Morphing techniques for manipulating face images

This paper describes morphing techniques to manipulate two-dimensional human face images and three-dimensional models of the human head. Applications of these techniques show how to generate composite faces based on any number of component faces, how to change only local aspects of a face, and how to generate caricatures and anticaricatures of faces. These techniques are potentially useful for many psychological studies because they permit realistic images to be generated with precise control.     

Is face distinctiveness gender based?

Two experiments were carried out to study the role of gender category in evaluations of face distinctiveness. In Experiment 1, participants had to evaluate the distinctiveness and the femininity-masculinity of real or artificial composite faces. The composite faces were created by blending either faces of the same gender (sexed composite faces, approximating the sexed prototypes) or faces of both genders (nonsexed composite faces, approximating the face prototype). The results show that the distinctiveness ratings decreased as the number of blended faces increased. Distinctiveness and gender ratings did not covary for real faces or sexed composite faces, but they did vary for nonsexed composite faces. In Experiment 2, participants were asked to state which of two composite faces, one sexed and one nonsexed, was more distinctive. Sexed composite faces were selected less often. The results are interpreted as indicating that distinctiveness is based on sexed prototypes. Implications for face recognition models are discussed.     

Attractiveness of own-race, other-race, and mixed-race faces

Averaged face composites, which represent the central tendency of a familiar population of faces, are attractive. If this prototypicality contributes to their appeal, then averaged composites should be more attractive when their component faces come from a familiar, own-race population than when they come from a less familiar, other-race population. We compared the attractiveness of own-race composites, other-race composites, and mixed-race composites (where the component faces were from both races). In experiment 1, Caucasian participants rated own-race composites as more attractive than other-race composites, but only for male faces. However, mixed-race (Caucasian/Japanese) composites were significantly more attractive than own-race composites, particularly for the opposite sex. In experiment 2, Caucasian and Japanese participants living in Australia and Japan, respectively, selected the most attractive face from a continuum with exaggerated Caucasian characteristics at one end and exaggerated Japanese characteristics at the other, with intervening images including a Caucasian averaged composite, a mixed-race averaged composite, and a Japanese averaged composite. The most attractive face was, again, a mixed-race composite, for both Caucasian and Japanese participants. In experiment 3, Caucasian participants rated individual Eurasian faces as significantly more attractive than either Caucasian or Asian faces. Similar results were obtained with composites. Eurasian faces and composites were also rated as healthier than Caucasian or Asian faces and composites, respectively. These results suggest that signs of health may be more important than prototypicality in making average faces attractive.     

NEWBORNS FORM "PROTOTYPES" IN LESS THAN 1 MINUTE

Abstract— Past research demonstrated that newborns looked longer at live faces and two-dimensional still images of the mother than at the faces and images of strangers (Bushnell, Sai, & Mullin, 1989; Field, Cohen, Garcia, & Greenberg, 1985; Walton, Bower, & Bower, 1992). This study examined schema (prototype) theory as a basis for newborn's preference for the mother's face by creating composite (prototype) faces with a pixel averaging technique. Results obtained with a preferential operant-sucking procedure indicated that formation of a representation of faces was rapid. Newborns preferred to look at a composite of presented faces rather than a composite of unseen faces on the first-look presentation of each. The effect faded over the duration of the experiment.     

The effect of face orientation on holistic processing

Holistic processing, a hallmark of face processing, can be measured by the composite face effect: adults have difficulty recognising the top half of a face when it is aligned with a new bottom half, unless holistic processing is disrupted by misaligning the two halves. Like the recognition of facial identity, holistic processing is impaired when faces are inverted. To obtain a more refined measure of the influence of orientation on holistic face processing, we administered the composite face task to adults at each of seven orientations. In both experiment 1 (in which orientations were randomly intermixed) and experiment 2 (in which orientations were blocked), the composite face effect decreased linearly with rotation. We conclude that holistic processing is tuned to upright faces, diminishes as faces deviate from upright, and becomes insignificant once faces reach a sideways orientation. We propose a hierarchical model of face perception in which linear decreases in holistic processing underlie qualitative shifts in other aspects of face perception.     

