The science of body metrics

Philip Treleaven, Adrian Furnham and Viren Swami on a new area with implications for psychological research.
Your body shape and language contain a wealth of personal information that has a major influence on your life. Now 3D photo-booths are able to instantaneously capture a highly accurate digital map of your body and gestures, and feed this to computers for analysis of your body metrics.

Your body shape and language contain a wealth of personal information that has a major influence on your life. Now 3D photo-booths are able to instantaneously capturea highly accurate digital map of your body and gestures, and feed this to computers for analysis of your body metrics.

The belief that you can read the character of another from their outward appearance has an exceptionally long history. The first indications of a developed theory of ‘physiognomy’ appear in fifth-century bc Athens, where one Zopyrus was said to be an expert in the art. By the fourth century, Aristotle and Plato were making frequent reference to physiognomy, linking physical beauty with moral goodness. Historically, however, physiognomy reached its zenith in the second half of the 18th century, when Johann Caspar Lavater, a Swiss theologian and mystic, published his four-volume Essays on Physiognomy. Though this particular version and its modern variants are quite rightly characterised as pseudosciences, measurement of the human body has been fundamental to the study of psychology.

In recent decades, technological advances have allowed the provision of detailed information about the internal structure of the human body. Magnetic resonance imaging (MRI), CT scans and ultrasound have revolutionised our capacity to study physiology and anatomy in vivo, and to diagnose and monitor a multitude of disease states. The external measurements of the body can be just as useful. For instance, the production of prostheses is dependent on obtaining detailed and highly accurate data on human body topography (Jones & Rioux, 1997). But despite these needs, until recently there had been no parallel evolution of technology providing data on body size, shape and surface topography. Auxologists continue to measure head and limb circumferences with tape measures, while clinicians and psychologists use equally outmoded forms of measurement.

Now new technology – three-dimensional (3D) body scanners and image processing – are transforming our ability to accurately measure, visualise and interpret an individual’s ‘body metrics’ (Treleaven, 2004). The capacity to automatically collect digital data on body size, shape and surface area represents a significant innovation, capable of having a major impact on psychological research and practice. First, more detailed information on body size and shape is transforming the way we look at human physical attractiveness. Second, the capacity to obtain such information has considerably enhanced psychological research concerning an individual’s health and well-being. Third, such information could guide psychological practice, both in terms
of fitness and in other less obvious applications. This article looks at the contribution of the new science of body metrics to these aspects of psychology.

The technology
Research on the problem of building robust and accurate models to represent bodies of real humans began in the mid- and late 1990s, when the first 3D scanning devices emerged on the market (Demers et al., 1997; Jones et al., 1995). Such scanners have been developed either for measurement (the geometry of size, shape, surface area and volume) or visualisation (texture).

Several different technologies have been employed for 3D body measurement, including video silhouette images, white light phase measurement, laser-based scanning and radio wave linear analysis. The most common form of body scanners are based on an array of depth sensors, which fire a laser beam at an object and measure the direction of the reflected ray in order to calculate the distance between the sensor lens and the incident surface. With the location and orientation of the sensor known from the scanner’s construction and calibration, it is relatively easy to calculate the coordinates of the point in space.

Hundreds of measurements can be extracted using a scan lasting only several seconds. This produces a so-called point cloud (see Figure 1) from which a computer extracts surface details, such as measurements associated with surface features (body landmarks). The successful location of body landmarks is the key to reliable automated extraction of measurements from 3D body scans. If a group of participants are scanned, the database can be searched efficiently for comparison, analysis and viewing (see over) of averaged scans within a range of body shapes, using software tools developed for this purpose (Ruiz et al., 2002).

In this article we are primarily concerned with 3D whole body scanners, most of which have been designed for measurement capture rather than visualisation (e.g. scanners that visualise skin pores, freckles and so on). The former are relatively cheap compared with other imaging devices, they are relatively small, the running costs negligible, and there is minimal need for operator training. The technique also has few ethical issues aside from data security, and can be readily used longitudinally. Data from different scanners may be merged, allowing use in different contexts, such as clinical specialities and research studies. Now we turn to some of the areas in which body scanners have begun to have an impact.

