Video Processing

• US 20080166025A1
• Filed: 11/04/2005
• Published: 07/10/2008
• Est. Priority Date: 12/02/2004
• Status: Active Grant

• Current Assignee: British Telecommunications PLC (BT Group PLC)
• Original Assignee: British Telecommunications PLC (BT Group PLC)
• Inventors: Thorne, Jeremy M
• Granted Patent: US 8,098,902 B2
• Time in Patent Office: Days
• US Class Current: 382/118
• CPC Class Codes: G06V 40/161 Detection; Localisation; Normalisation
• CPC Class Codes: H04N 7/147 Communication arrangements, e.g. identifying the communication as a video-communication, intermediate storage of the signals selecting H04Q

Abstract

A method of processing video data (300) is disclosed. The video data (300) comprises a plurality of sets of frame data and is captured by mobile video data capture means. The method comprises: (a) finding a face in each frame of the video (400); and (b) processing a corresponding set of frame data to: (i) maintain the area of the image occupied by the face substantially constant (400); and (ii) maintain the apparent direction of light incident upon the face substantially constant (500); and/or (iii) maintain the apparent colour of the face substantially constant (600).

Claims

1. A method of processing video data captured by mobile video data capture means, said video data comprising a plurality of sets of frame data, said method comprising:

(a) finding a face in each frame of said video; and

(b) processing a corresponding set of frame data to;

(i) maintain the area of the image occupied by said face substantially constant; and

(ii) maintain the apparent direction of light incident upon said face substantially constant; and

/or

(iii) maintain the apparent colour of said face substantially constant.

2. A method according to claim 1, wherein finding a face comprises identifying one ore more features on said face.

3. A method according to claim 2, wherein identifying one or more features comprises:

comparing regions of said frame data with stored feature templates, each of said feature templates comprising an image of a facial feature and corresponding in size to said regions; and

identifying each feature by identifying a match between a region of said frame data and one of said templates.

4. A method according to claim 3, wherein comparing regions comprises comparing each pixel in a region with its corresponding pixel in a template, said method further comprising generating a score for each comparison of a pixel in a region with its corresponding pixel in a stored eye template and identifying a feature by selecting the pixel having the greatest score.

5. A method according to claim 2, wherein said features comprise a pair of eyes on said face.

6. A method according to claim 5, further comprising checking the distance and angle of rotation between each of the eyes in said identified pair of eyes.

7. A method according to claim 1, wherein maintaining the apparent direction of light incident upon said face comprises:

low pass filtering said frame data

subtracting said low pass filtered version of said frame data from said frame data

adding a low pass filtered version of previously stored reference frame data to said low pass filtered version of said frame data, wherein said previously stored reference frame data comprises an image of said face under neutral lighting.

8. A method according to claim 7, wherein low pass filtering is achieved by convolving said frame data with a predefined filter kernel.

9. A method according to claim 1, wherein maintaining the apparent colour of said face comprises individually adjusting each colour channel of said frame data to maintain the relative offsets and absolute values of each colour channel substantially constant.

10. A method according to claim 9, further comprising shifting the mean pixel value in each colour channel of said frame data by a predetermined shift amount.

11. A method according to claim 10, wherein, for a selected colour channel, said shift amount corresponds to the difference between the mean pixel value of said selected channel in said frame data and the mean pixel value of said selected channel in previously stored reference frame data comprising an image of said face under neutral lighting.

15. A digital data carrier carrying a program of instructions executable by processing apparatus to perform the method steps as set out in any claim 1.

12. Video processing means including:

a storage medium having recorded therein processor readable code processable to process video data captured by mobile video data capture means, said video data comprising a plurality of sets of frame data, said code comprising;

identification code processable to identify a face in each frame of said video; and

frame data processing code processable to process a corresponding set of frame data to;

(i) maintain the area of the image occupied by said face substantially constant; and

(II) maintain the apparent direction of light incident upon said face substantially constant; and

/or

(iii) maintain the apparent colour of said face substantially constant.

13. A portable device comprising:

video data capture means, said video data comprising a plurality of sets of frame data;

a storage medium having recorded therein processor readable code processable to process said captured video data, said code comprising;

identification code processable to identify a face in each frame of said video; and

frame data processing code processable to process a corresponding set of frame data to;

(i) maintain the area of the image occupied by said face substantially constant; and

(ii) maintain the apparent direction of light incident upon said face substantially constant; and

/or

(iii) maintain the apparent colour of said face substantially constant.

14. A portable device according to claim 13, said portable device comprising a handheld device.