Automatic pose estimation from uncalibrated unordered spherical panoramas

A computer implemented system, comprising:

• a panoramic image storage module configured to store a plurality of panoramic images, which are uncalibrated and unordered, wherein at least one of the panoramic images includes an interior building image;
• one or more processors configured to:
• extract, for each panoramic image, one or more feature points;
• and
• generate a match matrix for all the panoramic images based on the one or more feature points;
• construct a minimal spanning tree based on the match matrix;
• identify, from the minimal spanning tree, one or more image triplets, wherein each image triplet comprises a panoramic image associated with both a parent panoramic image and a child panoramic image;
• identify a first panoramic image, based on the minimal spanning tree;
• and
• identify a second panoramic image, based on the minimal spanning tree, wherein the second panoramic image is associated with the first panoramic image providing a navigation from the first panoramic image to the second panoramic image.

A computer program product comprising at least one non-transitory computer readable storage medium encoding instructions thereon that, in response to execution by a computing device, cause the computing device to perform operations comprising:

• receiving a plurality of panoramic images, which are uncalibrated and unordered, wherein at least one of the panoramic images includes an interior building image;
• extracting, for each panoramic image, one or more feature points;
• generating a match matrix for all the panoramic images based on the one or more feature points;
• constructing a minimal spanning tree based on the match matrix;
• identifying a first panoramic image, based on the minimal spanning tree;
• identifying a second panoramic image, based on the minimal spanning tree, wherein the second image is associated with the first panoramic image providing a navigation from the first panoramic image to the second panoramic image;
• identifying, from the minimal spanning tree, one or more image triplets wherein each image triplet comprises a panoramic image associated with both a parent panoramic image and a child panoramic image;
• estimating a relative pose of each image triplet;
• and
• estimating a relative scale of a triplet that contains the first panoramic image;
• and propagating the relative scale of the triplet containing the first panoramic image to another triplet that does not contain the first panoramic image, wherein the first panoramic image represents a root of the minimal spanning tree.