Physical attraction
A long-standing debate within the psychological literature concerns the attributes that people use when judging the physical attractiveness of potential partners (see Swami & Furnham, in press). Early studies highlighted the waist-to-hip ratio (WHR) as a measure of body shape used in attractiveness judgements, especially of women (e.g. Singh, 1993), but such studies typically made use of two-dimensional line drawings. Such stimuli are thought to lack ecological validity (Tassinary & Hansen, 1998); they are often impoverished and unrealistic, relying on a single original image from which modifications are made. In addition, the WHR can be measured either as the circumference of the waist divided by the circumference of the hips, or across the waist divided by across the hips. Two-dimensional line drawings, of course, are only able to make the latter measurement.

Technological advance has meant that photographic stimuli have replaced line drawings as the dominant feature of such psychological studies. When these
are used, body weight rather than the WHR appears as the dominant cue
of attractiveness, at least of female attractiveness (Swami, & Tovée, 2005a; Tovée et al., 2002); conversely, the waist-to-chest ratio is the primary cue of male attractiveness (Maisey et al., 1999; Swami & Tovée, 2005b).

Although the use of high-resolution photographic images is more realistic than the line drawings used previously, two-dimensional images are unlikely to capture all the visual cues available from a three-dimensional image seen from the same viewing point (DeSoto & Kopp, 2003). Because of this, 3D body scanners have had a major influence on the way psychologists now study physical attraction. For example, using the [TC]2 body scanner, Fan and his colleagues (2004, 2005) obtained 3D body measurements which were then used to create short movie clips for viewing and rating the attractiveness. Their results highlight the relative power of body scanners. In viewing 2D female images, Tovée et al. (2002) believed that the perimeter–area ratio (PAR) was used as a visual cue to body weight, given the fact that PAR has a strong linear relationship with body weight. In viewing 3D female body images or real female bodies, the use of PAR becomes unnecessary: human observers are able to use other visual cues for assessing attractiveness.

Fan et al. (2004, 2005) showed that the body volume divided by the square of the height (the volume–height index) explained about 90 per cent of the variance of attractiveness ratings, which was more than body weight (which explained about 80 per cent of the variance). Three-dimensional body scanners have therefore led to the belief that the volume–height index is the most direct visual determinant of male and female physical attractiveness. In the future, it is expected that body scanners could play a more important role in understanding such issues as obesity stigmatisation and attractiveness stereotypes, providing an avenue into research previously closed off by limitations in design and technology.

Health and well-being

Three-dimensional body scanners also have the potential to change the way clinicians and psychologists examine the health and well-being of individuals. For example, large-scale anthropometric surveys have been fundamental to epidemiology for over half a century. Growth has long been used as a key proxy for health during childhood, and longitudinal cohort studies have provided crucial information about the relationships between early life experience and later outcomes. In this context, a 3D scan provides a wealth of information about the size and shape the body, from which sophisticated indices can be extracted.

An illustration of the potential for 3D body scanning for epidemiological surveys is the recent adult sizing surveys conducted for the clothing industry, SizeUK and SizeUSA (Treleaven, 2004). The United Kingdom National Sizing Survey (see used 3D whole body scanners to measure 11,000 adult participants, from which 130 body measurements were automatically extracted (leaving only about 10 measurements to be taken by hand, such as head girth and special measurements of the hands and feet). This highly accurate size and shape data, together with each participant’s 3D point cloud, is held securely in an anthropometrics web-database, available for online data mining. Information in the database includes personal details on each of the 11,000 adult participants, 140 measurements per participant (over 1.5 million measurements in total) and 22,000 scans (one standing and one seated per participant). The database also contains 24,000 pages of analysis, including statistical analyses of the measurement information, 3D shape analysis and cross-correlations charts of measurements and 3D shape. The SizeUK body data, originally collected for the clothing industry, is now being used for medical and psychological research into obesity, burns treatments and other applications.

Due to its non-invasiveness and ease of use, 3D body scanning is ideal for screening large populations of participants, where multiple conditions could be screened simultaneously. For example, scanners could be used to more accurately classify participants suffering from weight-related problems (e.g. anorexia, obesity). Body weight is typically classified according to BMI (Bray, 1998), though the BMI is unable to distinguish between fat and lean tissue. Three-dimensional scanning would dispense with this problem, though ethical and data protection issues will need to be addressed satisfactorily.

