VIDEO SEGMENTATION METHOD

• US 20160189388A1
• Filed: 12/18/2015
• Published: 06/30/2016
• Est. Priority Date: 12/24/2014
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: PARTIS, ASHLEY JOHN, SEE, DAVID KENJI, GUPTA, AMIT KUMAR
• Granted Patent: US 10,096,117 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation

Abstract

A method for classifying a visual element in a video frame as either foreground or background, the video frame being characterised using a scene model comprising a plurality of modes, the method comprising the steps of: determining a blob boundary characteristic for a blob comprising at least the visual element; identifying a mode matched to the visual element; classifying the visual element and the matched mode as foreground dependent upon a match mode boundary characteristic of the matched mode; and updating the scene model dependent upon the blob boundary characteristic and the matched mode boundary characteristic.

Claims

1. A method for classifying a visual element as either foreground or background in a video frame method comprising the steps of:

computing a blob boundary characteristic for a blob, wherein said blob comprises at least one visual element corresponding to the video frame and the scene model and said visual element are labelled foreground based on at least a match mode boundary characteristic in the scene model;

updating the scene model based on at least the computed blob boundary characteristic and the match mode boundary characteristic; and

classifying the visual element based on the scene model.

2. A method for classifying a visual element in a video frame as either foreground or background, the video frame being characterised using a scene model comprising a plurality of modes, the method comprising the steps of:

determining a blob boundary characteristic (b_bc) for a blob comprising at least the visual element;

identifying a mode matched to the visual element;

classifying the visual element as foreground dependent upon a match mode boundary characteristic (m_bc) of the matched mode; and

updating the scene model dependent upon the blob boundary characteristic (b_bc) and the matched mode boundary characteristic (m_bc).

3. A method according to claim 2, wherein determining the blob boundary characteristic comprises determining a gradient across a boundary between the visual element of the blob and a neighbouring visual element.

4. A method according to claim 2, wherein identifying the mode matched to the visual element comprises the steps of:

comparing visual descriptors of modes in the scene model against a visual feature of the visual element;

determining a match score for each comparison; and

identifying the mode matched to the visual element as the mode having the highest match score.

5. A method according to claim 2, wherein classifying the visual element as foreground 5 comprises the steps of:

determining a probability that the visual element is foreground dependent upon a value of the match mode boundary characteristic over time; and

classifying the visual element as foreground dependent upon the probability and a threshold.

6. An apparatus for classifying a visual element in a video frame as either foreground or background, the video frame being characterised using a scene model comprising a plurality of modes, the apparatus comprising:

a camera module for capturing the video frame;

a processor; and

a memory storing a computer executable software program for directing the processor to perform a method comprising the steps of;

determining a blob boundary characteristic for a blob comprising at least the visual element;

identifying a mode matched to the visual element;

classifying the visual element and the matched mode as foreground dependent upon a match mode boundary characteristic of the matched mode; and

updating the scene model dependent upon the blob boundary characteristic and the matched mode boundary characteristic.

7. A non-transitory computer readable memory medium storing a computer executable software program for directing a processor to perform a method for classifying a visual element in a video frame as either foreground or background, the video frame being characterised using a scene model comprising a plurality of modes, the method comprising the steps of:

determining a blob boundary characteristic for a blob comprising at least the visual element;

identifying a mode matched to the visual element;

classifying the visual element and the matched mode as foreground dependent upon a match mode boundary characteristic of the matched mode; and

updating the scene model dependent upon the blob boundary characteristic and the matched mode boundary characteristic.