# US 2009/0213222 A1 · System for tracking a moving object, by using particle filtering

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2009/0213222 A1
- Document ID: US20090213222A1
- Status: Published application
- Kind: A1
- Application number: 12/388,967
- Filed: 2009-02-19
- Published: 2009-08-27
- Earliest claimed: 2008-02-21
- Assignees: Kabushiki Kaisha Toshiba
- Inventors: Kenji BABA; Takaaki Enohara; Yoshihiko Suzuki; Yusuke Takahashi
- Web page: https://patentgrep.com/patent/US20090213222A1

## CPC

- G06T 7/277: involving stochastic approaches, e.g. using Kalman filters
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

In a tracking system for tracking a moving object by using a particle filter, the particle filter is configured to arrange particles initially, in a standby state, in a given background region provided in the screen of a camera and to rearrange the particles with respect to the moving object in accordance with a change in likelihood that the object has with respect to the particles.

## Claims

### Claim 1

A system for tracking a moving object, by using particle filtering, the system comprising:
a camera configured to photograph the object;
an initial arrangement module configured to arrange particles, in a standby-state, in a background included in an image photographed by the camera; and
a rearrangement module configured to rearrange the particles with respect to the object included in the image, in accordance with a change in likelihood of the particles arranged in the standby state.

### Claim 2 (depends on 1)

The system according to claim 1, wherein the rearrangement module rearranges the particles with respect to the object when the likelihood of the particles initially arranged becomes low in comparison with the likelihood of the object existing in the image photographed by the camera.

### Claim 3 (depends on 1)

The system according to claim 1, further comprising a control module configured to cause the initial arrangement module to arrange, in a standby mode, the particles in the standby state in a given background region; to detect that an operating mode of the system changes from the standby mode to a tracking mode for performing a tracking process of the object, when the object appears in the image photographed by the camera and the likelihood of the particles initially arranged in the standby mode decreases; to cause the rearrangement module to rearrange the particles with respect to the object when the operating mode of the system is detected to change from the standby mode to the tracking mode, thereby detecting completion of the initial arrangement of particles, and to change the operating mode of the system from the standby mode to the tracking mode after the completion of the initial arrangement of particles.

### Claim 4 (depends on 3)

The system according to claim 3, wherein after the operating mode has been changed to the tracking mode, the control module changes the operating mode from the tracking mode to the standby node when the object disappears from the image photographed by the camera.

### Claim 5 (depends on 1)

The system according to claim 1, further comprising:
a module configured to cause the particles to follow the object, thereby performing a tracking process after the initial arrangement module has initially arranged the particles; and
a module configured to arrange particles preliminarily near the particles following the object in the tracking process, in order to cancel occlusion occurring after the rearrangement module initially arranged the particles.

### Claim 6 (depends on 3)

The system according to claim 3, further comprising:
a module configured to cause the particles to follow the object, thereby performing a tracking process after the initial arrangement module has initially arranged the particles; and
a module configured to arrange particles preliminarily near the particles following the object in the tracking process, in order to cancel occlusion occurring after the rearrangement module initially arranged the particles.

### Claim 7 (depends on 5)

The system according to claim 5, further comprising a module configured to cause, in the tracking process, the preliminarily arranged particles to follow a new object that appears when the occlusion is cancelled.

### Claim 8 (depends on 3)

The system according to claim 3, further comprising a likelihood inference module configured to perform a dynamic threshold value control based on a likelihood distribution inferred of particle groups, thereby to determine which particles should be rearranged, in the tracking process, in a complicated background or a background similar in characteristic to the object.

### Claim 9 (depends on 1)

The system according to claim 1, in which the rearrangement module is configured to rearrange particles in accordance with a plurality of system models, and which further comprises a module configured to select the system model having more high-likelihood particles than any other system model and to perform a tracking process that uses the system model selected.

### Claim 10 (depends on 1)

The system according to claim 1, wherein the rearrangement module is configured to rearrange particles in accordance with a plurality of system models, and which further comprises a module configured to select a high-likelihood particle from the particles arranged in a prediction process based on each system model.

### Claim 11 (depends on 1)

The system according to claim 1, further comprising:
a module configured to cause the particles to follow the object, thereby performing a tracking process after the initial arrangement module has initially arranged the particles; and
a module configured to hold attribute data representing the distribution and likelihood that particle groups at the object have during the tracking process.

### Claim 12 (depends on 3)

The system according to claim 3, further comprising:
a module configured to cause the particles to follow the object, thereby performing a tracking process after the initial arrangement module has initially arranged the particles; and
a module configured to hold attribute data representing the distribution and likelihood that particle groups at the object have during the tracking process.

### Claim 13 (depends on 11)

The system according to claim 11, further comprising a decision module configured to determine whether the object appears again in the image after disappearing from the image, from the attribute data in the tracking process.

### Claim 14 (depends on 1)

The system according to claim 1, which further comprises a module configured to learn the position and distribution of particles at the object beforehand, thereby generating data, and in which the rearrangement module rearranges the particles with respect to the object in accordance with the data generated through learning.

### Claim 15 (depends on 3)

The system according to claim 3, which further comprises a module configured to learn the position and distribution of particles at the object beforehand, thereby generating data, and in which the rearrangement module rearranges the particles with respect to the object in accordance with the data generated through learning.

### Claim 16 (depends on 7)

The system according to claim 7, further comprising:
a module configured to learn the position and distribution of particles at a specific object beforehand, thereby generating data through learning; and
a module configured to cause particles to follow a new object appearing upon cancellation of the occlusion, in the tracking process in accordance with the data generated through learning.

### Claim 17 (depends on 16)

The system according to claim 16, further comprising a nodule configured to delete arranged with respect to the new object when the new object is determined not identical to the specified object, from the data generated through learning, after the particles have been caused to follow the new object appearing upon cancellation of the occlusion.

### Claim 18 (depends on 1)

The system according to claim 1, which further comprises a likelihood inference module which is included in the module configured to perform a tracking process to make the particles follow the moving object, after the initial arrangement module has initially arranged the particles, and which is configured to define a characteristic value of the object as a likelihood function,
wherein the likelihood inference module performs likelihood inference processes that use various likelihood functions, respectively.