Judging approachability on the face of it: the influence of face and body expressions on the perception of approachability

The aim of the current study was to examine how emotional expressions displayed by the face and body influence the decision to approach or avoid another individual. In Experiment 1, we examined approachability judgments provided to faces and bodies presented in isolation that were displaying angry, happy, and neutral expressions. Results revealed that angry expressions were associated with the most negative approachability ratings, for both faces and bodies. The effect of happy expressions was shown to differ for faces and bodies, with happy faces judged more approachable than neutral faces, whereas neutral bodies were considered more approachable than happy bodies. In Experiment 2, we sought to examine how we integrate emotional expressions depicted in the face and body when judging the approachability of face-body composite images. Our results revealed that approachability judgments given to face-body composites were driven largely by the facial expression. In Experiment 3, we then aimed to determine how the categorization of body expression is affected by facial expressions. This experiment revealed that body expressions were less accurately recognized when the accompanying facial expression was incongruent than when neutral. These findings suggest that the meaning extracted from a body expression is critically dependent on the valence of the associated facial expression.     

Robust representations for face recognition: the power of averages

We are able to recognise familiar faces easily across large variations in image quality, though our ability to match unfamiliar faces is strikingly poor. Here we ask how the representation of a face changes as we become familiar with it. We use a simple image-averaging technique to derive abstract representations of known faces. Using Principal Components Analysis, we show that computational systems based on these averages consistently outperform systems based on collections of instances. Furthermore, the quality of the average improves as more images are used to derive it. These simulations are carried out with famous faces, over which we had no control of superficial image characteristics. We then present data from three experiments demonstrating that image averaging can also improve recognition by human observers. Finally, we describe how PCA on image averages appears to preserve identity-specific face information, while eliminating non-diagnostic pictorial information. We therefore suggest that this is a good candidate for a robust face representation.     

Facial composite production: a comparison of mechanical and computer-driven systems

Computer-driven systems for constructing composite faces of suspects (E-fit; Mac-a-Mug) have largely replaced mechanical systems (Photofit; the Identikit) in police use, yet little is known of their comparative effectiveness in rendering an accurate likeness. Participants (N = 24) constructed 2 of 4 familiar or unfamiliar faces, for one of which they used Photofit and for the other, E-fit. A likeness of each face was made first under target-absent conditions and then with photographs of the target present. The accuracy of the resulting composites was assessed by familiarity ratings, names elicited, and matching accuracy. The computer-driven system showed consistent superiority only when a familiar face was constructed in the presence of photographs; when participants worked from memory, E-fit was no better than Photofit. The implications of these findings for theories of face retrieval and the operational use of composites are discussed.     

Holistic face processing can be independent of gaze behaviour: Evidence from the composite face illusion

People tend to perceive identical top halves (i.e. above the nose) of two face stimuli as being different when they are aligned with distinct bottom halves. This composite face illusion is generally considered as the most compelling evidence that facial features are integrated into a holistic representation. Here, we recorded eye‐movements during the composite face illusion in a delayed matching task of top halves of faces. Behavioural results showed a strong composite face effect, participants making more mistakes and taking longer time to match two identical top halves of faces when they were aligned (vs. misaligned) with different bottom halves. Importantly, fixation sites and eye‐movements were virtually identical when the top and bottom parts were aligned (composite illusion) or misaligned (no illusion), indicating that holistic face processing can be independent of gaze behaviour. These findings reinforce the view that holistic representations of individual faces can be extracted early on from information at a relatively coarse scale, independently of overt attention.     

Are faces perceived as configurations more by adults than by children?

Abstract

Adult face recognition is severely hampered by stimulus inversion. Several investigators have attributed this vulnerability to the effect of orientation on encoding relational aspects of faces. Previous work has also demonstrated that children are less sensitive to orientation of faces than are adults. This has been interpreted as reflecting an increasing reliance on configural aspects of faces with increasing age and expertise.

Young, Hellawell, and Hay (1987) demonstrated that for adults the encoding of relations among facial parts is, indeed, sensitive to orientation. When chimeric faces are upright, the top half of one face fuses with the bottom half of the other, making the person depicted in the top half difficult to recognize. This effect (the composite effect) is not seen when the faces are inverted. The present study obtained the composite effect for 6-year-old and 10-year-old children, just as for adults. The composite effect was found to an equal degree at all ages tested and was seen both in tasks involving highly familiar faces and in those involving newly learned, previously unfamiliar faces. Thus, these data provided no support for the hypothesis of increasing reliance on configural aspects of faces with increasing age, at least in the sense tapped by this procedure.

