根据视觉感知特点,用主分量图作为亮度初级视觉特征,将形状边缘信息融入视觉注意模型,获得改进的Itti视觉注意模型,并基于该改进视觉注意模型得到50个视觉特征图;...     

基于视觉显著计算的图像语义检索方法

网络标签已经开始广泛地用于图像内容的标注和分享,由于图像本身的差异和人们对图像的不同理解,对图像语义检索提出了新的挑战。该文首先引入视觉显著模型,突出图像的显著信息;然后提取视觉显著特征,建立图像内容的相似关系;最后基于随机漫步模型平衡图像内容及网络标签间的关系。实验表明该文提出的方法能够有效地实现图像的语义理解并用于图像检索。     

基于生成感知差异的无参考图像质量评价模型

人眼视觉系统中的视觉感知差异是图像质量评价过程中的重要组成部分,通过感知失真图像与原始图像之间的视觉差异,可对图像的失真程度进行判断,然而在无参考图像质量评价中无法获取原始未失真的图像,且缺乏对失真图像的视觉感知差异。通过对深度学习中的生成对抗网络进行分析,提出一种基于生成视觉感知差异的无参考图像质量评价模型。利用生成对抗网络产生与失真图像相对应的视觉感知差异图像,并将其与失真图像输入质量评价网络以进一步学习图像的失真信息,从而达到评估图像质量的目的。在TID2008和TID2013数据库上的实验结果表明,与CNN、SOM、CORNIA等模型相比,该模型能够使失真图像质量预测准确度提升1个百分点以上,且对不同种类失真也表现出良好的预测性能。     

基于生物视觉的目标识别与跟踪

在对视觉系统中各层的生物原型进行分析抽象和简化的基础上,提出了基于生物视觉的目标识别与跟踪模型.该模型采用CNN来模拟内视网膜激发模块,用在高分辨率下具有旋转、缩放、平移(RST)不变性的HU矩实现测量空间到特征空间的转化,最后采用BP神经网络来实现目标分类、质心(COG)算法实现目标跟踪.经过大量真实图像的试验,证明了图像分辨率达到1024* 024时,目标识别率达到95％以上,同时具有良好的跟踪效果,该模型中的算法具有良好的鲁棒性.     

结合引导解码和视觉注意力的图像语义描述模型

针对输入的图像视觉信息不能在每一步解码过程中动态调整,同时为了提高图像语义描述模型的精度和泛化能力,提出了一种结合引导解码和视觉注意力机制的双层长短时记忆（long short term memory,LSTM）网络的图像语义描述模型。将提取到的图像的视觉和目标特征通过一个引导网络建模后送入LSTM网络的每一时刻,实现端到端的训练过程;同时设计了基于图像通道特征的视觉注意力机制,提高了模型对图像细节部分的描述。利用MSCOCO和Flickr30k数据集对模型进行了训练和测试,结果显示模型性能在不同的评价指标上都得到了提升。     

一种基于视觉特性的仿生图像增强算法

常见的基于人类视觉特性的图像增强算法由于是同时完成动态范围压缩和对比度增强,导致增强图像的整体对比度不高、边缘部分效果不佳.通过分析人类视觉系统的全局和局部自适应调节原理及人眼视网膜神经节细胞感受野的传输特性,提出一种仿生图像增强算法.为适应人类视觉系统对光强的主观感觉特性,对图像作全局亮度对数变换;并利用人眼的主观亮度感觉与实际光强的对数呈局部线性关系的特性,采用视网膜神经元感受野三高斯模型来调整亮度图像的局部对比度;最后利用线性变换恢复图像的彩色信息.实验结果表明,该算法的增强效果良好,特别是对于图像边界处,既能很好地增强边缘对比,又可有效地提升区域亮度对比和亮度梯度信息.     

