立体图对匹配基元的心理物理学实验的探讨

Marr和Poggio等人从视觉信息加工理论中提出零交叉(Zero-Crossing)是视觉早期加工过程中首先得到的最基本的信息。这在生理上有二个高斯分布叠加的感受野以及侧抑现象等可以作为依据,但在心理物理上仍缺乏佐证。我们利用不同密度的随机点,加上了不同百分比的反差噪声的立体图对,观察了各种条件下体视对噪声的耐受特性,得出了有利于说明黑白相间的边(零交叉)为匹配基元的结果。本文报导了这些实验结果并进行了讨论。     

Image Registration Algorithm Using Mexican Hat Function-Based Operator and Grouped Feature Matching Strategy

Feature detection and matching are crucial for robust and reliable image registration. Although many methods have been developed, they commonly focus on only one class of image features. The methods that combine two or more classes of features are still novel and significant. In this work, methods for feature detection and matching are proposed. A Mexican hat function-based operator is used for image feature detection, including the local area detection and the feature point detection. For the local area detection, we use the Mexican hat operator for image filtering, and then the zero-crossing points are extracted and merged into the area borders. For the feature point detection, the Mexican hat operator is performed in scale space to get the key points. After the feature detection, an image registration is achieved by using the two classes of image features. The feature points are grouped according to a standardized region that contains correspondence to the local area, precise registration is achieved eventually by the grouped points. An image transformation matrix is estimated by the feature points in a region and then the best one is chosen through competition of a set of the transformation matrices. This strategy has been named the Grouped Sample Consensus (GCS). The GCS has also ability for removing the outliers effectively. The experimental results show that the proposed algorithm has high registration accuracy and small computational volume.     

Fuzzy Logic Based Edge Detection in Smooth and Noisy Clinical Images

Edge detection has beneficial applications in the fields such as machine vision, pattern recognition and biomedical imaging etc. Edge detection highlights high frequency components in the image. Edge detection is a challenging task. It becomes more arduous when it comes to noisy images. This study focuses on fuzzy logic based edge detection in smooth and noisy clinical images. The proposed method (in noisy images) employs a 3×3 mask guided by fuzzy rule set. Moreover, in case of smooth clinical images, an extra mask of contrast adjustment is integrated with edge detection mask to intensify the smooth images. The developed method was tested on noise-free, smooth and noisy images. The results were compared with other established edge detection techniques like Sobel, Prewitt, Laplacian of Gaussian (LOG), Roberts and Canny. When the developed edge detection technique was applied to a smooth clinical image of size 270×290 pixels having 24 dB ‘salt and pepper’ noise, it detected very few (22) false edge pixels, compared to Sobel (1931), Prewitt (2741), LOG (3102), Roberts (1451) and Canny (1045) false edge pixels. Therefore it is evident that the developed method offers improved solution to the edge detection problem in smooth and noisy clinical images.     

Transient information flow in a network of excitatory and inhibitory model neurons: Role of noise and signal autocorrelation

We investigate the performance of sparsely-connected networks of integrate-and-fire neurons for ultra-short term information processing. We exploit the fact that the population activity of networks with balanced excitation and inhibition can switch from an oscillatory firing regime to a state of asynchronous irregular firing or quiescence depending on the rate of external background spikes. We find that in terms of information buffering the network performs best for a moderate, non-zero, amount of noise. Analogous to the phenomenon of stochastic resonance the performance decreases for higher and lower noise levels. The optimal amount of noise corresponds to the transition zone between a quiescent state and a regime of stochastic dynamics. This provides a potential explanation of the role of non-oscillatory population activity in a simplified model of cortical micro-circuits.     

Edge extraction by active defocusing.

A novel edge extraction method that employs an active defocusing technique is presented. The method is based on the principle that a Laplacian-of-Gaussian (LOG) operation can be approximated by a Difference-of-Gaussian (DOG) operation. While such an operation is usually done in digital image processing, it can also be very effective conducted in a combination of optical techniques and digital processing. In this edge extraction method, a focused image of an object in a scene is first acquired. The image of the scene is then slightly defocused by changing the focal length of the camera. A real time subtraction operation is applied to subtract the defocused image from the previously acquired image. It produces a residual image that emphasizes abrupt intensity variations. An objective evaluation, called an edge index, is performed on the resulting image. The amount of defocusing is carefully adjusted according to this measurement so that a desired edge image is generated. Boundaries of objects can then be obtained by further enhancement of the edge image. Since this edge detection method is an optical-based process aided by digital processing, it is fast and relatively inexpensive.     

Zero-crossing detectors in primary visual cortex?

David Marr and others have hypothesized that the visual system processes complex scene information in stages, the first of which involves the detection of light intensity edges or zero-crossings (Marr, 1982). Ideal zero-crossing detector mechanisms have been described and modeled in terms of their possible physiological implementation (Marr and Hildreth, 1980; Poggio, 1983). We now present evidence of visual cortical receptive fields which resemble in spatial organizational terms the requirements of zero-crossing analysis. The linear and nonlinear summation within and between the receptive field subunits are described and compared with predicted processes. The relative subunit sizes and separations are analyzed in these terms. Our findings support the notion that receptive fields may correspond with zero-crossing filters rather than zero-crossing detector gates.     

