How to “hear” visual disparities: real-time stereoscopic spatial depth analysis using temporal resonance

In a stereoscopic system, both eyes or cameras have a slightly different view. As a consequence, small variations between the projected images exist (`disparities') which are spatially evaluated in order to retrieve depth information (Sanger 1988; Fleet et al. 1991). A strong similarity exists between the analysis of visual disparities and the determination of the azimuth of a sound source (Wagner and Frost 1993). The direction of the sound is thereby determined from the temporal delay between the left and right ear signals (Konishi and Sullivan 1986). Similarly, here we transpose the spatially defined problem of disparity analysis into the temporal domain and utilize two resonators implemented in the form of causal (electronic) filters to determine the disparity as local temporal phase differences between the left and right filter responses. This approach permits real-time analysis and can be solved analytically for a step function contrast change, which is an important case in all real-world applications. The proposed theoretical framework for spatial depth retrieval directly utilizes a temporal algorithm borrowed from auditory signal analysis. Thus, the suggested similarity between the visual and the auditory system in the brain (Wagner and Frost 1993) finds its analogy here at the algorithmical level. We will compare the results from the temporal resonance algorithm with those obtained from several other techniques like cross-correlation or spatial phase-based disparity estimation showing that the novel algorithm achieves performances similar to the `classical' approaches using much lower computational resources. Received: 9 April 1997 / Accepted in revised form: 15 January 1998     

Stereo disparity computation using Gabor filters

A solution to the correspondence problem for stereopsis is proposed using the differences in the complex phase of local spatial frequency components. One-dimensional spatial Gabor filters (Gabor 1946; Marcelja 1980), at different positions and spatial frequencies are convolved with each member of a stereo pair. The difference between the complex phase at corresponding points in the two images is used to find the stereo disparity. Disparity values are combined across spatial frequencies for each image location. Three-dimensional depth maps have been computed from real images under standard lighting conditions, as well as from random-dot stereograms (Julesz 1971). The algorithm can discriminate disparities significantly smaller than the width of a pixel. It is possible that a similar mechanism might be used in the human visual system.     

The Role of Binocular Disparity in Stereoscopic Images of Objects in the Macaque Anterior Intraparietal Area

Neurons in the macaque Anterior Intraparietal area (AIP) encode depth structure in random-dot stimuli defined by gradients of binocular disparity, but the importance of binocular disparity in real-world objects for AIP neurons is unknown. We investigated the effect of binocular disparity on the responses of AIP neurons to images of real-world objects during passive fixation. We presented stereoscopic images of natural and man-made objects in which the disparity information was congruent or incongruent with disparity gradients present in the real-world objects, and images of the same objects where such gradients were absent. Although more than half of the AIP neurons were significantly affected by binocular disparity, the great majority of AIP neurons remained image selective even in the absence of binocular disparity. AIP neurons tended to prefer stimuli in which the depth information derived from binocular disparity was congruent with the depth information signaled by monocular depth cues, indicating that these monocular depth cues have an influence upon AIP neurons. Finally, in contrast to neurons in the inferior temporal cortex, AIP neurons do not represent images of objects in terms of categories such as animate-inanimate, but utilize representations based upon simple shape features including aspect ratio.     

Brain damage and global stereopsis.

When a single object lies in front of or beyond the plane of fixation its retinal image lies on disparate positions in the two eyes. This 'local' retinal disparity is an excellent cue to depth, and retinal disparties of a few seconds of arc are detectable by people and monkeys. However, most visual scenes produce a complex array of contours in each eye and we can detect the disparity in the arrays despite the ambiguous nature of the disparities, i.e. each contour in one eye could be related to any of several similar contours in the other eye. This ability, known as 'global' stereopsis, may be selectively impaired following brain damage in man. Global stereopsis was measured in rhesus monkeys before and after removing a different cortical visual area in different groups of animals. Only removal of the inferotemporal cortex impaired global stereopsis. The result is related to the findings with human patients and to receptive field properties of neurons in the inferotemporal cortex of monkeys.     

