Home > Methodology, The Face > Standardizing Facial Photographs via IPD May Introduce Error

Standardizing Facial Photographs via IPD May Introduce Error

February 1, 2009 Leave a comment Go to comments

Much of beauty research requires comparing facial measurements from photographs of different subjects. Optimally, the distance from the sensor in the digital camera (or the film plane) to each subject’s face should be identical. Identical head-camera distances ensure that measurements taken from the photographs retain identical proportionality to measurements taken directly from the faces themselves.

When researchers are unable to control the head-camera distance when making the photographs, it is common to standardize the photographs based on interpupillary distance (Little et.al., 2008). Most often, this sort of standardization is required for photographs obtained outside of a laboratory. To illustrate, imagine that the two images below were taken from different head-camera distances.

1. Copyright Christiaan Briggs/Wikipedia Commons/http://commons.wikimedia.org/wiki/File:Sabaa_Nissan_Militiaman.jpg

The image on the left could have been taken with a shorter head-camera distance than the one on the right. Under such circumstances, even though the actual face is identical, any measures made from these images would be different. Of course, any conclusion drawn from comparing these different measurements would be inaccurate.

The usual solution to this problem – standardizing on interpupillary distance – is illustrated below:

By enlarging the smaller image until the distance between the pupils in both images is identical, we ensure that any measures we take from these two images will be identical. Clearly, this procedure works on images like those above: when the image is identical. What happens, though, when head-camera distance is standardized via equating the interpupillary distances of different faces?

The answer depends on how similar the interpupillary distances are between the faces. Problematically, there are significant differences within and between groups (Dodgson, 2004). The data are reported in millimeters.

SEXUAL DIMORPHISM 

Sex 

Male 

Female 

Range 

52-78 

52-76 

Mean 

64.67 

62.31 

St. Dev. 

3.708 

3.599 

 

Clearly, matching interpupillary distance between images of men and women will introduce bias and some degree of error. Male faces have a greater mean and standard deviation. Thus, for example, male faces might appear smaller or female faces might appear larger relative to one another. Error can also result from standardizing within sex. Any facial feature whose dimensions vary along with interpupillary distance will be similarly distorted. The sexual dimorphism apparent in this table suggests standardizing images for interpupillary distance can blur significant sex differences measureable in faces of both sexes.

ETHNIC/RACIAL GROUPS

Race

White

Black

Hispanic

Pacific Islander

American Indian

Other

Range

52-75

56-78

57-71

55-71

59-72

56-70

Mean

61.99

65.62

63.54

63.17

65.12

63.26

St. Dev

3.429

3.489

3.104

3.146

4.003

3.315

 

The same sorts of error can be introduced in data when equalizing interpupillary distance across ethnic/racial groups.

In my view, data gathered from images that have been standardized by means of equating interpupillary distance should be interpreted cautiously and tentatively. Further, I recommend that researchers avoid this practice.

 

Dodgson, N.A. (2004). Variation and extrema of human interpupillary distance. In Stereoscopic displays and virtual reality systems, XI. Proceedings of the SPIE, 18-22 January 2004. . A.J. Woods, J.O. Merritt, S.A. Benton & M.T. Bolas (eds.). San Jose California.

Anthony C. Little, Benedict C. Jones, Corri Waitt, Bernard P. Tiddeman, David R. Feinberg, David I. Perrett, Coren L. Apicella, Frank W. Marlowe (2008). Symmetry Is Related to Sexual Dimorphism in Faces: Data Across Culture and Species PLoS ONE, 3 (5) DOI: 10.1371/journal.pone.0002106

Advertisements
Categories: Methodology, The Face
  1. No comments yet.
  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: