De-identification Project

Face De-Identification

by Ralph Gross, Latanya Sweeney, Jeffrey Cohn, Fernando de la Torre, and Simon Baker

Abstract

With the emergence of new applications centered around the sharing of image data, questions concerning the protection of the privacy of people visible in the scene arise. In most of these applications knowledge of the identity of people in the image is not required. This makes the case for image de-identification, the removal of identifying information from images, prior to sharing of the data. Privacy protection methods are well established for field-structured data, however, work on images is still limited. In this chapter we review previously proposed naȧve and formal face de-identification methods. We then describe a novel framework for the deidentification of face images using multi-factor models which unify linear, bilinear, and quadratic data models. We show in experiments on a large expression-variant face database that the new algorithm is able to protect privacy while preserving data utility. The new model extends directly to image sequences, which we demonstrate on examples from a medical face database.

Keywords: Video surveillance, privacy, de-identification, privacy-preserving data mining, k-anonymity, microaggregation

Citation:
Ralph Gross, Latanya Sweeney, Jeffrey Cohn, Fernando de la Torre, and Simon Baker. In: Protecting Privacy in Video Surveillance, A. Senior, editor. Springer, 2009 Preserving Privacy by De-identifying Facial Images. (PDF).

Related Links


Copyright © 2011. President and Fellows Harvard University.   |   IQSS   |    Data Privacy Lab   |    [info@dataprivacylab.org]