US 10,706,587 B1Granted

Calibration of multiple cameras

AssigneeAmazon Technologies, Inc.

InventorsLorenzo Sorgi, Pramod Sreenivasa

Life of this patent

Estimating the term…

Loading patent…

Abstract

A plurality of cameras obtain images in an environment. Calibration data about the translation and rotation of the cameras is used to support functionality such as determining a location of an object that appears in those images. To determine the calibration data, images are processed to determine features produced by points in the environment. A cluster is designated that includes the images of the same point as viewed from at least some of the cameras. A triangulation matrix and back-projection error are calculated for each cluster. These triangulation matrices and back-projection errors are then perturbed to find the translation and rotation that minimize eigenvalues and eigenvectors representative of the feature. The eigenvalues and eigenvectors are then used to determine the rotation and translation of the respective cameras with respect to an origin.

Claims20

Dependent claims

A system comprising:

a plurality of cameras in a material handling facility;
a memory storing first computer-executable instructions;
and
a hardware processor to execute the first computer-executable instructions to:
receive a plurality of images acquired by the plurality of cameras;
determine a plurality of features that are present the plurality of images, wherein each feature is representative of a three-dimensional (3D) point in the material handling facility as depicted in one of the plurality of images;
determine, based on the plurality of features, a plurality of clusters wherein each cluster is a group of images from the plurality of images that depicts the same feature;
determine a subset of the plurality of cameras that are associated with the each of the cluster;
determine, for the each cluster, a triangulation matrix representative of a first translation and first rotation of each camera in the subset of the plurality of cameras associated with the each cluster;
determine, by adjusting the first translation and the first rotation of the each camera, a second rotation and second translation for each of the plurality of cameras, wherein the second rotation and the second translation minimize a sum of:
a minimum of a triangulation cost using the triangulation matrix for the each cluster;
and
a minimum of a back-projection error for each cluster.

10 more elements

A system comprising:

a plurality of cameras in an environment;
a memory storing first computer-executable instructions;
and
a hardware processor to execute the first computer-executable instructions to:
acquire a plurality of images, wherein each image is acquired by one of the plurality of cameras;
determine a plurality of features that are present in each of at least a portion of the plurality of images, wherein at least a portion of the features are representative of three-dimensional (3D) points in the environment;
determine, based on the plurality of features, a plurality of clusters wherein each cluster comprises a subset of the plurality of images that are associated with a particular 3D point as indicated by at least a portion of the plurality of features;
determine a subset of the plurality of cameras that are associated with the each cluster;
determine, for the each cluster, a triangulation matrix representative of a first translation and first rotation of each camera in the subset of the plurality of cameras associated with the each cluster;
and
determine calibration data indicative of a second rotation and second translation for each of the plurality of cameras.

8 more elements

Dependencies are inferred from the claim text.