Pseudo-constant bit rate video coding with quantization parameter adjustment

• US 5,241,383 A
• Filed: 05/13/1992
• Issued: 08/31/1993
• Est. Priority Date: 05/13/1992
• Status: Expired due to Term

• Current Assignee: Shingo Limited Liability Company
• Original Assignee: Bell Communications Research, Inc. (Telefonaktiebolaget LM Ericsson)
• Inventors: Chen, Cheng-Tie, Wong, Andria H.
• Primary Examiner(s): Britton, Howard W.
• Assistant Examiner(s): LIN, BRYAN T
• Application Number:
  US07/882,487
• Time in Patent Office: 475 Days
• Field of Search: 358/133, 358/136, 358/261.2, 358/404, 358/430, 358/432, 358/466
• US Class Current: 375/240.04
• CPC Class Codes: H04N 19/107 between spatial and temporal predictive coding, e.g. picture refresh
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/126 Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
• CPC Class Codes: H04N 19/139 Analysis of motion vectors, e.g. their magnitude, direction, variance or reliability
• CPC Class Codes: H04N 19/146 Data rate or code amount at the encoder output
• CPC Class Codes: H04N 19/149 by estimating the code amount by means of a model, e.g. mathematical model or statistical model
• CPC Class Codes: H04N 19/152 by measuring the fullness of the transmission buffer
• CPC Class Codes: H04N 19/172 the region being a picture, frame or field
• CPC Class Codes: H04N 19/177 the unit being a group of pictures [GOP]
• CPC Class Codes: H04N 19/577 Motion compensation with bidirectional frame interpolation, i.e. using B-pictures
• CPC Class Codes: H04N 19/60 using transform coding
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

A method for coding frames of video wherein a coding circuit (14,16,18) includes a processor (30) for performing an orthogonal transform such as a discrete

cosine transform and a quantizer (32) for quantizing the resulting transform coefficients. The coding circuit codes the video frames using intra-frame, predictive or interpolative coding to generate code bits at a variable rate. The code bits are stored at a variable rate in a rate buffer (22), which transmits the code bits into a communication channel (24) at a pseudo-constant rate, i.e. a rate which is constant in every time interval of one frame. To maintain the contents of the rate buffer (22) within predetermined limits, the quantization parameters utilized by the quantizer (32) are periodically adjusted to increase or decrease the amount of code bits generated by the coding circuit. The quantization parameters are changed on a global SGOP level to avoid changes of quantization parameters and corresponding changes in decoded image quality within particular frames. The change in quantization parameters for coding the next SGOP is determined by a deviation measure between the actual number of code bits generated by the coding circuit (14,16,18) for the previous SGOP and an estimate of the number of code bits for the previous SGOP. The estimated number of code bits is determined based on the contents of the rate buffer (22) such that the rate buffer (22) will be emptied in a predetermined time period.

Claims

1. A method for coding video frames comprising the steps of

coding said video frames using a coding circuit including a processor for performing an orthogonal transform to produce transform coefficients and a quantizer for quantizing the transform coefficients to generate code bits at a variable bit rate,

storing the code bits in a rate buffer at a variable bit rate and transmitting the code bits out of the rate buffer and into a communication channel at a substantially constant bit rate and,

maintaining the contents of said rate buffer in a predetermined range by adjusting quantization parameters utilized by said quantizer to increase or decrease the amount of code bits generated by said coding circuit for a next set of one or more frames, the adjustment of quantization parameters being in response to a deviation between the actual number of code bits generated by the coding circuit for a previous set of one or more frames and an estimate of the number of code bits for said previous set of one or more frames.

2. The method of claim 1 wherein said estimate is determined in accordance with the contents of the rate buffer such that the rate buffer will be emptied in a predetermined time period.

3. The method of claim 1 wherein said coding circuit codes said video frames using intra-frame, predictive and interpolative coding.

4. The method of claim 1 wherein said step of maintaining the contents of said rate buffer in a predetermined range by adjusting quantization parameters comprises adjusting the quantization parameters within a predetermined range.

5. The method of claim 1 wherein said video frames are organized into groups of frames (GOPs) with each group of frames being in turn organized into a first frame which is intra-frame coded by said coding circuit and a plurality of sub-groups of frames (SGOPs) whose frames are predictively or interpolatively coded by said coding circuit, said step of maintaining the contents of said rate buffer in a predetermined range by adjusting said quantization parameter taking place once for each sub-group of frames.

6. The method of claim 5 wherein said estimate is determined such that said rate buffer will be emptied within a predetermined number of GOPs.

7. A method for coding video frames organized into sets of one or more video frames comprising the steps of:

generating code bits at a variable bit rate by a coding circuit to code said video frames,

storing said code bits in a rate buffer, and

transmitting said code bits from said rate buffer via a communication channel at a substantially constant bit rate,

the step of generating code bits comprising the step of selecting a next quantizer in response to a deviation of an actual number of code bits generated by said coding circuit for a previous set of frames from an estimated number of code bits for said previous set of frames to maintain the contents of said rate buffer in a predetermined range.

