Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2001-10-09
2004-07-13
Kelley, Chris (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S240060
Reexamination Certificate
active
06763067
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to video processing, and, in particular, to processing existing compressed video bitstreams to generate altered bitstreams.
2. Description of the Related Art
MPEG (Motion Picture Experts Group) and MPEG-like compression systems create a stream of bits wherein successive frames do not contain equal numbers of bits. This requires that decoders have a bitstream buffer at their inputs, and it requires encoders to generate bitstreams that avoid overflowing or underflowing the decoder buffers. MPEG-2 13818-2 Annex C (Normative) describes a buffer model (referred to as the VBV (video buffering verifier) model) that conforming bitstreams must neither overflow nor underflow. A bitstream that overflows or underflows the VBV buffer is not compliant with the standard. The process by which an encoder creates a bitstream constrained to comply with these buffer requirements is commonly known as “rate control.”
Many rate control schemes have been developed. Most are variations of the widely-known “TM5” (Test Model 5) developed during the standardization of MPEG-2. In TM5, rate control has three main steps. The first is budgeting or allocating a certain number of bits for the next picture (based on the intended GOP (group of pictures) structure and the current relative complexity of the three picture coding types (I, P, and B). The second is creating a quantizer scale value that is adjusted macroblock by macroblock based on how the previous macroblocks in the picture have or have not stayed on budget. The third is to modify this quantizer scale value based on psychovisual error masking principles. When the current picture is completely coded, the extent that the target bitcount for the picture was missed is added to or subtracted from the number of bits available to code the remaining pictures in the GOP. At the end of the current GOP, the extent that the target bitcount for the GOP was missed is added to or subtracted from the allocation for the next GOP.
There exists a need to change the bit rate characteristic of existing compressed video bitstreams such that a bitstream is received and converted to a different (usually lower) bit rate. Examples of this include (a) reducing a 45-Mbps (megabits per second) network distribution stream to a 19.39-Mbps ATSC broadcast stream, (b) inserting a logo into a 4-Mbps stream, thus increasing the information load, while maintaining the 4-Mbps bit rate, (c) restructuring the mix of multiple streams in a multiplex (e.g., fitting a 5-Mbps stream into a 4-Mbps channel), and (d) simply reducing the bit rate of a bitstream so that other data may be sent at the same time over a channel having limited bandwidth. In each of these situations, there are two conflicting goals: first, maintaining compliance by avoiding overflow or underflow of the buffer, and, second, maintaining maximum quality by making full use of the buffer size available.
TM5 and similar rate control schemes presuppose that the encoder has control over the GOP structure; that is, the encoder chooses the sequence of picture coding types. When processing an existing compressed video bitstream (e.g., inserting a logo into an existing bitstream by decoding, inserting, and then re-encoding), the best picture quality is typically obtained by using the same sequence of picture coding types. This means, however, that the GOP structure is determined by the incoming bitstream, not by the re-encoder. As such, the GOP and picture bit targeting approach of TM5 and similar rate control schemes will not adequately work when processing existing compressed video bitstreams.
SUMMARY OF THE INVENTION
The present invention is directed to certain applications in which an existing compressed video bitstream is processed to generate an output bitstream that is different from the input bitstream. These applications may involve situations where the imagery represented in the bitstream is changed (such as when a logo is inserted into the imagery encoded in the bitstream) or where the size of the bitstream is changed (such as when the bit rate is reduced) or both.
According to embodiments of the present invention, the input bitstream is used as a reference to which the output bitstream is conformed. In particular, a controller compares measurements of the input and output bitstreams and adjusts the processing applied to the input bitstream so as to attempt to cause future measurements of the output bitstream to match future measurements of the input bitstream.
In one particular class of embodiments, the bitstream processing involves at least partially decoding the input bitstream, optionally applying one or more image processing functions to the decoded video data (e.g., logo insertion), followed by re-encoding of the video data to generate the output bitstream, where the output bitstream may (but does not have to) have a bit rate that differs from that of the input bitstream.
In one implementation, analogous buffer models are applied to both the input and output bitstreams to generate arrival margin values that are then compared to generate Quantization Re-scale Factors (QRFs) that are used to scale quantization data recovered from the input bitstream during the decoding process, where the scaled quantization data are then used to quantize DCT coefficients that are based on the input bitstream to generate quantized DCT coefficients for the output bitstream. The arrival margin for a frame is defined as the amount of time between when the last bit of a frame has entered the buffer and when that frame is removed from the buffer.
According to one embodiment, the present invention is a method for converting an input compressed video bitstream into an output compressed video bitstream, comprising the steps of (a) applying a process to convert the input bitstream into the output bitstream; (b) generating one or more input measurements from the input bitstream; (c) generating one or more output measurements from the output bitstream; and (d) comparing the one or more input measurements to the one or more output measurements to generate one or more control parameters that are used to adjust the process so that subsequent output measurements will more closely match subsequent input measurements.
REFERENCES:
patent: 6023553 (2000-02-01), Boyce et al.
patent: 6181743 (2001-01-01), Bailleul
patent: 6226326 (2001-05-01), Mihara
patent: 6310915 (2001-10-01), Wells et al.
patent: 6493388 (2002-12-01), Wang
patent: 6587508 (2003-07-01), Hanamura et al.
patent: 2002/0181595 (2002-12-01), Obata et al.
patent: 0 739 138 (1996-10-01), None
patent: 0 977 439 (2000-02-01), None
patent: WO 00 46997 (2000-08-01), None
patent: WO 00 51357 (2000-08-01), None
Burke, Esq. William J.
Czekaj Dave
Kelley Chris
Sarnoff Corporation
LandOfFree
Rate control for bitstream re-encoding does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Rate control for bitstream re-encoding, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Rate control for bitstream re-encoding will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3188922