Method and device for anonymizing reidentification data in a visual tracking system

1 . A visual tracking system comprising:

a plurality of image sources configured to provide images M within respective fields-of-view, each image source configured to:

detect, in images obtained from the image sources, subareas m i which each contain a tracking target; and

compute, for each subarea, a feature vector f(m i ) which represents a visual appearance of the tracking target therein;

an anonymization layer configured to:

provide first reidentification data items g i,1 =h([f(m i ), σ 1 ]) by anonymizing each feature vector using a predefined one-way function h modified by a first tracking-rights token σ 1 , which first tracking-rights token is specific to a first group of at least one of the image sources;

provide second reidentification data items g i,2 =h([f(m i ), σ 2 ]) by anonymizing each feature vector using the predefined one-way function h modified by a second tracking-rights token σ 2 , which second tracking-rights token is specific to a second group of at least one of the image sources and is distinct from the first tracking-rights token, wherein at least one image source belongs to both the first group and second group of the image sources; and

disclose the first and second reidentification data items annotated with locations X(m i ) of the respective subareas to tracking

clients, while preventing access to the feature vectors;

a plurality of tracking clients configured to perform reidentification of tracking targets using data obtained from the anonymization layer, including:

a first tracking client, which is authorized to perform reidentification in fields-of-view of the first group of image sources and to acquire the location-annotated first reidentification data items; and

a second tracking client, which is authorized to perform reidentification in fields-of-view of the second group of image sources and to acquire the location-annotated second reidentification data items; and

wherein the anonymization layer is implemented in processing circuitry separate from the tracking clients.

2 . The visual tracking system of claim 1 , wherein the anonymization layer is implemented in processing circuitry of a coordinating entity in the visual tracking system and/or is implemented in processing circuitry which is co-located with respective ones of the image sources, wherein the coordinating entity is implemented in a dedicated server or host computer in a computer network, or as a process executing on a computer or a virtual machine with additional responsibilities in the visual tracking system.

3 . A method of providing anonymized data for facilitating reidentification in a visual tracking system, the method comprising:

detecting, in images M obtained from a plurality of image sources, subareas m i which each contain a tracking target;

computing, for each subarea, a feature vector f(m i ) which represents a visual appearance of the tracking target therein;

for a first group of the image sources;

providing first reidentification data items g i,1 =h([f(m i ), σ 1 ]) by anonymizing each feature vector using a predefined one-way function h modified by a first tracking-rights token σ 1 ; and

disclosing the first reidentification data items annotated with locations X (m i ) of the respective subareas to a first tracking client;

for a second group of the image sources;

providing second reidentification data items g i,2 =h([f(m i ), σ 2 ]) by anonymizing each feature vector using the predefined one-way function h modified by a second tracking-rights token σ 2 which is distinct from the first tracking-rights token; and disclosing the second reidentification data items annotated with locations of the respective subareas to a second tracking client; and

preventing access to the feature vectors,

wherein the first tracking-rights token σ 1 is specific to a group of image sources that includes the first group, and the second tracking-rights token σ 2 is specific to a group of image sources that includes the second group, wherein the first and second groups of the image sources overlap at least partially; and

wherein at least one image source belongs to both the first group and second group of the image sources.

4 . The method of claim 3 , further comprising:

ceasing to use the first tracking-rights token σ 1 after expiry of a validity period thereof.

5 . The method of claim 4 , further comprising:

performing a negotiation between a device performing the method and at least one further device which uses the first tracking-rights token σ 1 concerning the expiry of the validity period thereof, wherein the negotiation ends with a unanimous approval of a proposed expiry time.

6 . The method of claim 4 , further comprising:

receiving an indication of the expiry of the validity period of the first tracking-rights token σ 1 decided by a coordinating entity in a visual tracking system,

wherein the coordinating entity is implemented in a dedicated server or host computer in a computer network, or as a process executing on a computer or a virtual machine with additional responsibilities in the visual tracking system.

7 . The method of claim 4 , wherein, after the expiry of the validity period of the first tracking-rights token σ 1 , the feature vectors for the first group of the image sources are anonymized using the predefined one-way function h modified by a substitute first tracking-rights token σ 1 ′.

8 . The method of claim 3 , wherein:

the computation of the feature vectors f(m i ) takes values in a discrete set; and/or

the one-way function h used for providing the reidentification data items g i,1 , g i,2 is proximity-preserving.

9 . The method of claim 3 , wherein the providing of the reidentification data items includes applying the tracking-rights tokens σ 1 , σ 2 that modify the predefined one-way function as cryptographic salts.

10 . The method of claim 3 , wherein the tracking targets include a person and/or an inanimate object.

11 . A method for performing visual tracking of tracking targets in fields-of-view of a first group of image sources, the method comprising:

at an anonymization layer in a visual tracking system, providing anonymized data by performing the method of claim 3 ; and,

at a first tracking client in the visual tracking system, performing the following steps:

receiving the first reidentification data items g i,1 , each first reidentification data item annotated with the location of the respective subarea;

searching for matching reidentification data items among the first reidentification data items; and

for a set of mutually matching reidentification data items, following one tracking target on the basis of the locations with which the mutually matching reidentification data items are annotated.

12 . The method of claim 11 , wherein said searching for matching reidentification data items is restricted in time to a validity period of the first tracking-rights token σ 1 .

13 . The method of claim 11 , further comprising:

at a second tracking client in the visual tracking system, performing the following steps:

receiving the second reidentification data items, each second reidentification data item annotated with the location of the respective subarea;

searching for matching reidentification data items among the first reidentification data items; and

for a set of mutually matching reidentification data items, following one tracking target on the basis of the locations with which the mutually matching reidentification data items are annotated.

14 . The method of claim 12 , further comprising:

at a second tracking client in the visual tracking system, performing the following steps:

receiving the second reidentification data items, each second reidentification data item annotated with the location of the respective subarea;

searching for matching reidentification data items among the first reidentification data items; and

for a set of mutually matching reidentification data items, following one tracking target on the basis of the locations with which the mutually matching reidentification data items are annotated.

15 . A device or device cluster comprising processing circuitry configured to perform the method of claim 3 .

16 . A non-transitory computer-readable storage medium comprising instructions which, when the instructions are executed by a computer, cause the computer to perform the method of claim 3 .