System and method for identifying a particular human in images using an artificial image composite or avatar

1. A method of detecting a particular human in a plurality of images of humans, the method comprising:

using one or more processors:

receiving input data describing the appearance of the particular human;

generating a representation of the particular human based on the input data;

generating an avatar signature for the particular human based on said representation;

automatically extracting images of humans from the plurality of images and generating respective human signatures for at least some of the humans based on the extracted images;

using the representation to identify one or more of the humans in said plurality of images as a candidate for the particular human by comparing the avatar signature with the human signatures,

wherein the generation of the avatar signature comprises:

dividing the representation into avatar body parts;

determining a first set of dominant colors for each avatar body part;

dividing the avatar body parts into avatar layers; and

determining a first distribution of only the first set of dominant colors on the avatar layers, and

wherein the generation of each human signature comprises:

dividing an image corresponding to the human into human body parts;

determining a second set of dominant colors for each human body part;

dividing the human body parts into human layers; and

determining a second distribution of only the second set of dominant colors on the human layers.

2. The method of claim 1 wherein at least some of operations performed on said representation in order to generate the avatar signature for the particular human are also performed on the extracted images in order to generate said human signatures for humans included in the images.

3. The method of claim 1 comprising using one or more identified candidates to identify one or more humans in further images as further candidates for the particular human.

4. The method of claim 3 wherein identifying one or more humans in further images comprises:

comparing a human signature for an identified candidate with further human signatures for humans generated using images extracted from further video footage.

5. The method of claim 1 wherein the input is received via a graphical user interface (GUI) and said GUI presents a set of visual descriptors of a human figure and the input data comprises a selection of said visual descriptors.

6. The method of claim 1 wherein the input is received via a graphical user interface (GUI) and wherein said GUI presents a figure representing a human and is configured to receive input to change the figure to more closely resemble the particular human.

7. The method of claim 1 wherein said input data comprises a sketch of at least part of the particular human.

8. The method of claim 1 , wherein each avatar body part is divided into 3 avatar layers, and wherein each human body part is divided into 3 human layers.

9. The method of claim 1 , wherein the avatar layers and the human layers are horizontal layers.

10. The method of claim 1 , wherein the avatar layers and the human layers are of even depth.

11. A system for detecting a particular human in a plurality of images of humans, the system comprising:

one or more processors;

at least one display device for displaying a graphical user interface (GUI);

one or more input devices enabling input of data describing the appearance of the particular human via the GUI;

the one or more processors being configured to:

generate a representation of the particular human based on the input data;

generate an avatar signature for the particular human based on said representation;

extract images of humans from the plurality of images and generate respective human signatures for at least some of the humans based on the extracted images;

use the representation to identify one or more of the humans in said plurality of images as a candidate for the particular human by comparing the avatar signature with the human signatures,

wherein the generation of the avatar signature comprises:

dividing the representation into avatar body parts;

determining a first set of dominant colors for each avatar body part;

dividing the avatar body parts into avatar layers; and

determining a first distribution of only the first set of dominant colors on the avatar layers, and

wherein the generation of each human signature comprises:

dividing an image corresponding to the human into human body parts;

determining a second set of dominant colors for each human body part;

dividing the human body parts into human layers; and

determining a second distribution of only the second set of dominant colors on the human layers.

12. The system of claim 11 , wherein each avatar body part is divided into 3 avatar layers, and wherein each human body part is divided into 3 human layers.

13. The system of claim 11 , wherein the avatar layers and the human layers are horizontal layers.

14. The system of claim 11 , wherein the avatar layers and the human layers are of even depth.