However, the data did confirm an Age X Orientation interaction. In recognizing both familiar and previously unfamiliar faces, 6-year-olds were less affected by inversion than were 10-year-olds, who, in turn, were less affected than were adults. Increasing vulnerability to inversion of faces with age was independent of the composite effect. Apparently, there are two distinct sources to the large effect of inversion that characterizes adult face encoding: one seen throughout development and one acquired only with expertise.     

Searching for faces in scrambled scenes

While much is known about how faces are recognized, little is known about how a face is first detected. Five face-detection experiments investigated how faces are localized and detected among either scrambled natural backgrounds or inverted faces. The first two experiments revealed that faces pop out among the former but must be searched for serially among the latter. Two subsequent experiments investigated what properties of a face allow for this pop-out. Colour removal or reversal, blurring, and inversion all had small effects on the visual search slope. The only transformation that reduced the parallel nature of face detection was luminance reversal. These results are considered in the light of models of face processing and systems for automatic face detection.     

Processing Differences between Feature‐Based Facial Composites and Photos of Real Faces

Computer‐generated faces (composites) constructed by select2ing individual facial features (e.g. eyes, nose, mouth) are poorly recognized because this process contrasts with the natural holistic processing of real faces. This result suggests that there should be differences in the cognitive processing of these composites compared with photos of real faces, which would make these stimuli problematic for theories seeking to explain real face processing. We conducted five experiments to test potential conditions for moving composite processing closer to how real face photos are processed, first taking the perspective of researchers who construct composites with a random selection of available features and then taking a perspective closer to police by creating each composite to match a real face photo. Composites with randomly selected features (but configured like real faces) showed no face inversion effect and recognition memory for these composites benefited from increased encoding time, unlike real face photos. Although composites constructed to match real face photos yielded an inversion effect, they still were remembered differently than the photos. Researchers should not use feature‐based composites as proxies for real face photos. We conclude with a discussion of alternative methods of constructing composites. Copyright © 2012 John Wiley & Sons, Ltd.     

The composite face illusion: A whole window into our understanding of holistic face perception

Two identical top halves of a face are perceived as being different when their bottom halves belong to different faces, showing that the parts of a face cannot be perceived independently from the whole face. When this visual illusion is inserted in a matching task, observers make more mistakes and/or are slower at matching identical top face halves aligned with different bottom halves than when the bottom halves are spatially offset: The composite face effect. This composite face paradigm has been used in more than 60 studies that have provided information about the specificity and nature of perceptual integration between facial parts (“holistic face perception”), the impairment of this process in acquired prosopagnosia, its developmental course, temporal dynamics, and neural basis. Following a review of the main contributions made with the paradigm, I explain its rationale and strengths, and discuss its methodological parameters, making a number of proposals for its optimal use and refinement in order to improve our understanding of holistic face perception. Finally, I explain how this standard composite face paradigm is fundamentally different than the application to facial parts of a congruency/interference paradigm that has a long tradition in experimental psychology since Stroop (1935), and which was originally developed to measure attentional and response interference between different representations rather than perceptual integration. Moreover, a version of this congruency/interference paradigm used extensively over the past years with composite faces lacks a baseline measure and has decisional, attentional, and stimulus confounds, making the findings of these studies impossible to interpret in terms of holistic perception. I conclude by encouraging researchers in this field to concentrate fully on the standard composite face paradigm, gaze contingency, and other behavioural measures that can help us take one of the most important challenges of visual perception research: Understanding the neural mechanisms of holistic face perception.     

Configural information in gender categorisation

The role of configural information in gender categorisation was studied by aligning the top half of one face with the bottom half of another. The two faces had the same or different genders. Experiment 1 shows that participants were slower and made more errors in categorising the gender in either half of these composite faces when the two faces had a different gender, relative to control conditions where the two faces were nonaligned or had the same gender. This result parallels the composite effect for face recognition (Young et al, 1987 Perception 16 747-759) and facial-expression recognition (Calder et al, 2000 Journal of Experimental Psychology: Human Perception and Performance 26 527-551). Similarly to responses to face identity and expression, the composite effect on gender discrimination was disrupted by inverting the faces (experiment 2). Both experiments also show that the composite paradigm is sensitive to general contextual interference in gender categorisation.     