Shape identification using a charge simulation retina model

Investigations were conducted to explore the feasibility of a prototype charge simulation retina machine vision system to identify shape and size, when different three-dimensional objects were arbitrarily located in the vision field of the retina. The system consisted of a light source, light beam conditioner, artificial retina installed with photo sensors, data transfer unit, and a computer installed with analogue to digital converter peripherals. The retina was used to acquire image features for regular prisms. The features were transferred to a charge simulation retina model that was identical to the prototype retina and were compressed using a charge simulation method (CSM) algorithm by computing output signals at work cells located in the retina model. With these signals, neural networks were trained to classify each image sample, to identify shape and size. The results showed that object displacement, especially for locations beyond a circle with a radius one-tenth that of the retina and measured from the centre of the base of the retina, significantly affected shape and size classification performance. Despite this, overall shape and size classification rates of 75% and above were obtained when the retina discriminated between different prisms. The results indicate that it is feasible for the charge simulation retina based on the CSM algorithm to identify three-dimensional shapes.     

基于视中枢神经机制的层次网络计算模型

在视皮层区中,有许多非常规整的柱形功能结构,它们形成的局域网络具有抽取视图像中最基本的特征的计算能力,相邻视神细胞则抑制机制和神经细胞的感受野为实现这样的并行计算能力提供了保证,并且以层的这种等级组构为许多心理现像提供了生理解释,这不仅对模式识别、计算机视觉有重要的价值,而且对人工智能系统的知识获取和知识表示都具有非常重要的意义,文中通过构造一个金字塔状的神经网络层次模型,来对模拟视网膜的输入点阵     

An image representation of infrastructure based on non-classical receptive field

Biological vision systems have become highly optimized over millions of years of evolution, developing complex neural structures to represent and process stimuli. Moreover, biological systems of vision are typically far more efficient than current human-made machine vision systems. The present report describes a non-task-dependent image representation schema that simulates the early phase of a biological neural vision mechanism. We designed a neural model involving multiple types of computational units to simulate ganglion cells and their non-classical receptive fields, local feedback control circuits and receptive field dynamic self-adjustment mechanisms in the retina. We found that, beyond the pixel level, our model was able to represent images self-adaptively and rapidly. A series of statistical analyses revealed that this model not only produces compact and abstract approximations of images, but also retains their primary visual features. In addition, the improved representation was found to substantially facilitate contour detection and image segmentation. We propose that this improvement arose because ganglion cells can resize their receptive fields, enabling multi-scale analysis functionality, a neighborhood referring function and a localized synthesis function. The ganglion cell layer is the starting point of subsequent diverse visual processing. The universality of this cell type and its functional mechanisms suggests that it will be useful for designing image processing algorithms in future.     

基于Retinex和视觉适应性的图像增强

根据人眼视网膜上的锥细胞和柱细胞的视觉特性,提出了用于彩色图像增强的视觉适应性模型。基于Retinex和视觉适应性模型提出了一种新的图像增强算法,先将图像进行简单去光照分量处理,得到反射分量的近似解,再根据视觉适应性模型对反射图像的近似解进行全局对比度和亮度的调整,使之适应于人的视觉。实验中使用的算法和经典Retinex算法处理相同的RGB退化图像,对处理结果进行了定性和定量比较,结果表明提出的算法在增强图像细节,提高全局对比度方面优于已有的Retinex算法。     

Visual information processing in primate cone pathways. I. A model

At the retinal level, the strategies utilized by biological visual systems allow them to outperform machine vision systems, serving to motivate the design of electronic or "smart" sensors based on similar principles. Design of such sensors in silicon first requires a model of retinal information processing which captures the essential features exhibited by biological retinas. In this paper, a simple retinal model is presented, which qualitatively accounts for the achromatic information processing in the primate cone system. The computer retina model exhibits many of the properties found in biological retinas such as data reduction through nonuniform sampling, adaptation to a large dynamic range of illumination levels, variation of visual acuity with illumination level, and enhancement of spatiotemporal contrast information. The main emphasis of the model presented here is to demonstrate how different adaptation mechanisms play a role in extending the operating range of the primate retina.     

Real-time tracking of multiple objects in space-variant vision based on magnocellular visual pathway

In this paper, we propose a space-variant image representation model based on properties of magnocellular visual pathway, which perform motion analysis, in human retina. Then, we present an algorithm for the tracking of multiple objects in the proposed space-variant model. The proposed space-variant model has two effective image representations for object recognition and motion analysis, respectively. Each image representation is based on properties of two types of ganglion cell, which are the beginning of two basic visual pathways; one is parvocellular and the other is magnocellular. Through this model, we can get the efficient data reduction capability with no great loss of important information. And, the proposed multiple objects tracking method is restricted in space-variant image. Typically, an object-tracking algorithm consists of several processes such as detection, prediction, matching, and updating. In particular, the matching process plays an important role in multiple objects tracking. In traditional vision, the matching process is simple when the target objects are rigid. In space-variant vision, however, it is very complicated although the target is rigid, because there may be deformation of an object region in the space-variant coordinate system when the target moves to another position. Therefore, we propose a deformation formula in order to solve the matching problem in space-variant vision. By solving this problem, we can efficiently implement multiple objects tracking in space-variant vision.     