The spatial metric of brain underlying the temporal metric of EEG and thought

Multiplying memory span by mental speed, we obtain the information entropy of short-term memory capacity, which is rate-limiting for cognitive functions and corresponds with EEG power spectral density. From psychometric and EEG data follows a fundamental of about 3.14 Hz. The number of harmonics (n = 1, 2,...,9) is identical with memory span, and the eigenvalues of the EEG impulse response are represented by the zero-crossings up to the convolved fundamental, the P300. The difference T of 4.42 ms between the 8th and the 9th harmonic is the smallest time window of conscious information processing. Shannon's sampling theorem allows the replacement of any band-limited signal by m discrete sequences of T without loss of any information. Brain architecture can be understood in terms of sequences of delaying chains. Acting as a wavefront tracking array, scaled in relation of mT and in such a way also expressing the metric of eigenvalues, widths of orientation columns match with phase reversal after a zero-crossing and lengths of dendritic trees with run-length of travelling harmonics.     

What Is the Primary Cause of Individual Differences in Contrast Sensitivity?

One of the primary objectives of early visual processing is the detection of luminance variations, often termed image contrast. Normal observers can differ in this ability by at least a factor of 4, yet this variation is typically overlooked, and has never been convincingly explained. This study uses two techniques to investigate the main source of individual variations in contrast sensitivity. First, a noise masking experiment assessed whether differences were due to the observer’s internal noise, or the efficiency with which they extracted information from the stimulus. Second, contrast discrimination functions from 18 previous studies were compared (pairwise, within studies) using a computational model to determine whether differences were due to internal noise or the low level gain properties of contrast transduction. Taken together, the evidence points to differences in contrast gain as being responsible for the majority of individual variation across the normal population. This result is compared with related findings in attention and amblyopia.     

Dependence of functional organization of receptive fields of the cat lateral geniculate body on visual stimulus contrast

The spatial organization of receptive fields of the lateral geniculate body in response to visual stimuli with different degrees of contrast was studied in cats. During variation of contrast changes in organization of the central zone were found to take place in some receptive fields. Inside the central zone of the receptive field as revealed by the use of low stimulus contrasts, an additional inhibitory ring appears in response to a stimulus of high contrast. The weighting function of the central zone of the receptive field becomes variable in sign. The role of this phenomenon in transmission of information on high spatial frequencies (increase in visus) at high contrasts is discussed.I. P. Pavlov Institute of Physiology, Academy of Sciences of the USSR, Leningrad. Translated from Neirofiziologiya, Vol. 16, No. 6, pp. 789–796, November–December, 1984.     

The Gaussian derivative model for spatial vision: I. Retinal mechanisms

Physiological evidence is presented that visual receptive fields in the primate eye are shaped like the sum of a Gaussian function and its Laplacian. A new 'difference-of-offset-Gaussians' or DOOG neural mechanism was identified, which provided a plausible neural mechanism for generating such Gaussian derivative-like fields. The DOOG mechanism and the associated Gaussian derivative model provided a better approximation to the data than did the Gabor or other competing models. A model-free Wiener filter analysis provided independent confirmation of these results. A machine vision system was constructed to simulate human foveal retinal vision, based on Gaussian derivative filters. It provided edge and line enhancement (deblurring) and noise suppression, while retaining all the information in the original image.     

Color vision and image intensities: When are changes material?

The difficulty in understanding a biological system or its components without some idea of its goals has been emphasized by Marr. In this paper, a preliminary goal for color vision is proposed and analyzed. That goal is to determine where changes of material occur in a scene (using only spectral information). The goal is challenging because the effects of many processes (shadowing, shading from surface orientation changes, highlights, variations in pigment density) are confounded with the effects of material changes in the available image intensities. We show there is a minimal and unique condition, thespectral crosspoint, that rejects instances of these confounding processes. (If plots are made of image intensity versus wavelength from two image regions, and the plots intersect, we say that there is a spectral crosspoint.) An operator is designed to detect crosspoints; it turns out to resemble double-opponent cells described in primate visual cortex.     

Neural network model of visual cortex for determining surface curvature from images of shaded surfaces

The visual system can extract information about shape from the pattern of light and dark surface shading on an object. Very little is known about how this is accomplished. We have used a learning algorithm to construct a neural network model that computes the principal curvatures and orientation of elliptic paraboloids independently of the illumination direction. Our chief finding is that receptive fields developed by units of such model network are surprisingly similar to some found in the visual cortex. It appears that neurons that can make use of the continuous gradations of shading have receptive fields similar to those previously interpreted as dealing with contours (i.e. 'bar' detectors or 'edge' detectors). This study illustrates the difficulty of deducing neuronal function within a network solely from receptive fields. It is also important to consider the pattern of connections a neuron makes with subsequent stages, which we call the 'projective field'.     

