Object tracking device, object tracking method, and object tracking program

• US 8,891,821 B2
• Filed: 01/13/2011
• Issued: 11/18/2014
• Est. Priority Date: 02/19/2010
• Status: Expired due to Fees

• Current Assignee: Sovereign Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: Panasonic Corporation (Panasonic Holdings Corporation)
• Inventors: Sato, Yuji
• Primary Examiner(s): Akhavannik, Hadi
• Assistant Examiner(s): RASHIDIAN, MOHAMMAD M
• Application Number:
  US13/263,169
• Publication Number:
  US 20120093364A1
• Time in Patent Office: 1,405 Days
• Field of Search: 382/100, 382103-104, 345/169, 348/135, 702/181, 702/150, 703/6
• US Class Current: 382/103
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters

Abstract

An object tracking apparatus is provided that enables the possibility of erroneous tracking to be further reduced. An object tracking apparatus (300) is an apparatus that tracks the position of an object displayed in video using a particle filter, and has: a feature calculation section (340) that generates a plurality of particles indicating candidates for the position of the object, and calculates a feature of an image of the object and features of images of the particles; a likelihood calculation section (350) that calculates, for each particle, the likelihood of that particle being the position of the object from similarity between a feature of an image of that particle and a feature of an image of the object; a position estimation section (360) that estimates the position of the object based on the calculated particle likelihood; and a likelihood correction section (390) that performs likelihood correction when there are a plurality of objects and a plurality of positions estimated in correspondence to these overlap.

Claims

1. An object tracking apparatus that tracks a position of an object displayed in video using a particle filter, the position of the object being defined by an area corresponding to an image area of the object in the video and the object tracking apparatus comprising:

a feature calculation section that generates a plurality of particles indicating candidates for a position of the object, and calculates a feature of an image of the object and features of images of the particles;

a likelihood calculation section that calculates, for each particle, a likelihood of that particle being a position of the object from similarity between a feature of an image of that particle and a feature of an image of the object;

a position estimation section that estimates a position of the object based on the calculated particle likelihood;

an overlap proportion calculation section that calculates a size of an overlap area of a particle of a second object with respect to a position of a first object; and

a likelihood correction section that, when there is a particle for which a size of the overlap area is greater than or equal to a first predetermined value and the likelihood is greater than or equal to a second predetermined value, corrects a likelihood of a corresponding particle.

2. The object tracking apparatus according to claim 1, wherein a position of the first object is defined by a rectangular area that circumscribes an image area of the first object in the video, and a size of the overlap area is a proportion of an area of an overlap area with respect to an area of a particle of the second object.

3. The object tracking apparatus according to claim 1, wherein the similarity is similarity between a color histogram of an image of the object and a color histogram of an image of the particle.

4. The object tracking apparatus according to claim 1, wherein the second predetermined value is an average value of particle likelihood of the second object.

5. The object tracking apparatus according to claim 1, wherein the second predetermined value is a mode of a particle likelihood histogram of the second object.

6. The object tracking apparatus according to claim 1, wherein the likelihood correction section corrects the likelihood by a size according to a difference of a size of the overlap area and a predetermined value of the likelihood.

7. An object tracking method that tracks a position of an object displayed in video using a particle filter, the position of the object being defined by an area corresponding to an image area of the object in the video and the object tracking method comprising the steps of:

generating a plurality of particles indicating candidates for a position of the object, and calculating a feature of an image of the object and features of images of the particles;

calculating, for each particle, a likelihood of that particle being a position of the object from similarity between a feature of an image of that particle and a feature of an image of the object;

estimating a position of the object based on calculated the particle likelihood;

calculating a size of an overlap area of a particle of a second object with respect to a position of a first object; and

correcting, when there is a particle for which a size of the overlap area is greater than or equal to a first predetermined value and the likelihood is greater than or equal to a second predetermined value, a likelihood of a corresponding particle.

8. An object tracking program stored on a non-transitory computer medium for tracking a position of an object displayed in video using a particle filter, the position of the object being defined by an area corresponding to an image area of the object in the video and the object tracking program causing a computer to execute:

processing that generates a plurality of particles indicating candidates for a position of the object, and calculates a feature of an image of the object and features of images of the particles;

processing that calculates, for each particle, a likelihood of that particle being a position of the object from similarity between a feature of an image of that particle and a feature of an image of the object;

processing that estimates a position of the object based on calculated the particle likelihood;

processing that calculates a size of an overlap area of a particle of a second object with respect to a position of a first object; and

processing that corrects, when there is a particle for which a size of the overlap area is greater than or equal to a first predetermined value and the likelihood is greater than or equal to a second predetermined value a likelihood of a corresponding particle.