Component based correspondence matching for reconstructing cables

摘要

In a stereoscopic pair of images, global homography at the image level is applied to feature points extracted from connected components (CC) to identify corresponding CC's and feature points, and to discard any CC's that do not have a corresponding pair in the stereoscopic pair of images. Local homography at the CC level is then applied to individual footprint areas of the previously identified paired CC to further clean feature point correspondence. Any CC or feature point or pixel within a paired CC footprint not satisfying local homography constraint is discarded. A correspondence is also extrapolated between unknown pixels within a paired CC footprint using a weighing mechanism and the unknown pixel's surrounding pixels that do have a known correspondence. This provides a dense correspondence of pixels, or feature points, which is then used to create a dense 3D point cloud of identified objects within a 3D space.

主权项

1. A method of identifying corresponding objects in first and second images of a stereoscopic pair of images, said method comprising the following steps:
(a) determining a first set of connected components from said first image, and a second set of connected components from said second image; (b) determining a first set of feature descriptors from said first set of connected components and a second set of feature descriptors from said second set of connected components; (c) matching pairs of corresponding feature descriptors between the first and second sets of feature descriptors; (d) applying global homography to said first and second images to reject matched pairs of corresponding feature descriptors that do not satisfy global homography constraints; (e) for each pair of corresponding feature descriptors not rejected in step (d), applying local homography to the pair of corresponding feature descriptors in accordance to the pair's corresponding connected components, and rejecting any pair of corresponding feature descriptors that does not satisfy its local homography constraints, the remaining pairs of corresponding feature descriptors defining current connected components in their respective first and second images; and (f) within the footprints of each pair of corresponding current connected components, for each unknown pixel defined as a pixel not known as being a part of the connected component within whose footprint the unknown pixel resides, determining if a corresponding match for the unknown pixel can be identified in the corresponding paired current connected component, the matched pixels of each pair of current corresponding components identifying a corresponding object in said first and second images of said stereoscopic pair of images.