Evaluating the effectiveness of pixelation and blurring on masking the identity of familiar faces

Abstract

Two experiments are reported that assess how well the identity of highly familiar "famous# faces can be masked from short naturalistic television clips[ Recognition of identity was made more di.cult by either pixelating "Experiment 0# or blurring "Experiment 1# the viewed face[ Participants were asked to identify faces from both moving and static clips[ Results indicated that participants were still able to recognize some of the viewed faces\ despite these image degradations[ In addition\ moving images of faces were recognized better than static ones[ The practical and theoretical implications of these _ndings are discussed[ Copyright Þ 1990 John Wiley + Sons\ Ltd[ Correspondence to] Karen Lander\ Department of Psychology\ University of Stirling\ Stirling FK8 3LA\ UK[ E!mail] kl2Ýstir[ac[uk Contract grant sponsor] University of Stirling[ Contract grant sponsor] ESRC[ Contract grant number] R999 12 55 77[ 0 In both experiments reported\ famous faces rather than personally familiar ones have been used[ Personally familiar faces have typically been viewed in person on many occasions[ Famous face identity is much more likely to be learned from a small number of {stereotypical| instances[ Despite this di}erence\ a number of experimental studies have found no qualitative di}erence between the processing of personally familiar and famous faces "see Campbell and De Haan\ 0887^Roberts and Bruce\ 0878#[

Pixelating Familiar People in the Media: Should Masking Be Taken at Face Value?

“…Harmon and Julesz (1973) were the first to demonstrate that pixelation reduces people's ability to recognise an image. Ever since, this finding has been replicated and extended by many others (e.g., Bachmann, 1991;Costen, Parker, & Craw, 1994, 1996Lander, Bruce, & Hill, 2001;Morrone, Burr, & Ross, 1983). Bachmann (1991) for instance, found evidence for a critical amount of pixels that is sufficient for recognising a face (i.e., 18 pixels horizontally and vertically; see also Costen et al, 1994Costen et al, , 1996.…”

“…As recently noted by Lander et al (2001), the stimuli used by Bachmann (1991) contained pictures that only showed the face area, i.e., without the hair or any other body part. Moreover, the depicted people did not have any special characteristics like beards or glasses.…”

“…To clarify this issue, Lander et al (2001) conducted two experiments to investigate the effectiveness of pixelation (Experiment 1) and blurring (Experiment 2). In both experiments, participants saw television images of familiar (famous) and unfamiliar people who had to be identified.…”

Predicting Human Performance for Face Recognition

“…For example, participants can recognize the faces of well-known politicians and celebrities more accurately from videotaped images than from static images. This finding is especially salient when the faces are presented in sub-optimal viewing formats (e.g., blurred, inverted, or pixilated displays; [60,64,66,67]). Thus it seems that motion becomes more important as a cue to identity when the viewing conditions are sub-optimal.…”

A review of the changing culture and social context relating to forensic facial depiction of the dead

“…Since this research relates to facial images subject anonymity is impossible and the usual methods utilised to reduce recognition, such as the use of eye blocks or pixilation, are inappropriate for research where recognition is paramount. Some psychology research [76,77] suggests that pixelated faces and eye blocks do not hide the identity of the subject and covering up some of the face does not allow satisfactory resemblance evaluation, as we can recognise familiar faces even from poor quality images [78] with as few as 7 x 10 pixels representing the face [79]. Consent for publication may be provided by the subject and these must follow ethical research guidelines [80].…”