Characteristics of dynamic reorganization of receptive fields of cortical and geniculate visual neurons of cat in response to changes in photic stimulation parameters

The effect of parameters of local photic stimulation of different points of the receptive field on the characteristics of dynamic reorganization of receptive fields of cortical and geniculate visual neurons within microintervals of time observed previously was studied in computer-controlled experiments on unanesthetized, curarized cats. Dependence on the degree of widening of the receptive field and the temporal characteristics of this process on the background illumination level, intensity, contrast, area, duration, energy, and orientation of a local rectangular or circular photic stimulus flashing in random order at 100 points of the tested part of the visual field was studied. It was concluded that the background illumination level and the intensity, size, duration, and orientation of the stimulus have a specific effect on dynamic reorganization of the receptive field. The effects of all the parameters studied on the dynamics of the receptive field were shown to be nonlinear functions with optimal values that differed for different cells. The possible functional role of this effect and also the probability that it may participate in information coding in the visual system are discussed.Institute of Higher Nervous Activity and Neurophysiology, Academy of Sciences of the USSR, Moscow. Translated from Neirofiziologiya, Vol. 15, No. 4, pp. 339–346, July–August, 1983.     

Initial processing of visual information within the retina and the LGN

The initial stage of information processing by the visual system reduces the information contained in the continuous image on the retina into a discrete set of responses which are carried from the lateral geniculate nucleus (LGN) to the visual cortex.-1. The optimal sampling of the light intensity distribution in the visual environment is achieved only if each channel in the visual pathways carries undistorted information corresponding to an image element. The visual system approaches as closely as possible the scheme of optimal spatial sampling, retaining the full information on the low spatial frequency content of the object light intensity. The ideal receptive field of a sustained LGN cell is then of the form J ₁ (Kr)/Kr.-2. The experimentally determined receptive fields of sustained LGN cells (and to some extent retinal ganglion cells as well) in cat closely resemble the functional form J ₁ (Kr)/Kr. The centre-surround organization of the receptive fields is therefore understood as a scheme which leads to a maximal information flow through the visual pathways.-3. The optimal sampling scheme cannot be realized by the retina alone, because of restrictions on the size of neural networks. It is therefore constructed in two stages, ending at the LGN level. A recombination of ganglion cell signals into optimal receptive fields is a major role of the LGN.     

A Simple Model of Optimal Population Coding for Sensory Systems

A fundamental task of a sensory system is to infer information about the environment. It has long been suggested that an important goal of the first stage of this process is to encode the raw sensory signal efficiently by reducing its redundancy in the neural representation. Some redundancy, however, would be expected because it can provide robustness to noise inherent in the system. Encoding the raw sensory signal itself is also problematic, because it contains distortion and noise. The optimal solution would be constrained further by limited biological resources. Here, we analyze a simple theoretical model that incorporates these key aspects of sensory coding, and apply it to conditions in the retina. The model specifies the optimal way to incorporate redundancy in a population of noisy neurons, while also optimally compensating for sensory distortion and noise. Importantly, it allows an arbitrary input-to-output cell ratio between sensory units (photoreceptors) and encoding units (retinal ganglion cells), providing predictions of retinal codes at different eccentricities. Compared to earlier models based on redundancy reduction, the proposed model conveys more information about the original signal. Interestingly, redundancy reduction can be near-optimal when the number of encoding units is limited, such as in the peripheral retina. We show that there exist multiple, equally-optimal solutions whose receptive field structure and organization vary significantly. Among these, the one which maximizes the spatial locality of the computation, but not the sparsity of either synaptic weights or neural responses, is consistent with known basic properties of retinal receptive fields. The model further predicts that receptive field structure changes less with light adaptation at higher input-to-output cell ratios, such as in the periphery.     

The computational measurement of apparent motion: A recurrent pattern recognition strategy as an approach to solve the correspondence problem

In short, the model consists of a two-dimensional set of edge detecting units, modelled according to the zero-crossing detectors introduced first by Marr and Ullman (1981). These detectors are located peripherally in our synthetic vision system and are the input elements for an intelligent recurrent network. The purpose of that network is to recognize and categorize the previously detected contrast changes in a multi-resolution representation of the original image in such a manner that the original information will be decomposed into a relatively small numberN of well-defined edge primitives. The advantage of such a construction is that time-consuming pattern recognition has no longer to be done on the originally complex motion-blurred images of moving objects, but on a limited number of categorized forms. Based on a numberM of elementary feature attributes for each individual edge primitive, the model is then able to decompose each edge pattern into certain features. In this way anM-dimensional vector can be constructed for each edge. For each sequence of two successive frames a tensor can be calculated containing the distances (measured inM-dimensional feature space) between all features in both images. This procedure yields a set ofK—1 tensors for a sequence ofK images. After cross-correlation of allN ×M feature attributes from image (i) with those from image (i+1), wherei = 1, ...,K - 1, probability distributions can be computed. The final step is to search for maxima in these probability functions and then to construct from these extremes an optimal motion field. A number of simulation examples will be presented.