METHOD FOR DETECTING TARGETS ON THE GROUND AND IN MOTION, IN A VIDEO STREAM ACQUIRED WITH AN AIRBORNE CAMERA

• US 20170213078A1
• Filed: 01/13/2017
• Published: 07/27/2017
• Est. Priority Date: 01/26/2016
• Status: Active Grant

• Current Assignee: Thales SA
• Original Assignee: Thales SA
• Inventors: Guerrini, Gilles, Camus, Fabien, Richard, Fabien
• Granted Patent: US 10,387,718 B2
• Time in Patent Office: Days
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/30212 Military
• CPC Class Codes: G06T 2207/30252 Vehicle exterior; Vicinity of vehicle
• CPC Class Codes: G06T 5/80 Geometric correction
• CPC Class Codes: G06T 7/254 involving subtraction of images
• CPC Class Codes: G06V 20/48 Matching video sequences
• CPC Class Codes: G06V 40/20 Movements or behaviour, e.g. gesture recognition recognition of facial expressions G06V40/16

Abstract

A target detection method for detecting targets on the ground and in motion, in a video stream that is acquired by an airborne digital camera, which includes the steps of: processing a plurality of successive frames in the video frame in a manner so as to stabilize the frames as if they had been acquired by a fixed camera; and comparing two processed frames, temporally separated from the other, so as to identify the zones of pixels in motion from one processed frame to the other, one zone of pixels in motion constituting a detected target.

Claims

1. A method for detecting targets on the ground and in motion, within a video stream that is acquired by an airborne digital camera, including the steps of:

processing a plurality of successive frames of the video stream in a manner so as to stabilize the frames as if they had been acquired by a fixed camera and,

comparing two processed frames, temporally separated one from the other, in a manner so as to identify a pixel zone in motion from one processed frame to the other, the pixel zone in motion constituting a detected target.

2. The method according to claim 1, wherein the step of processing a plurality of successive frames includes an error correction step of correcting a relative parallax error affecting each frame of the plurality of successive frames in a manner so as to obtain, for each frame, a corrected frame, the error correlation step implementing an algorithm for determining a projective mapping that makes it possible to pass from a frame at a current time to a corrected frame associated with a frame at a preceding time.

3. The method according to claim 2, wherein the step of processing a plurality of successive frames includes a motion vector determination step of determining a motion vector of the camera relative to the ground, the motion vector determination step implementing an algorithm for determining an optical flow that makes it possible to pass from a corrected frame at a current time to a corrected frame at a past time that is separated from the current time by an integer number of sampling time intervals.

4. The method according to claim 3, wherein the algorithm for determining an optical flow uses a homogenous grid of points.

5. The method according to claim 3, wherein the integer number is selected to be between 5 and 15, preferably equal to 10.

6. The method according to claim 3, wherein, once the motion vector of the camera relative to the ground has been determined, a corresponding transformation is applied to the corrected frame at the current time in order to compensate for an effect of a movement of the camera relative to the ground and obtain a final frame that may be superimposed on the corrected at a past time, considered as being an initial frame.

7. The method according to claim 6, wherein, the step of comparing two processed frames consists of comparing the final frame and the initial frame to each other by carrying out successively the following sub-steps:

subtracting pixel by pixel the initial frame from the final frame in a manner so as to obtain, for each pixel, a color distance value;

applying a color distance threshold, each pixel having a color distance value that is less than or equal to the color distance threshold, taking the null value, and each pixel having a color distance value that is greater than the color distance threshold taking the unit value, so as to obtain a binary map; and

applying a contour determination algorithm on the binary map in order to group together pixels having a unit value into at least one pixel zone in motion.

8. The method according to claim 7, wherein, the binary map obtained at the end of the sub-step of applying a color distance threshold being an intermediate binary map, the following sub-steps are carried out successively on the intermediary binary map to obtain a binary map:

applying a morphological erosion mapping using an appropriate mask; and

applying a morphological dilatation mapping that uses the mask of the morphological erosion mapping.

9. The method according to claim 7, including a supplementary step of verifying a consistency of one pixel zone in motion identified at the end of the step of comparing two processed frames, consisting of determining a correlation index between values of the pixels of the two processed frames, the pixel zone in motion being deemed to correspond to a target when the correlation index is close to the value −

1.

10. The method according to claim 1, wherein the pixel zone in motion identified is displayed in an appropriate manner over the video stream displayed on a control monitor.

11. The method according to claim 1, wherein the video stream is a video stream in a visible spectral domain or an infra-red domain.

12. The method according to claim 1, executed in real time or in constrained time on the video stream.

13. The method according to claim 1, making possible to detect a target of a size, of a few pixels.

14. A data recording medium for recording information that contains computer program instructions suitable for being executed by a computer in order to implement the method according to claim 1.