US 12,073,533 B1
Detecting specular surfaces
Anselm Grundhoefer, Campbell, CA (US); Mohamed Selim Ben Himane, Milpitas, CA (US); and Daniel Kurz, Sunnyvale, CA (US)
Assigned to Apple Inc., Cupertino, CA (US)
Filed by Apple Inc., Cupertino, CA (US)
Filed on Aug. 17, 2022, as Appl. No. 17/820,442.
Claims priority of provisional application 63/240,427, filed on Sep. 3, 2021.
Int. Cl. G06T 3/60 (2024.01); G06V 10/60 (2022.01)
CPC G06T 3/60 (2013.01) [G06V 10/60 (2022.01); G06T 2207/30244 (2013.01)] 14 Claims
OG exemplary drawing
 
1. A method for identifying a specular surface in a scene, comprising:
at one or more processors:
receiving an image of the scene;
extracting, from the image, a set of natural features;
generating, from the image, a set of synthesized mirrored features;
determining one or more correspondences between the set of natural features and the set of synthesized mirrored features;
identifying, based on the one or more correspondences, that the set of natural features comprises one or more features representing a specular surface in the scene, the identifying further comprising:
identifying a first subset of natural features from the determined one or more correspondences;
identifying a second subset of natural features from the determined one or more correspondences;
determining that the first subset of natural features comprise a mirrored version of the second subset of natural features; and
identifying that the first subset of natural features represents the specular surface in the scene by determining that the location of the first subset of natural features is farther from a device from which the image is captured than the second subset of natural features is from the device; and
determining a location of the specular surface in the scene based on a location of the one or more features representing the specular surface in the scene.