A model of binocular stereopsis including a global consistency constraint

 The binocular correspondence problem was solved by implementing the uniqueness constraint and the continuity constraint, as proposed by Marr and Poggio [Marr D, PoggioT (1976) Science 194: 283–287]. However, these constraints are not sufficient to define the proper correspondence uniquely. With these constraints, random-dot stereograms (RDSs), consisting of the periodic textures in each image, are treated as a correspondence of surfaces composed of patches of alternating values of disparity. This is quite different from the surface we perceive through the RDSs, that is a surface characterized by a single depth. Because these constraints are local, they cannot produce the global optimum of correspondence. To obtain the global optimum of correspondence, we propose a model of binocular stereopsis in which a global measure of correspondence is explicitly employed. The model consists of two hierarchical systems. First, the lower system processes various correspondences based on the uniqueness constraint. Second, the higher system provides a global measure of correspondence for the disparity in question. The higher system uniquely determines the global optimum of correspondence in the lower system through the recurrent loop between hierarchical systems. The convergence of the recurrent loop is determined by the consistency between the hierarchical systems. The condition is termed the `global consistency constraint. Received: 27 August 1998 / Accepted in revised form: 8 November 1999     

基于动态聚焦的声聚焦光声内窥成像算法研究

目的 声聚焦光声内窥成像具有成像深度大的优点,是一种非常有前景的功能成像技术,该技术被广泛应用于直肠、食道等内窥成像中。声聚焦光声内窥成像通常采用基于单个聚焦超声传感器的侧向扫描方式,同时采用传统的B扫描方法进行重建,会大大降低图像质量。为了获得高质量的图像,本文提出了几种动态聚焦的声聚焦光声内窥成像算法。方法 本文使用几种动态聚焦算法进行了数值仿真,并搭建系统进行了仿体实验验证,从横向分辨率和信噪比等多方面比较了各算法在动态聚焦中的成像效果。结果 相比B扫描方法,动态聚焦后的图像在离焦区域的横向分辨率与信噪比方面都有提升,仿真模拟中最高可将离焦区域的成像目标分辨率提升约26倍,其信噪比经动态聚焦后最高可提高2.3倍左右,实验中的远距离点目标经动态聚焦重建后分辨率提升3～6倍。结论 整体而言,基于时空响应的算法和合成孔径聚焦重建算法是在实验条件下更为适用的算法。本工作对后续的声聚焦光声内窥成像的设计具有指导意义。     

Dynamic Lung Modeling and Tumor Tracking Using Deformable Image Registration and Geometric Smoothing

A greyscale-based fully automatic deformable image registration algorithm, based on an optical flow method together with geometric smoothing, is developed for dynamic lung modeling and tumor tracking. In our computational processing pipeline, the input data is a set of 4D CT images with 10 phases. The triangle mesh of the lung model is directly extracted from the more stable exhale phase (Phase 5). In addition, we represent the lung surface model in 3D volumetric format by applying a signed distance function and then generate tetrahedral meshes. Our registration algorithm works for both triangle and tetrahedral meshes. In CT images, the intensity value reflects the local tissue density. For each grid point, we calculate the displacement from the static image (Phase 5) to match with the moving image (other phases) by using merely intensity values of the CT images. The optical flow computation is followed by a regularization of the deformation field using geometric smoothing. Lung volume change and the maximum lung tissue movement are used to evaluate the accuracy of the application. Our testing results suggest that the application of deformable registration algorithm is an effective way for delineating and tracking tumor motion in image-guided radiotherapy.     

Spatial stereoresolution for depth corrugations may be set in primary visual cortex

