Method and device for detecting falls by image analysis

• US 20150023551A1
• Filed: 01/17/2014
• Published: 01/22/2015
• Est. Priority Date: 01/17/2013
• Status: Active Grant

• Current Assignee: Creative Specific Software
• Original Assignee: Cash Systems Security
• Inventors: RAMZI, Larbi
• Granted Patent: US 9,202,283 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments
• CPC Class Codes: G06V 40/23 Recognition of whole body movements, e.g. for sport training

Abstract

The present invention relates to a method for detecting a fall of a person by analysis of a stream of video images originating from an image capture device, comprising:

• • acquisition (21) of successive images of a determined scene;
  • selection (23, 25, 27) of at least one point of the scene in said images;
  • determination of the movement (29) of each point selected by analysis of the displacement of said point in the successive images in the form of a temporal succession of vectors oriented proportionally to the instantaneous velocity of the movement of said point;
  • computation (31) of the instantaneous acceleration of each point associated with each vector representing the instantaneous velocity;
  • detection of a fall (33) when the determined instantaneous velocity is above a predefined threshold velocity and the instantaneous acceleration is above a predefined threshold acceleration.

Claims

1. Method for detecting a fall of a person by analysis of a stream of video images originating from an image capture device, comprising:

acquisition of successive images of a determined scene;

selection of at least one point of the scene in said images;

determination of the movement of each point selected by analysis of the displacement of said point in the successive images in the form of a temporal succession of vectors oriented proportionally to the instantaneous velocity of the movement of said point;

computation of the instantaneous acceleration of each point associated with each vector representing the instantaneous velocity;

detection of a fall when the determined instantaneous velocity is above a predefined threshold velocity and the instantaneous acceleration is above a predefined threshold acceleration.

2. Method according to claim 1, wherein a capture of the last acquired image of the scene is recorded, and in that a fall is detected only if the recorded image capture or a copy of said image capture is identified as a fall scene by shape recognition software.

3. Method according to claim 1, wherein the fall is detected when, moreover, the orientation of the instantaneous velocity vector is included in a predefined directional sector.

4. Method according to claim 1, wherein several regions of interest are defined inside the scene so that the fall detection is performed only for movements inside these regions of interest;

5. Method according to claim 1, wherein a fall is detected only if the fall detection is performed for a number of points above a predetermined value.

6. Method according to claim 1, wherein an alarm is triggered when, moreover, no movement is detected during a determined period.

7. Method according to claim 1, wherein as the person is wearing a fall detection bracelet, an alarm is triggered when a fall detected by analysis of the video stream is confirmed by the bracelet.

8. Method according to claim 6, wherein, apart from the alarm, a message accompanied by a photo of the scene is sent.

9. Method according to claim 1, wherein, with the fall detection taking place in real time, the velocity vector and the acceleration of each point are computed for each new image taking as parameter the variation of the position of the point in the new image and the preceding images.

10. Computer program product comprising program instructions suitable for implementing the method according to claim 1 when the program product is executed on a computer.

11. System for detecting the fall of a person by analysis of a stream of video images, comprising:

a device for acquiring successive images of a determined scene;

a means for selecting at least one point of the scene in said images;

a computer of the movement of each point selected by analysis of the displacement of said point in the successive images in the form of a temporal succession of vectors oriented proportionally to the instantaneous velocity of the movement of said point;

a computer of the instantaneous acceleration of each point associated with each vector representing the instantaneous velocity;

a means for detecting a fall when the determined instantaneous velocity is above a predefined threshold velocity and the instantaneous acceleration is above a predefined threshold acceleration.