US 12,236,631 B2
Keypoint detection and feature descriptor computation
Pushkar Gorur Sheshagiri, Bengaluru (IN); Ajit Deepak Gupte, Bangalore (IN); Chiranjib Choudhuri, Bangalore (IN); Gerhard Reitmayr, Del Mar, CA (US); and Youngmin Park, San Diego, CA (US)
Assigned to QUALCOMM Incorporated, San Diego, CA (US)
Filed by QUALCOMM Incorporated, San Diego, CA (US)
Filed on Sep. 24, 2021, as Appl. No. 17/484,601.
Prior Publication US 2023/0095621 A1, Mar. 30, 2023
Int. Cl. G06T 7/73 (2017.01); G06T 7/246 (2017.01); G06V 10/46 (2022.01); G06V 20/40 (2022.01)
CPC G06T 7/73 (2017.01) [G06T 7/246 (2017.01); G06V 10/462 (2022.01); G06V 20/46 (2022.01)] 27 Claims
OG exemplary drawing
 
16. A method of processing one or more frames, comprising:
determining, based on pixel values of a current frame obtained from an image sensor, a location of a feature in the current frame;
obtaining, by a computing device, feature information associated with the feature in the current frame, wherein the feature information is based on one or more previous frames;
determining an estimated pose of the computing device associated with the current frame;
obtaining, from a depth sensor, a distance associated with the feature in the current frame;
determining, by the computing device, an estimated scale of the feature in the current frame based on the feature information associated with the feature, the estimated pose, and the distance associated with the feature;
determining, based on the estimated scale of the feature, a nearest scale space layer associated with the feature, wherein a scale space comprises a plurality of scale space layers associated with different feature scales, the plurality of scale space layers comprising the nearest scale space layer; and
determining, based on searching for an additional feature corresponding to the feature in the nearest scale space layer, a scale of the feature in the current frame, wherein the determined scale is different from the estimated scale.