DYNAMIC PARTICLE FILTER PARAMETERIZATION

• US 20170178347A1
• Filed: 12/16/2016
• Published: 06/22/2017
• Est. Priority Date: 12/17/2015
• Status: Active Grant

• Current Assignee: Stmicroelectronics France (STMicroelectronics NV)
• Original Assignee: STMicroelectronics SA (STMicroelectronics NV)
• Inventors: ALIBAY, Manu, AUBERGER, Stphane
• Granted Patent: US 10,229,508 B2
• Time in Patent Office: Days
• CPC Class Codes: G06T 2207/10004 Still image; Photographic image
• CPC Class Codes: G06T 2207/10028 Range image; Depth image; 3D point clouds
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06T 2207/30244 Camera pose
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters

Abstract

Method of estimating a position variation of a motion of an apparatus between a first instant and a second instant, said motion including a rotation of the apparatus and said position variation, said position variation including a position and a velocity, wherein estimating said position variation comprises performing a particles filtering for estimating said position and velocity from the probabilistic-weighted average of the particles, said particles filter using a known estimation of said rotation and being parameterized for taking into account a quality of said rotation estimation.

Claims

1. A method, comprising:

estimating a position variation of a motion of an apparatus between a first instant and a second instant, said motion including a rotation of the apparatus and said position variation, said position variation including a position and a velocity, wherein estimating said position variation comprises performing a particles filtering for estimating said position and velocity from the probabilistic-weighted average of the particles, said particle filter using a known estimation of said rotation and being parameterized for taking into account a quality of said rotation estimation.

2. The method according to claim 1, wherein said parameterization is used in the velocity estimation.

3. The method according to claim 1, wherein said apparatus is intended to capture a sequence of images, a first image being captured at said first instant, a second image being captured at said second instant, said motion is an ego-motion of the apparatus between the first image and the second image including a 3D rotation of the apparatus and a position variation of the apparatus in the 3D space, and each particle is a virtual apparatus.

4. The method according to claim 3, wherein said 3D rotation is estimated by performing a RANSAC type algorithm providing an inliers rate, and said parameterization of the particle filter taking into account a quality of said rotation estimation includes said inliers rate.

5. The method according to claim 4, wherein said parameterization of the particle filter comprises the calculation of a first parameter being a relationship between said inliers rate, the reprojection error of the best particle, and targeted reprojection error.

6. The method according to claim 2, wherein estimating said velocity further comprises guiding the particle spreading with a pedestrian step detection technique using accelerometers data.

7. The method according to claim 6, wherein the velocity estimation takes into account said first parameter and a random 3D vector generated according to a Gaussian distribution involving an expected velocity for the second image, and said expected velocity for the second image is calculated by taking into account a current estimated rotation matrix, a velocity of step and the frame rate of the apparatus.

8. A device, comprising processing means for estimating a position variation of a motion of an apparatus between a first instant and a second instant, said motion including a rotation of the apparatus and said position variation, said position variation including a position and a velocity, wherein said processing means comprises particle filter means for particle filtering of particles for estimating said position and velocity from a probabilistic-weighted average of the particles, said particle filtering using a known estimation of said rotation and being parameterized for taking into account a quality of said rotation estimation.

9. The device according to claim 8, wherein said particle filter means are configured to use said parameterization in the velocity estimation.

10. The device according to claim 8, wherein said apparatus is configured to capture a sequence of images, including a first image captured at said first instant, and a second image captured at said second instant, said motion is an ego-motion of the apparatus between the first image and the second image including a 3D rotation of the apparatus and a position variation of the apparatus in the 3D space, and each particle is a virtual apparatus.

11. The device according to claim 10, further comprising a processing module configured to estimate said 3D rotation by performing a RANSAC type algorithm providing an inliers rate, and said parameterization of the particle filter taking into account a quality of said rotation estimation includes said inliers rate.

12. The device according to claim 11, wherein said parameterization of the particle filter comprises a first parameter being a relationship between said inliers rate, the reprojection error of the best particle, and targeted reprojection error.

13. The device according to claim 9, wherein said processing means comprises a processing module configured to estimate said velocity and comprising a guiding module configured to guide the particle spreading with a pedestrian step detection technique using accelerometer data.

14. The device according to claim 13, wherein said processing module is configured to take into account said first parameter and a random 3D vector generated according to a Gaussian distribution involving an expected velocity for the second image, and comprises calculation means for calculating said expected velocity for the second image by taking into account a current estimated rotation matrix, a velocity of step and the frame rate of the apparatus.

15. A platform, comprising

an apparatus;

and

a device including processing means for estimating a position variation of a motion of an apparatus between a first instant and a second instant, said motion including a rotation of the apparatus and said position variation, said position variation including a position and a velocity, wherein said processing means comprises particle filter means for particle filtering of particles for estimating said position and velocity from a probabilistic-weighted average of the particles, said particle filtering using a known estimation of said rotation and being parameterized for taking into account a quality of said rotation estimation.

16. The platform according to claim 15, wherein said apparatus is configured to capture a sequence of images, including a first image captured at said first instant, and a second image captured at said second instant, said motion is an ego-motion of the apparatus between the first image and the second image including a 3D rotation of the apparatus and a position variation of the apparatus in the 3D space, and each particle is a virtual apparatus.

17. The platform according to claim 16, wherein the processing means includes a processing module configured to estimate said 3D rotation by performing a RANSAC type algorithm providing an inliers rate, and said parameterization of the particle filter taking into account a quality of said rotation estimation includes said inliers rate.

19. The platform according to claim 15, wherein said processing means comprises a processing module configured to estimate said velocity and comprising a guiding module configured to guide the particle spreading with a pedestrian step detection technique using accelerometer data.

20. The platform according to claim 19, wherein said processing module is configured to take into account said first parameter and a random 3D vector generated according to a Gaussian distribution involving an expected velocity for the second image, and comprises calculation means for calculating said expected velocity for the second image by taking into account a current estimated rotation matrix, a velocity of step and the frame rate of the apparatus.