Object Tracking Sensor and Mask Generator Effectively Anonymize Camera Images to Protect Individual Privacy
This de-identification assembly preserves personal anonymity by masking faces that are captured by camera/video enabled goggles, smart glasses, wearable cameras, or other imaging devices while ensuring that the cameras maintain nearly 100 percent utility. An estimated 30 million surveillance cameras operate in the United States, constantly observing, and personal anonymity is at an all-time low. This is especially unsettling to members of the public who need to preserve their anonymity for their own protection, such as members of the witness protection program, domestic abuse victims, and individuals with stalkers. University of Florida researchers have found a way to restore individual privacy by utilizing camera technology and k-anonymity. This de-identification assembly will preserve privacy by masking faces observed through imaging devices. Rather than blurring or blocking facial images, the assembly applies k-Same principles, taking the average of k face images similar to the original and imposing this image as a mask. In some cases this pre-capture de-identification assembly may be used to preserve privacy while maintaining an ability to recognize individuals in a specific membership class, such as terrorists.
De-identification assembly that masks facial features in order to protect personal privacy
- Masks facial features for k-anonymity, reducing liability in the use of publicly displayed feeds being
- Can maintain the ability to recognize individuals such as criminals or terrorists, maintaining functional use of surveillance while also preserving privacy
This de-identification assembly utilizes k-anonymity, which is a method of preserving the privacy of data. K-anonymity ensures privacy in data sets, making it impossible to distinguish the identity of a person included in a data set while allowing the use of the data. The k-Same principle applies to images, wherein k number of images similar to the original, “averaged” together, are used as a mask to cloak faces. Researchers at the University of Florida utilized k-anonymity with imaging devices in a pre-capture de-identification assembly that can mask a face or other objects, such as well-known structures or vehicles or license tags, that might threaten privacy. In addition to surveillance cameras, the imaging device could be interfaced with wearable smart glasses, de-identifiying individuals without the wearer ever having seen the original face or image.