Biomedical subjects

Chang Hong Liu

Publications and source records attributed to Chang Hong Liu.

10 recordsLinked to original sources

PubMed2005

The accessibility of spatial channels for stereo and motion.

Using fractal noise images, we measured the dependence of D(min) on the spatial passband (spatial frequency and orientation) over which information was correlated either between the eyes for stereo or between sequential frames for motion. Without affecting the amplitude spectrum of the noise stimulus we used idealized filters to scramble the phase of components outside a pre-defined passband. Using a simple Gaussian model in which performance depends on the signal/noise within a restricted spatial region, we obtained estimates of the bandwidth of the narrowest underlying spatial frequency and orientation spectral region subserving these two comparable tasks. Spatial bandwidths varied with peak spatial frequency but were very broad approximating the spectrum of the stimulus itself. Orientation properties of the underlying mechanisms were isotropic. These results suggest that the independent activity of individual narrowband spatial channels is not perceptually accessible for these tasks.

Depth Perception↗

PubMed2005

The use of 3D information in face recognition.

Effects of shading in face recognition have often alluded to 3D shape processing. However, research to date has failed to demonstrate any use of important 3D information. Stereopsis adds no advantage in face encoding [Liu, C. H., Ward, J., & Young, A. W. (in press). Transfer between 2D and 3D representations of faces. Visual Cognition], and perspective transformation impairs rather than assists recognition performance [Liu, C. H. (2003). Is face recognition in pictures affected by the center of projection? In IEEE international workshop on analysis and modeling of faces and gestures (pp. 53-59). Nice, France: IEEE Computer Society]. Although evidence tends to rule out involvement of 3D information in face processing, it remains possible that the usefulness of this information depends on certain combinations of cues. We tested this hypothesis in a recognition task, where face stimuli with several levels of perspective transformation were either presented in stereo or without stereo. We found that even at a moderate level of perspective transformation where training and test faces were separated by just 30 cm, the stereo condition produced better performance. This provides the first evidence that stereo information can facilitate face recognition. We conclude that 3D information plays a role in face processing but only when certain types of 3D cues are properly combined.

Adolescent↗

PubMed2005

VRVision: a new tool for the display of 3-D images in behavioral research.

Researchers using three-dimensional (3-D) scenes in their studies often spend a lot of time prerendering the scenes into a set of images for later presentations. To improve efficiency, we have developed a new plug-in program for MATLAB that eliminates this step. The program is able to manipulate and present 3-D scenes in real time. It allows fine control over numerous parameters, including the mode of projection, angular rotation, lighting, surface photo textures, and animation. Images can be presented with monoscopic or stereoscopic display. It can also dynamically track viewer position and update and display an appropriate 3-D projection. The program is fully compatible with existing MATLAB utilities such as the Psychophysics Toolbox (psychtoolbox.org). It is made freely available to the research community (www.hive.hull .ac.uk/software/vrvision).

Behavior↗

PubMed2005

Recognizing faces defined by texture gradients.

Texture gradients can reveal surface orientation in a manner similar to shape from shading, and therefore provide an important cue for object recognition. In this study, we tested whether a complex 3-D object, such as a face, can be identified from texture gradients alone. The stimuli were laser-scanned faces for which the texture element was a fractal-noise pattern mapped onto the 3-D surface. An eight-alternative forced choice task was used in which participants matched a face defined by texture gradients to one of eight faces defined by shape from shading (Experiment 1) or by texture gradients (Experiment 2). On average, participants scored 24% and 18%, respectively, above chance in these experiments. Although this performance was much poorer than the performance based entirely on shape-from-shading stimuli (Experiment 3), the results suggest that texture gradient information may be used to recover surface geometry of complex objects.

Contrast Sensitivity↗

PubMed2004

Face recognition is affected by similarity in spatial frequency range to a greater degree than within-category object recognition.

Previous studies have suggested that face identification is more sensitive to variations in spatial frequency content than object recognition, but none have compared how sensitive the 2 processes are to variations in spatial frequency overlap (SFO). The authors tested face and object matching accuracy under varying SFO conditions. Their results showed that object recognition was more robust to SFO variations than face recognition and that the vulnerability of faces was not due to reliance on configural processing. They suggest that variations in sensitivity to SFO help explain the vulnerability of face recognition to changes in image format and the lack of a middle-frequency advantage in object recognition.

