US 11,956,441 B2
Identifying long term reference frame using scene detection and perceptual hashing
Sunil Gopal Koteyar, Burlington (CA); Sonu Thomas, Markham (CA); Ihab M. A. Amer, Markham (CA); and Haibo Liu, Markham (CA)
Assigned to ATI Technologies ULC, Markham (CA)
Filed by ATI Technologies ULC, Markham (CA)
Filed on Dec. 23, 2021, as Appl. No. 17/561,275.
Prior Publication US 2023/0209064 A1, Jun. 29, 2023
Int. Cl. H04N 19/136 (2014.01); H04N 19/105 (2014.01); H04N 19/142 (2014.01); H04N 19/167 (2014.01); H04N 19/172 (2014.01); H04N 19/423 (2014.01)
CPC H04N 19/136 (2014.11) [H04N 19/105 (2014.11); H04N 19/142 (2014.11); H04N 19/167 (2014.11); H04N 19/172 (2014.11); H04N 19/423 (2014.11)] 20 Claims
OG exemplary drawing
 
1. A method of encoding a video stream, the method comprising:
encoding a plurality of frames of video acquired from different scenes or different points of view;
for a frame of the plurality of frames of video, generating a perceptual indicator based on an array of values, generated after the frame is encoded, which comprises a first set of statistical values and a second set of statistical values, different from the first set of statistical values, wherein the first set of statistical values of the array of values and the second set of statistical values of the array of values are determined from values of pixels of the frame; and
when a first perceptual indicator for a current frame matches a second perceptual indicator of a previously encoded reference frame, encoding the current frame using the previously encoded reference frame.