Bust detection algorithm

Around 2006, concerns were being raised regarding the (what was then) the newly emerging technology of 3-D full-body scanners.

“These high-resolution scanned images reveal human body details and have raised privacy concerns. Airport and transport officials in several countries are refusing to run a test trial with the scanners until a more suitable way to conceal certain parts of the human body is found.”

This prompted a research team from the Visual Intel. Studio at Carnegie Mellon University to develop computer algorithms which could identify specific parts of the human body and then block their display to scanner operators. Their technique was aimed at detecting characteristically curved areas of the female human form – creating what was in effect an automatic bust-detector.

Bust-detector-01

The bust areas were then hidden using a modesty filter for the appropriate parts of the scanner image.

Bust-detector-02

The algorithm proved to be around 90% reliable, but there was a ‘dilemma zone’ :

“[…] some over-weight males do have the curvature features. However, the over-lapped zone is small, less than 8% of the total one hundred samples.”

see: Feature hiding in 3D human body scans [InformationVisualization (2006)5, 271–278] and A Privacy Algorithm for 3D Human Body Scan

[Lecture Notes in Computer Science, LNCS 3394, Springer, 2006.]