Person tracking method, person tracking apparatus, and person tracking program storage medium

• US 8,374,392 B2
• Filed: 04/26/2010
• Issued: 02/12/2013
• Est. Priority Date: 05/21/2009
• Status: Active Grant

• Current Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Original Assignee: Fujifilm Corporation (Fujifilm Holdings Corporation)
• Inventors: Hu, Xuebin
• Primary Examiner(s): Park, Chan S
• Assistant Examiner(s): NAKHJAVAN, SHERVIN K
• Application Number:
  US12/767,452
• Publication Number:
  US 20100296702A1
• Time in Patent Office: 1,023 Days
• Field of Search: 382/103, 382/107, 382/115, 382/116, 382/118, 382/181, 382/190, 382/192, 382/203, 348/169
• US Class Current: 382/103
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/248 involving reference images or patches
• CPC Class Codes: G06V 10/62 relating to a temporal dimension, e.g. time-based feature extraction; Pattern tracking
• CPC Class Codes: G06V 30/19147 Obtaining sets of training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 30/2504 Coarse or fine approaches, e.g. resolution of ambiguities or multiscale approaches
• CPC Class Codes: G06V 40/171 Local features and components; Facial parts eye characteristics G06V40/18; Occluding parts, e.g. glasses; Geometrical relationships

Abstract

A person tracking method capable of obtaining information representing a correspondence between a shot image and a three-dimensional real space, without actual measurement, thereby enabling lighter processing is provided. The method includes: calculating a statistically average correspondence between a size of person'"'"'s head and a position representing a height of the head on the shot image, the camera looking down a measured space and taking the measured space; detecting a position and a size of a head on each of measured frame images; calculating, based on positions and sizes of heads on plural past measured frame images and the correspondence, a movement feature quantity representing a possibility that a head on a current measured frame image is of the same person on the past measured frame images; and determining that the head on the current measured frame image is of the same person on the past measured frame images.

Claims

1. A person tracking method comprising:

a correspondence calculation step of calculating a statistically average correspondence between a size of a head of a person appearing on an image taken by a camera and a position representing a height of the head on the taken image, the camera being disposed at a position looking down a measured space and generates image data representing a moving image of a predetermined frame rate by taking the moving image of the measured space;

a head detecting step of detecting, based on each piece of image data representing each of a plurality of measured frame images among a plurality of frame images obtained through image-taking by the camera, a position and a size of a head on each of the measured frame images;

a movement-feature-quantity calculation step of calculating, based on positions and sizes of heads on a plurality of past measured frame images detected in the head detecting step and the correspondence calculated in the correspondence calculation step, a movement feature quantity that represents a possibility that a head on a current measured frame image detected in the head detecting step is a head that belongs to the same person as a person having the heads on the past measured frame images; and

a same-person determination step of determining, by using the movement feature quantity calculated in the movement-feature-quantity calculation step, whether the head on the current measured frame image is the head that belongs to the same person as the person having the heads on the past measured frame images.

2. The person tracking method according to claim 1, wherein the movement-feature-quantity calculation step is a step of calculating the movement feature quantity based on a ratio between a uniform movement distance and a between-head distance, the uniform movement distance being a distance between an intersection point and a first head position, the intersection point being formed by a border line and a straight line, when the border line of a movement range of the head that belongs to the same person as the person of the heads on the past measured frame images is set on the current measured frame image based on the positions and sizes of the heads on the plurality of past measured frame images detected in the head detecting step as well as based on the correspondence calculated in the correspondence calculation step, the straight line passing through the first head position that is a position of a head on an immediately preceding past measured frame image and a second head position that is the position of the head on the current measured frame image, and the between-head distance being a distance between the first head position and the second head position.

3. A person tracking apparatus comprising:

a storage section that stores a statistically average correspondence between a size of a head of a person appearing on an image taken by a camera and a position representing a height of the head on the taken image, the camera being disposed at a position looking down a measured space and generates image data representing a moving image of a predetermined frame rate by taking the moving image of the measured space;

a head detecting section that detects, based on each piece of image data representing each of a plurality of measured frame images among a plurality of frame images obtained through image-taking by the camera, a position and a size of a head on each of the measured frame images;

a movement-feature-quantity calculation section that calculates, based on positions and sizes of heads on a plurality of past measured frame images detected by the head detecting section and the correspondence stored in the storage section, a movement feature quantity that represents a possibility that a head on a current measured frame image detected by the head detecting section is a head that belongs to the same person as a person having the heads on the past measured frame images; and

a same-person determination section that determines, by using the movement feature quantity calculated by the movement feature-quantity calculation section, whether the head on the current measured frame image is the head that belongs to the same person as the person having the heads on the past measured frame images.

4. The person tracking apparatus according to claim 3, wherein the movement-feature-quantity calculation section calculates the movement feature quantity based on a ratio between a uniform movement distance and a between-head distance, the uniform movement distance being a distance between an intersection point and a first head position, the intersection point being formed by a border line and a straight line, when the border line of a movement range of the head that belongs to the same person as the person of the heads on the past measured frame images is set on the current measured frame image based on the positions and sizes of the heads on the plurality of past measured frame images detected in the head detecting section as well as based on the correspondence calculated in the correspondence calculation section, the straight line passing through the first head position that is a position of a head on an immediately preceding past measured frame image and a second head position that is the position of the head on the current measured frame image, and the between-head distance being a distance between the first head position and the second head position.

5. A non-transitory computer-readable storage medium storing a program that is executed in an operational apparatus and causes the operational apparatus to perform a person tracking method comprising:

storing a statistically average correspondence between a size of a head of a person appearing on an image taken by a camera and a position representing a height of the head on the taken image, the camera being disposed at a position looking down a measured space and generates image data representing a moving image of a predetermined frame rate by taking the moving image of the measured space;

detecting, based on each piece of image data representing each of a plurality of measured frame images among a plurality of frame images obtained through image-taking by the camera, a position and a size of a head on each of the measured frame images;

a calculating, based on positions and sizes of detected heads on a plurality of past measured frame images and the correspondence stored in the step of storing, a movement feature quantity that represents a possibility that a detected head on a current measured frame image is a head that belongs to the same person as a person having the heads on the past measured frame images; and

a determining, by using the calculated movement feature quantity whether the head on the current measured frame image is the head that belongs to the same person as the person having the heads on the past measured frame images.

6. The person tracking program storage medium according to claim 5 wherein the step of calculating movement-feature-quantity calculates the movement feature quantity based on a ratio between a uniform movement distance and a between-head distance, the uniform movement distance being a distance between an intersection point and a first head position, the intersection point being formed by a border line and a straight line, when the border line of a movement range of the head that belongs to the same person as the person of the heads on the past measured frame images is set on the current measured frame image based on the positions and sizes of the detected heads on the plurality of past measured frame images as well as based on the calculating correspondence, the straight line passing through the first head position that is a position of a head on an immediately preceding past measured frame image and a second head position that is the position of the head on the current measured frame image, and the between-head distance being a distance between the first head position and the second head position.