8. The method of claim 7 wherein quantization parameters of the selected next quantizer do not deviate by more than a predetermined amount from quantization parameters of a quantizer previously selected for the previous set of frames.

9. The method of claim 7 wherein the estimated number of code bits for the previous set of frames is determined so that the rate buffer will be emptied after a predetermined number of sets of frames are coded.

10. The method of claim 7 wherein said method includes the step of halting the coding of said frames until the contents of said rate buffer fall below a second specific limit, if the contents of said rate buffer exceed a first specific limit.

11. The method of claim 7 wherein particular ones of said video frames are intra-frame coded, predictively coded, and interpolatively coded by said coding circuit.

12. The method of claim 7 wherein said step of generating code bits at a variable bit rate by a coding circuit comprises the step of generating orthogonal transform coefficients from pixel values contained in said video frames, quantizing the coefficients using the selected quantizer and converting the quantized transform coefficients into fixed or variable length code words.

13. The method of claim 7 wherein said video frames are organized into groups of frames (GOPs), wherein each group of frames is further organized into a first frame and a plurality of sub-groups of frames (SGOPs), wherein the first frame of each group of frames is intra-frame coded using said coding circuit, wherein the frames of each sub-group of frames are interpolatively or predictively coded by said coding circuit, and wherein said step of selecting a next quantizer is carried out once for each sub-group of frames.

14. A method for coding a series of video frames organized into groups of frames (GOPs), which in turn are organized into sub-groups of frames (SGOPs), through use of a coding circuit comprising a processor for carrying out an orthogonal transform and a quantizer for quantizing the coefficients generated by said orthogonal transform, and a rate buffer for receiving code bits from said coding circuit at a variable rate and outputting the code bits into a communication channel at a substantially constant rate, said method comprising the steps of:

after coding a previous SGOP by said coding circuit utilizing a previous quantizer, selecting a next quantizer from a plurality of available quantizers for the next SGOP by determining a deviation measure of the actual number of code bits generated by said coding circuit for the previous SGOP from an estimated number of code bits for the previous SGOP, and utilizing the deviation measure to select said next quantizer,

determining an estimated number of code bits for the next SGOP based on the contents of the frame buffer after coding the previous SGOP,

coding said next SGOP using said processor and said next quantizer and storing the coded bits of the next SGOP in said rate buffer,

said next quantizer being chosen such that the next quantizer has quantization parameters which are constrained to be within a predetermined range of quantization parameters of the previous quantizer.

15. The method of claim 14 wherein said orthogonal transform is a discrete cosine transform.

16. The method of claim 14 wherein each GOP comprises N frames including a first frame and L SGOPs each of which comprises M frames, whereby N=ML+1, and wherein the first frame and first SGOP of a GOP are coded using a quantizer selected through use of a deviation measure of the actual number of code bits generated by said coding circuit for the last SGOP of the previous GOP from an estimated number of code bits for the last SGOP of the previous GOP.

17. The method of claim 14 wherein the first frame of each GOP is intra-frame coded by said coding circuit and the frames of each SGOP in a GOP are predictively or interpolatively coded by said coding circuit.

18. The method of claim 14 wherein said plurality of available quantizers comprises

S sets of quantization matrices Q_i, (i=0,1, . . . , S-1) and wherein each quantization matrix Q_i is comprised of elements q_i =(m,n) such that

19. The method of claim 18 wherein the step of selecting a next quantizer comprises the step of selecting a next quantization matrix such that i_next -i_previous ≦

J where i_next is the index of the next quantization matrix, i_previous is the index of a previous quantization matrix and J is a predetermined integer.

20. The method of claim 19 wherein the step of selecting a next quantization matrix comprises selecting the index i_next of the next quantization matrix such that ##EQU8## where j=1,2, . . . , J and where the values t_-31 j-1,t_-j, . . . , t_j, t_j+1 are predetermined thresholds and where ID_previous is said deviation measure.

21. The method of claim 20 wherein the deviation measure for selecting said next quantization matrix is determined by ##EQU9## where SGOPBITS_previous is the number of code bits actually generated by said coding circuit for said previous SGOP, EB_previous is the estimated number of code bits for said previous SGOP, and R_SGOP is a desired number of bits per SGOP transmitted from said rate buffer to said communication channel.

22. The method of claim 14 wherein the estimated number of code bits for the next SGOP is determined such that the rate buffer will be emptied after a predetermined number of GOPs.

23. The method of claim 14 wherein said method further includes the step of halting the coding of said video frames by said coding circuit until the contents of said rate buffer fall below a second predetermined level, if the contents of said rate buffer exceed a first predetermined level.