Stereo "3D" depth perception requires the visual system to extract binocular disparities between the two eyes' images. Several current models of this process, based on the known physiology of primary visual cortex (V1), do this by computing a piecewise-frontoparallel local cross-correlation between the left and right eye's images. The size of the "window" within which detectors examine the local cross-correlation corresponds to the receptive field size of V1 neurons. This basic model has successfully captured many aspects of human depth perception. In particular, it accounts for the low human stereoresolution for sinusoidal depth corrugations, suggesting that the limit on stereoresolution may be set in primary visual cortex. An important feature of the model, reflecting a key property of V1 neurons, is that the initial disparity encoding is performed by detectors tuned to locally uniform patches of disparity. Such detectors respond better to square-wave depth corrugations, since these are locally flat, than to sinusoidal corrugations which are slanted almost everywhere. Consequently, for any given window size, current models predict better performance for square-wave disparity corrugations than for sine-wave corrugations at high amplitudes. We have recently shown that this prediction is not borne out: humans perform no better with square-wave than with sine-wave corrugations, even at high amplitudes. The failure of this prediction raised the question of whether stereoresolution may actually be set at later stages of cortical processing, perhaps involving neurons tuned to disparity slant or curvature. Here we extend the local cross-correlation model to include existing physiological and psychophysical evidence indicating that larger disparities are detected by neurons with larger receptive fields (a size/disparity correlation). We show that this simple modification succeeds in reconciling the model with human results, confirming that stereoresolution for disparity gratings may indeed be limited by the size of receptive fields in primary visual cortex.     

Shading Beats Binocular Disparity in Depth from Luminance Gradients: Evidence against a Maximum Likelihood Principle for Cue Combination

Perceived depth is conveyed by multiple cues, including binocular disparity and luminance shading. Depth perception from luminance shading information depends on the perceptual assumption for the incident light, which has been shown to default to a diffuse illumination assumption. We focus on the case of sinusoidally corrugated surfaces to ask how shading and disparity cues combine defined by the joint luminance gradients and intrinsic disparity modulation that would occur in viewing the physical corrugation of a uniform surface under diffuse illumination. Such surfaces were simulated with a sinusoidal luminance modulation (0.26 or 1.8 cy/deg, contrast 20%-80%) modulated either in-phase or in opposite phase with a sinusoidal disparity of the same corrugation frequency, with disparity amplitudes ranging from 0’-20’. The observers’ task was to adjust the binocular disparity of a comparison random-dot stereogram surface to match the perceived depth of the joint luminance/disparity-modulated corrugation target. Regardless of target spatial frequency, the perceived target depth increased with the luminance contrast and depended on luminance phase but was largely unaffected by the luminance disparity modulation. These results validate the idea that human observers can use the diffuse illumination assumption to perceive depth from luminance gradients alone without making an assumption of light direction. For depth judgments with combined cues, the observers gave much greater weighting to the luminance shading than to the disparity modulation of the targets. The results were not well-fit by a Bayesian cue-combination model weighted in proportion to the variance of the measurements for each cue in isolation. Instead, they suggest that the visual system uses disjunctive mechanisms to process these two types of information rather than combining them according to their likelihood ratios.     

Synchronized audio-visual transients drive efficient visual search for motion-in-depth

In natural audio-visual environments, a change in depth is usually correlated with a change in loudness. In the present study, we investigated whether correlating changes in disparity and loudness would provide a functional advantage in binding disparity and sound amplitude in a visual search paradigm. To test this hypothesis, we used a method similar to that used by van der Burg et al. to show that non-spatial transient (square-wave) modulations of loudness can drastically improve spatial visual search for a correlated luminance modulation. We used dynamic random-dot stereogram displays to produce pure disparity modulations. Target and distractors were small disparity-defined squares (either 6 or 10 in total). Each square moved back and forth in depth in front of the background plane at different phases. The target's depth modulation was synchronized with an amplitude-modulated auditory tone. Visual and auditory modulations were always congruent (both sine-wave or square-wave). In a speeded search task, five observers were asked to identify the target as quickly as possible. Results show a significant improvement in visual search times in the square-wave condition compared to the sine condition, suggesting that transient auditory information can efficiently drive visual search in the disparity domain. In a second experiment, participants performed the same task in the absence of sound and showed a clear set-size effect in both modulation conditions. In a third experiment, we correlated the sound with a distractor instead of the target. This produced longer search times, indicating that the correlation is not easily ignored.     

Vergence eye movement control and multivalent perception of autostereograms

We introduce a dynamical model for automatic vergence eye movement control. In connection with our dynamical system of binocular model neurons that solves the correspondence problem of stereo-vision, we present a complete model for stereo-vision. Our automatic vergence eye movement control adjusts an image segment, which is of momentary interest to the observer. The adjustment is done in such a way that we only need to define a disparity search range of minimal extension. ecently, a new method of encoding (3D) three-dimenional information in 2D pictures was designed in the form of computer-generated patterns of colored dots. At first glimpse, these so-called autostereograms appear as structured but meaningless patterns. After a certain period of observation, a 3D pattern emerges suddenly in an impressive way. Applying our algorithm to autostereograms, we find a fully satisfactory agreement with the multivalent perception experienced by humans. As in nature, in our model the phase transition between the initial state and the 3D perception state takes place in a very short time. Our algorithm is very robust against noise, and there is no need to interpolate a sparse depth map.     