Optical flow in log-mapped image plane - a new approach

Foveating vision sensors are important in both machine and biological vision. The term space-variant or foveating vision refers to sensor architectures based on smooth variation of resolution across the visual field, like that of the human visual system. Traditional image processing techniques do not hold when applied directly to such an image representation since the translation symmetry and the neighborhood structure in the spatial domain is broken by the space-variant properties of the sensor. Unfortunately, there has been little systematic development of image processing tools that are explicitly designed for foveated vision. The author proposes a novel approach to compute the optical flow directly on log-mapped images. We propose the use of a generalized dynamic image model (GDIM) based method for computing the optical flow as opposed to the brightness constancy model (BCM) based method. We introduce a new notion of "variable window" and use the space-variant form of gradient operator while computing the spatio-temporal gradient in log-mapped images for a better accuracy and to ensure that the local neighborhood is preserved. We emphasize that the proposed method must be numerically accurate, provide a consistent interpretation, and be capable of computing the peripheral motion. Experimental results on both the synthetic and real images have been presented to show the efficacy of the proposed method     

基于V1细胞特性的边缘检测

针对均衡边缘检测精度和抗噪性能难度大的问题,借鉴初级视皮层（V1）细胞的动静态感知特性,建立具有方位选择性的V1细胞模型应用于图像边缘检测。采用时空滤波器来模拟简单细胞的感受野,通过使用能量模型和归一化来整合简单细胞的响应得到V1细胞模型,从而利用V1细胞静态感知特性来检测自然图像边缘。仿真结果表明,所提V1细胞模型能够基本拟合生物数据,具有生物上的普适性;与传统的边缘检测算子相比,该模型的性能更优,鲁棒性更强。依据生物实验结论来构建生物视觉模型并用于图像处理,对生物视觉和计算机视觉的融合进行了有益的探索。     

The resonant retina: exploiting vibration noise to optimally detect edges in an image

We show that, far from being a drawback, the ubiquitous presence of random vibrations in vision systems operating from mobile devices can advantageously be used as a fundamental tool for edge detection. Directly inspired by biology, the concept of dynamic retina uses the random spatiotemporal path, traced by a moving receptor that samples the image over time, as the basis for the edge detection operation. We propose a simple mathematical formalization of the dynamic retina concept that shows that the relevant information needed for edge detection is contained in the modulation of the variance of the output signal delivered by the retina. Based on a sequence of observations, we then use a variance estimator to determine the presence of the image edges. Following again a biological inspiration, more specifically focusing on neuron dynamics, we introduce a threshold type estimator and use its local asymptotic normality to optimize, via the Cramer-Rao relation, the value of the threshold. The optimal threshold value coincides with a maximum of the associated Fisher information and the overall process can therefore be directly interpreted as a stochastic resonance. We end our contribution by reporting some simple experimental illustrations.     

Ganglion-Based Balance Design of Multi-Layer Model and Its Watchfulness-Keeping

An important problem of machine vision is the balance among the efficiency, accuracy and computation cost. The visual system of man can keep watchfulness to the perimeter of a visual field and subtly process information emerging in the center of the visual field at the same time. This kind of requirement assignment of computation can virtually ease the demand of hardware both in quantity and complexity. Therefore designing an artificial model based on biological mechanism is an effective approach. In this paper a multi-layer neural model is designed based on the multi-scale receptive fields of ganglions in retina. The model can keep watch on the periphery part of a scene while processing the center information of the scene. And why it can balance the hardware complexity, processing precision and computational intensity is analyzed. An experiment is done to test the model's sensitivity in watchfulness keeping and its efficiency and veracity in environment sampling. This model may provide valuable inspiration in the implementation of real-time processing and the avoidance of expensive computation cost in machine vision.