Differences in holistic processing do not explain cultural differences in the recognition of facial expression

The aim of this study was to investigate the causes of the own-race advantage in facial expression perception. In Experiment 1, we investigated Western Caucasian and Chinese participants’ perception and categorization of facial expressions of six basic emotions that included two pairs of confusable expressions (fear and surprise; anger and disgust). People were slightly better at identifying facial expressions posed by own-race members (mainly in anger and disgust). In Experiment 2, we asked whether the own-race advantage was due to differences in the holistic processing of facial expressions. Participants viewed composite faces in which the upper part of one expression was combined with the lower part of a different expression. The upper and lower parts of the composite faces were either aligned or misaligned. Both Chinese and Caucasian participants were better at identifying the facial expressions from the misaligned images, showing interference on recognizing the parts of the expressions created by holistic perception of the aligned composite images. However, this interference from holistic processing was equivalent across expressions of own-race and other-race faces in both groups of participants. Whilst the own-race advantage in recognizing facial expressions does seem to reflect the confusability of certain emotions, it cannot be explained by differences in holistic processing.     

Learning influences the encoding of static and dynamic faces and their recognition across different spatial frequencies

Studies on face recognition have shown that observers are faster and more accurate at recognizing faces learned from dynamic sequences than those learned from static snapshots. Here, we investigated whether different learning procedures mediate the advantage for dynamic faces across different spatial frequencies. Observers learned two faces—one dynamic and one static—either in depth (Experiment 1) or using a more superficial learning procedure (Experiment 2). They had to search for the target faces in a subsequent visual search task. We used high-spatial frequency (HSF) and low-spatial frequency (LSF) filtered static faces during visual search to investigate whether the behavioural difference is based on encoding of different visual information for dynamically and statically learned faces. Such encoding differences may mediate the recognition of target faces in different spatial frequencies, as HSF may mediate featural face processing whereas LSF mediates configural processing. Our results show that the nature of the learning procedure alters how observers encode dynamic and static faces, and how they recognize those learned faces across different spatial frequencies. That is, these results point to a flexible usage of spatial frequencies tuned to the recognition task.     

基于儒家文化的“帝王相”之民间意象的首次探索

本研究通过评价不同性别二态线索和吸引力的面孔图片来考察儒家文化下人们心中帝王面孔形象。采用FaceGen Modeller 3.1操作面孔性别二态线索,并通过PhotoShop CS5合成面孔材料。研究发现:被试认为女性化的男性面孔比男性化的男性面孔更具"帝王相";低吸引力的女性化男性面孔比高吸引力的女性化男性面孔更具"帝王相";不同性别被试之间的评价无显著差异。上述结果显示,在儒家文化影响下,人们偏好具有女性化面孔特点的帝王。     

The role of color in human face detection

Significant advances have been made in understanding human face recognition . However, a fundamental aspect of this process, how faces are located in our visual environment, is poorly understood and little studied. Here we examine the role of color in human face detection. We demonstrate that detection performance declines when color information is removed from faces, regardless of whether the surrounding scene context is rendered in color. Furthermore, faces rendered in unnatural colors are hard to detect, suggesting a role beyond simple segmentation. When faces are presented such that half the surface is colored appropriately, and half unnaturally, performance declines. This suggests that observers are not simply using the presence of skin color "patches" to detect faces. Rather, our data suggest that detection operates via a face template combining diagnostic color and face-shape information. These findings are consistent with color-template approaches used in some computer-based face detection systems.     

Developmental changes in the recognition and comprehension of facial expression: implications for frontal lobe function.

Children aged 6-15 years old and adults (over 18) were given three tests designed to test perception and comprehension of facial expression. In the first test subjects were given two composite symmetrical faces made from the left or right half of a normal face, the subjects' task being to indicate which composite more closely resembled the original face. In the second test the subjects matched a series of photographs from Life magazine with key photographs of one of six distinct emotions (sad, fear, happy, anger, disgust, surprise). In the third test the subjects chose a key photograph that was appropriate for the face of a faceless character in a cartoon. On the composite faces test the subjects in all groups exhibited a preference for the left visual field composite, implying that all age groups were processing the faces in a similar manner. The results of the other two tests showed that there was an improvement in the perception of facial expression between the ages of 6 and 8 years, little change until about 13 years, and then a second improvement to adult performance at about 14 years. The performance of the 8- to 13-year-old children was similar to that of adult patients with frontal lobe injuries, which could be taken as evidence that the regions of the frontal lobe involved in the performance of these tasks may not be mature until about 14 years of age.