Method and apparatus to detect anchor frames from digital video streams
AssigneeGeneral Instrument Corporation
InventorsRobert S. Nemiroff, Yong He
Life of this patent
Estimating the term…
Abstract
A method (60) and apparatus process video in a multiple encoder system by identifying (61) those frames that were originally encoded as anchor frames in the first encoder and ensuring that a second encoder encodes (65) these identified frames as anchor frames rather than bidirectional frames or non-anchor frames. In addition, the second encoder identifies (62) those frames that were originally encoded as non-anchor frames in the first encoder and encodes (66) the non-anchor frames, such as bidirectional frames, as non-anchor frames during the second encoding process. Once the anchor and non-anchor frames are identified, a frame-type pattern is determined (63) by the first encoder and synchronizes (64) the second encoder to the frame-type pattern.
Claims44
A method (60) for processing video, the method comprising the steps of:
- identifying (61) one or more anchor frames among a sequence of frames encoded by a first encoder
- determining (63) a pattern of frame types present in the sequence of frames encoded by the first encoder by searching (91) for one of three frame type patterns, including:
- two non-anchor frames between each anchor frame;
- one non-anchor frame between each anchor frame, or no non-anchor frames between each anchor frame;
- and
- encoding (65) the one or more anchor frames as anchor frames in a second encoder.
A method (60) for processing video, the method comprising the steps of:
- identifying (61) one or more anchor frames among a sequence of frames encoded by a first encoder by using edge detection on each frame to identify a difference in a number and a strength of edges in a frame as a basis for determining frame type;
- and
- encoding (65) the one or more anchor frames as anchor frames in a second encoder.
An apparatus (20) for processing video that was previously encoded by a first encoder comprising:
- a second encoder (24) to re-encode the video;
- a processor (21) receiving the video coupled from a first encoder, determining frame activity data for each frame of the video by dividing a frame into a predetermined number of fields, dividing each field into a predetermined number of blocks, summing, for each of the blocks, an absolute difference between each horizontally adjacent pixel and each vertically adjacent pixel, and adding a result of the summing for each of the fields to obtain the frame activity data, and determining a potential frame type for subsequent encoding by the second encoder (24);
- a pattern detector (22) coupled to the processor to receive the frame activity data for each frame of the video and to determine an encoded frame type for each frame encoded by the first encoder by searching for one of three frame type patterns, including:
- two non-anchor frames between each anchor frame one non-anchor frame between each anchor frame or no non-anchor frames between each anchor frame;
- and
- a frame-type comparator and adjuster (23) coupled to the processor (21) and to the pattern detector (22) to receive the encoded frame type encoded by the first encoder and the potential frame type for subsequent encoding by the second encoder (24), to synchronize an actual frame type of subsequent encoding by the second encoder (24) to the encoded frame type of the first encoder by adjusting the potential frame type for subsequent encoding by the second encoder (24) to a modified frame type, wherein said second encoder (24) is coupled to the frame-type comparator and adjuster (23) and receives the modified frame type and re-encodes the video using the modified frame type.
Dependencies are inferred from the claim text.