Additionally, body scanners could potentially be used to improve the diagnosis of a wide range of body deformities that arise from genetic constitution, trauma, pathology, or occupation (Jones & Rioux, 1997), as well as certain body image or eating disorders. For instance, people suffering from body image disorders sometimes exhibit either a distorted perception of their body shape, or hold unrealistic ideals regarding body size. Current psychological instruments, such as questionnaires and line drawings, make it very difficult to describe these ideals and expectations, relating as they do to specific 3D features of the body (e.g. Parkinson et al., 1998). In this regard, 3D body scanners will allow the user to manipulate 3D images to describe their present and ideal body image.

Three-dimensional body scanning is also useful for the monitoring of longitudinal changes in body morphology, whether due to exercise, nutrition or diet programmes, or as part of clinical treatment. Repeated scans of a participant can be overlaid to show changes in body shape or size over time, allowing the clinician to monitor treatment efficacy while also engaging patient interest and hence increasing patient motivation, whether for the treatment of clinical obesity or sculpting the body for aesthetic appeal.

Also in the pipeline are scanners that would be able to extract not just geometric information about the participants but colour scans of their skin as well. In medicine, for example, doctors could use this technology to screen for any changes to melanomas or skin cancer. In terms of biophysiological research, 3D scanners have allowed important advances in, for instance, the developmental genetics of craniofacial variations, bone and joint disease, and the design and fit of products that go on or around the human body (see Hoppa & Nelson, in press).

In addition, using data from 3D scanners will make it possible to model changes that occur during speech and facial expressions. A model that can synthesise facial expression, jaw movement and speech could be used for animation, in the treatment of speech impairments, or for the study of certain eating behaviours. Improved simulation of effects of surgery or treatment could be obtained by creating a dynamic synthesis of the predicted face movement before and after treatment, as opposed to static views currently used. In particular, such applications have the potential to improve understanding of treatment effects since they may be generated using the patient’s own data, thus personalising the campaign.

Three-dimensional body scanning, we believe, is poised to become a mainstream psychological tool of major value. Who knows what other applications could be out there? For example, in the games industry, players frequently paste photographs of their faces on to the stock digital avatars used in online combat arenas. Body scanners could allow them to paste accurate models of their entire bodies online. More importantly, avatars are also used in psychological research into problems such as paranoia (see Freeman & Garety, 2004: Perhaps body scans could be used to increase the realism of such simulations.

As with any new healthcare technology, there are a number of ethics, privacy and data protection issues to be considered. Privacy and legal issues impact on almost all aspects of research with scanners, including the recruitment of participants, getting participants’ agreement to use the data, protecting personal data from unauthorised access, and how the data can be disseminated. In Britain for example, the Data Protection Act of 1998 stipulates that participants be given fair notice of how, why and by whom their personal data may be used (it is also necessary for participants to consent to data being used in such a way).

Fortunately, projects such as SizeUK have developed highly secure database systems and online data mining tools for protecting the integrity and privacy of participants’ body scanning data. For instance, the database on which personal data is stored in the SizeUK survey is maintained in accordance with data protection law and the security principle of the 1998 act. Online access to the SizeUK database by clothing companies and academics must also comply with the Data Protection Act. In this instance, disclosure can only be made to those parties of which the participant is aware and in respect of whom consent has been given. To further comply with the Act, the SizeUK software tools provide only anonymous data to users of the database, and do not allow users to download individual point clouds and measurements.

In our view, ethical and privacy issues are outweighed by the enormous potential benefits offered to psychological research as a result of 3D body scanning. Perhaps readers will be able to think of more applications in research and practice, to take this technological development to the next level.

Professor Philip Treleaven is at the Department of Computer Science, University College London. E-mail: [email protected].
Professor Adrian Furnham and Dr Viren Swami are at the Department of Psychology, University College London.
E-mail: [email protected]; [email protected].




BPS Members can discuss this article

Already a member? Or Create an account

Not a member? Find out about becoming a member or subscriber