Mooney faces are monochrome images of faces, heavily shadowed. They were introduced by Craig Mooney.
Mooney’s original paper is unclear on the construction of the faces used as stimuli: he states "The closure items were drawings of the heads and faces of miscellaneous people. They presented, in solid blacks and whites, only the salient shadows or high-lights, as revealed in strongly lighted photographs." (Mooney, 1957).
This python program generates Mooney style faces, using OpenCV and a fairly simple algorithm. It was written to investigate the effect of various parameters upon perception, including the presence or absence of a border, the highlight/shadow colour, and motion. The software was inspired by (and mostly written at) Dagstuhl, in seminar 15192 "The Message in the Shadow: Noise or Knowledge?" [https://www.dagstuhl.de/en/program/calendar/semhp/?semnr=15192].
A video of the software in action can be found on youtube here: [https://www.youtube.com/watch?v=CHvpeir3dDA]
A summary of the algorithm follows: set threshold t to mid-grey (126,126,126)
- Take input image
- Run face detector
- If face found set threshold t to mean face value
- Blur image
- Threshold by t
- Morphologically tidy up the thresholded image
- Colour in the image and draw a boundary line if required
C. M. Mooney. Age in the development of closure ability in children. Canadian Journal of Psychology, 11:219–226 1957 [http://d20uo2axdbh83k.cloudfront.net/20140401/4843d99c2f0957c1212ccbb770e363cc.pdf ]