De-identification in visual media data

摘要

A visual media de-identification system is described. The system includes an image merger and a de-identifying engine. The image merger is configured to merge a sequence of images from a set of visual media data into an averaged image. The de-identifying engine is configured to: bound portions of the averaged image that are determined to be relatively fixed, wherein each bounded portion is identified by a corresponding position in the averaged image; generate a template comprising the bounded portions and the corresponding position for each bounded portion in the averaged image; and de-identify the sequence of images by obfuscating content in the bounded portions.

主权项

1. A computer program product, comprising:
a computer readable storage device to store a computer readable program, wherein the computer readable program, when executed by a processor within a computer, causes the computer to perform operations for de-identification of visual media data, the operations comprising:
merging a sequence of images from a set of visual media data into an averaged image;bounding portions of the averaged image that are determined to be relatively fixed, wherein each bounded portion corresponds to identification information located at a relatively fixed position in a plurality of images in the sequence of images, wherein each bounded portion is identified by a corresponding position in the averaged image, and wherein bounding portions of the averaged image that are determined to be relatively fixed further comprises:
bounding connected components from the averaged image to find characters and to produce a character image;bounding words from the averaged image to produce a word image, wherein bounding words from the averaged image further comprises:
analyzing a portion of the averaged image to obtain a confidence level that the analyzed portion contains text; andestablishing the analyzed portion as a word candidate in response to determining that the confidence level meets a word threshold;retaining bounded portions in which a predetermined percentage of bounded characters from the character image and bounded words from the word image overlap;generating a template for de-identifying the sequence of images, wherein the template comprises the bounded portions and the corresponding position for each bounded portion in the averaged image;de-identifying the sequence of images by obfuscating content in the bounded portions; andestablishing viewing rights for the bounded portions, wherein at least two of the bounded portions comprise different viewing rights, wherein content in a first bounded portion is visible only to a first set of users, and content in a second bounded portion is visible only to a different, second set of users.