Depth Perception Not Found in Human Observers for Static or Dynamic Anti-Correlated Random Dot Stereograms

One of the greatest challenges in visual neuroscience is that of linking neural activity with perceptual experience. In the case of binocular depth perception, important insights have been achieved through comparing neural responses and the perception of depth, for carefully selected stimuli. One of the most important types of stimulus that has been used here is the anti-correlated random dot stereogram (ACRDS). In these stimuli, the contrast polarity of one half of a stereoscopic image is reversed. While neurons in cortical area V1 respond reliably to the binocular disparities in ACRDS, they do not create a sensation of depth. This discrepancy has been used to argue that depth perception must rely on neural activity elsewhere in the brain. Currently, the psychophysical results on which this argument rests are not clear-cut. While it is generally assumed that ACRDS do not support the perception of depth, some studies have reported that some people, some of the time, perceive depth in some types of these stimuli. Given the importance of these results for understanding the neural correlates of stereopsis, we studied depth perception in ACRDS using a large number of observers, in order to provide an unambiguous conclusion about the extent to which these stimuli support the perception of depth. We presented observers with random dot stereograms in which correlated dots were presented in a surrounding annulus and correlated or anti-correlated dots were presented in a central circular region. While observers could reliably report the depth of the central region for correlated stimuli, we found no evidence for depth perception in static or dynamic anti-correlated stimuli. Confidence ratings for stereoscopic perception were uniformly low for anti-correlated stimuli, but showed normal variation with disparity for correlated stimuli. These results establish that the inability of observers to perceive depth in ACRDS is a robust phenomenon.     

An ASIC-chip for stereoscopic depth analysis in video-real-time based on visual cortical cell behavior

In a stereoscopic system both eyes or cameras have a slightly different view. As a consequence small variations between the projected images exist ("disparities") which are spatially evaluated in order to retrieve depth information. We will show that two related algorithmic versions can be designed which recover disparity. Both approaches are based on the comparison of filter outputs from filtering the left and the right image. The difference of the phase components between left and right filter responses encodes the disparity. One approach uses regular Gabor filters and computes the spatial phase differences in a conventional way as described already in 1988 by Sanger. Novel to this approach, however, is that we formulate it in a way which is fully compatible with neural operations in the visual cortex. The second approach uses the apparently paradoxical similarity between the analysis of visual disparities and the determination of the azimuth of a sound source. Animals determine the direction of the sound from the temporal delay between the left and right ear signals. Similarly, in our second approach we transpose the spatially defined problem of disparity analysis into the temporal domain and utilize two resonators implemented in the form of causal (electronic) filters to determine the disparity as local temporal phase differences between the left and right filter responses. This approach permits video real-time analysis of stereo image sequences (see movies at http://www.neurop.ruhr-uni-bochum.de/Real- Time-Stereo) and a FPGA-based PC-board has been developed which performs stereo-analysis at full PAL resolution in video real-time. An ASIC chip will be available in March 2000.     