Adolescent↗

PubMed2003

Differential binocular input and local stereopsis.

Using fractal noise images, we measured the dependence of Dmin and Dmax for stereo on the interocular differences of spatial frequency and contrast. Dmin exhibits a strong dependence on the highest spatial frequency contained in the image, while Dmax exhibits a weaker dependence on the lowest spatial frequency contained within the image. Neither relationship was found to be different when the filtering was restricted to only one eye's image, although the effect of differential lowpass filtering for Dmin was greater than that of binocular lowpass filtering. Contrast is thought to affect stereo performance particularly when it is reduced in only one eye's image. We show that, at least for broadband fractal images representative of everyday natural images, interocular contrast differences are no more disruptive than binocular ones. These results bear upon the nature of the matching process in stereopsis. The fact that these interocular spatial frequency and contrast manipulations do not selectively degrade stereopsis beyond that expected from a consideration of purely monocular effects is consistent with matching occurring within multiple spatial channels prior to their combination.

Contrast Sensitivity↗

PubMed2003

Face recognition with perspective transformation.

The effect of perspective transformation on transfer of face training was investigated in a yes/no recognition task using face stimuli with 42 degrees, 10 degrees, or no perspective convergence. A strong dependence of recognition performance on the magnitude of perspective transformation was found, with large perspective changes such as from 42 degrees at learning to orthogonal at test producing the strongest impairment and small perspective changes such as from 10 degrees at learning to orthogonal at test the least. In a second experiment, the internal and external features of a face from different perspective convergence were artificially combined to produce identical local features between this composite image and the original but producing an impossible perspective transformation from either. The results of transfer between the composite and untouched images showed face recognition to be strongly affected by local featural similarities and relatively insensitive to global coherence of perspective transformation.

Adult↗

PubMed2003

Face recognition is robust with incongruent image resolution: relationship to security video images.

Identifying a criminal captured on conventional security video typically requires matching poor-quality video footage against a high-quality photograph. The authors examined the consequence of such a large discrepancy in image quality. Recognition and matching performance of this incongruent-quality condition was compared with that of a congruent one, in which a high-quality photograph was reduced to a low-quality video. Recognition memory was little affected by this manipulation, whereas matching performance of the incongruent condition enjoyed occasional advantage. The results show that person identification can tolerate a large discrepancy between image qualities of matching stimuli when one of the images is of poor quality.

Adolescent↗

PubMed2002

Reassessing the 3/4 view effect in face recognition.

It is generally accepted that unfamiliar faces are better recognized if presented in 3/4 view. A common interpretation of this result is that the 3/4 view represents a canonical view for faces. This article presents a critical review of this claim. Two kinds of advantage, in which a 3/4 view either generalizes better to a different view or produces better recognition in the same view, are discussed. Our analysis of the literature shows that the first effect almost invariably depended on different amounts of angular rotation that was present between learning and test views. The advantage usually vanished when angular rotation was equalized between conditions. Reports in favor of the second effect are scant and can be countered by studies reporting negative findings. To clarify this ambiguity, we conducted a recognition experiment. Subjects were trained and tested on the same three views (full-face, 3/4 and profile). The results showed no difference between the three view conditions. Our analysis of the literature, along with the new results, shows that the evidence for a 3/4 view advantage in both categories is weak at best. We suggest that a better predictor of performance for recognition in different views is the angular difference between learning and test views. For recognition in the same view, there may be a wide range of views whose effectiveness is comparable to the 3/4 view.

Adolescent↗

PubMed2002

Luminance spatial scale and local stereo-sensitivity.

Using filtered, broad band, fractal noise images we measured the dependence of D(min) and D(max) for stereo on luminance spatial frequency. D(min) was found to exhibit a simple dependence on the highest spatial frequency contained in the stimulus. D(max) depended on both image size and spatial frequency in a way that suggests an informational limit. Different rules govern D(min) and D(max) even for first order stereopsis, arguing against a common neural explanation based on independent access to the most pertinent spatial filter.

Depth Perception↗