Visual depth encoding in populations of neurons with localized receptive fields

 Stereopsis is the ability to perceive three-dimensional structure from disparities between the two-dimensional retinal images. Although disparity-sensitive neurons have been proposed as a neural representation of this ability many years ago, it is still difficult to link all qualities of stereopsis to properties of the neural correlate of binocular disparities. The present study wants to support efforts directed at closing the gap between electrophysiology and psychophysics. Populations of disparity-sensitive neurons in V1 were simulated using the energy-neuron model. Responses to different types of stimuli were evaluated with an efficient statistical estimator and related to psychophysical findings. The representation of disparity in simulated population responses appeared to be very robust. Small populations allowed good depth discrimination. Two types of energy neurons (phase- and position-type models) that are discussed as possible neural implementations of disparity-selectivity could be compared to each other. Phase-type coding was more robust and could explain a tendency towards zero disparity in degenerated stimuli and, for high-pass stimuli, exhibited the breakdown of disparity discrimination at a maximum disparity value. Contrast-inverted stereograms led to high variances in disparity representation, which is a possible explanation of the absence of depth percepts in large contrast-inverted stimuli. Our study suggests that nonlocal interactions destroy depth percepts in large contrast-inverted stereograms, although these percepts occur for smaller stimuli of the same class. Received: 21 December 2001 / Accepted: 29 April 2002 RID="*" ID="*" Present address: Bayer AG BTS-PT-MVT-MKM, Geb. K9, 51368 Leverkusen, Germany Acknowledgement. This work was supported by a scholarship from the Studienstiftung des deutschen Volkes to J.L. Correspondence to: J. Lippert (e-mail: joerg.lippert.jl@bayer-ag.de)     

基于空间频域滤波高动态光学投影层析的斑马鱼三维成像研究

本文提出了一种基于空间频率滤波的多曝光融合的高动态投影层析三维成像方法,实现了活体斑马鱼（17 mm × 4 mm,最大厚度为2.33 mm,最小厚度为0.29 mm）的三维结构成像. 通过相机采用不同曝光时间记录系列吸收图像,将每张图像取变换到频域去除低频后,将各张滤波后叠加并逆傅里叶变换回空域,对变换后的图像进行归一化处理,最终获得高动态图像. 在每个投影角度获得这种高动态吸收投影图像,进行滤波反投影算法重建,获得高动态的整条斑马鱼三维结构信息. 实验成像结果表明,这种空间频率滤波多曝光融合的高动态光学投影层析三维成像研究,可以获得复杂结构更丰富的空间信息,对斑马鱼等模式生物早期胚胎生长发育进程进行监测和定量评估有一定的应用前景.     

Locality Constrained Joint Dynamic Sparse Representation for Local Matching Based Face Recognition

Recently, Sparse Representation-based Classification (SRC) has attracted a lot of attention for its applications to various tasks, especially in biometric techniques such as face recognition. However, factors such as lighting, expression, pose and disguise variations in face images will decrease the performances of SRC and most other face recognition techniques. In order to overcome these limitations, we propose a robust face recognition method named Locality Constrained Joint Dynamic Sparse Representation-based Classification (LCJDSRC) in this paper. In our method, a face image is first partitioned into several smaller sub-images. Then, these sub-images are sparsely represented using the proposed locality constrained joint dynamic sparse representation algorithm. Finally, the representation results for all sub-images are aggregated to obtain the final recognition result. Compared with other algorithms which process each sub-image of a face image independently, the proposed algorithm regards the local matching-based face recognition as a multi-task learning problem. Thus, the latent relationships among the sub-images from the same face image are taken into account. Meanwhile, the locality information of the data is also considered in our algorithm. We evaluate our algorithm by comparing it with other state-of-the-art approaches. Extensive experiments on four benchmark face databases (ORL, Extended YaleB, AR and LFW) demonstrate the effectiveness of LCJDSRC.     

Bathymetric patterns of morphological disparity in deep-sea gastropods from the western North Atlantic basin

Understanding patterns of species richness requires knowledge of the individual roles species play in community structure. Here, I use gastropod shells as a source of information about both their ecological and their evolutionary functions in generating bathymetric gradients of diversity. Specifically, morphological disparity of shell architecture in deep-sea gastropods is evaluated over a depth gradient in the western North Atlantic by constructing an empirical morphospace based on an eigenshape analysis. Morphological disparity is quantified by calculating the centroid, total range, and dispersion of the morphospace at each station along the depth gradient. The results indicate that local faunas are drawn from a regional pool with the same variance but that average dissimilarity in forms reflects the number of species in the sample. The range of the morphospace at local scales is also less than at regional scales, resulting from the variability of the morphospace centroid over depth. Although the position of the morphospace changes with depth, morphological disparity remains unaffected. Despite the lack of bathymetric patterns in variance, patterns in nearest neighbor distance persist. The findings suggest the importance of interacting ecological and evolutionary processes at varying spatiotemporal scales for both morphological disparity and species richness.