Granted Patent B2

US 8,233,534 · App. 12/757,438 · Granted Jul 31, 2012

Abstract

Claims (44)

1. A method for frame buffer video compression on an input video stream to efficiently allocate and utilize memory used for storing reference frames in a video decoder, said method comprising:

parsing a video header to obtain a maximum number of reference frames N;

dividing an available frame buffer memory to hold N+1 compressed frames;

controlling a compression process to limit a size of each frame such that said size is not less than a frame buffer memory size divided by N+1;

parsing messages that contain reference frames by looking ahead in an input video stream buffer of said input video stream; and

allocating memory for compressing a given reference frame based on a number of needed frames at a frame buffer,

wherein a first frame is decoded by looking ahead three frames to know which frames are said reference frames, and only reference frames are compressed and stored in said frame buffer, and

wherein when a fourth frame is decoded, half of said frame buffer is allocated for a third reference frame and only one-fourth of said frame buffer is allocated for said fourth frame.

2. The method of claim 1 , further comprising adjusting a quantization of said compression process to improve a picture quality of said input video stream.

3. The method of claim 1 , wherein each of said reference frames is compressed using a JPEG image compression technique and stored in said memory, and said compression process is controlled in a quantization stage by dividing said input by a quantization matrix.

4. The method of claim 1 , wherein said messages are parsed based on information about said reference frames, said information comprising slice headers.

5. The method of claim 3 , wherein if only one reference frame is needed in which said reference frame requires a memory equal to said size of frame buffer memory divided by N+1, the quantization of the compression technique is adjusted to increase picture quality and avoid compression artifacts.

6. The method of claim 3 , wherein said quantization matrix is adjusted to limit an output size to said size of frame buffer memory divided by N+1.

7. The method of claim 1 , wherein said incoming video stream is stored in a buffer at an input to said video decoder.

8. The method of claim 1 , wherein said input video stream comprises an analog input video stream.

9. A video decoder that performs buffer video compression on an input video stream and efficiently allocate and utilize memory used for storing reference frames, said video decoder comprising:

a frame buffer;

a parsing unit that parses a video header of said input video stream to obtain a maximum number of reference frames N and parse messages that contain reference frames by looking ahead in said frame buffer;

a memory allocation unit that divides an available memory of a frame buffer to hold N+1 compressed frames and allocate memory for compressing a given reference frame based on a number of needed frames at said frame buffer at a given instance; and

a compression control unit that controls a compression process to limit the size of each frame such that said size is not less than the memory size divided by N+1, said compression control unit comprising a quantization adjustment unit that divides said input by a quantization matrix and controls said compression process in a quantization stage, wherein each frame is decoded by looking ahead in said frame buffer to determine which frames are said reference frames,

wherein a first frame is decoded by looking ahead three frames to know which frames are said reference frames, and only reference frames are compressed and stored in the frame buffer, and

wherein when a fourth frame is decoded, half of said frame buffer is allocated for a third reference frame and only one-fourth of said frame buffer is allocated for said fourth frame.

10. The video decoder of claim 9 , wherein each of said reference frames is compressed using a JPEG image compression technique and stored in said memory.

11. The video decoder of claim 10 , wherein if only one reference frame is needed in which said reference frame requires a memory equal to said size of frame buffer memory divided by two, said quantization adjustment unit adjusts a quantization of said compression technique to increase picture quality and avoid compression artifacts.

12. The video decoder of claim 9 , wherein said memory allocation unit removes frames that are not reference frames from said frame buffer.

13. The video decoder of claim 12 , wherein said parsing unit parses said messages based on information about said reference frames, said information comprising slice headers.

14. The video decoder of claim 9 , wherein said quantization matrix is adjusted to limit an output size to said size of frame buffer memory divided by N+1.

15. The video decoder of claim 9 , wherein said input video stream comprises an analog input video stream.

16. A system for performing frame buffer video compression on an input video stream to efficiently allocate and utilize memory used for storing reference frames in a video decoder, said system comprising:

means for parsing a video header to obtain a maximum number of reference frames N;

means for dividing an available frame buffer

memory to hold N+1 compressed frames; and

means for controlling a compression process to limit a size of each frame such that said size is not less than a frame buffer memory size divided by N+1;

means for parsing messages that contain reference frames by looking ahead in an input video stream buffer of said input video stream, wherein said messages are parsed based on information about said reference frames, said information comprising slice headers;

means for allocating memory for compressing a given reference frame based on a number of needed frames at a frame buffer,

wherein each of said reference frames is compressed using a JPEG image compression technique and stored in said memory, and said compression process is controlled in a quantization stage by dividing said input by a quantization matrix,

wherein said quantization matrix is adjusted to limit an output size to said size of frame buffer memory divided by N+1,

wherein only reference frames are compressed and stored in the frame buffer, and each frame is decoded by looking ahead in said frame buffer to determine which frames are reference frames,

wherein a first frame is decoded by looking ahead three frames to know which frames are said reference frames, and only reference frames are compressed and stored in said frame buffer, and

wherein when a fourth frame is decoded, half of said frame buffer is allocated for a third reference frame and only one-fourth of said frame buffer is allocated for said fourth frame.

17. The system of claim 16 , further comprising means for adjusting a quantization of said compression process to improve a picture quality of said input video stream.

18. The system of claim 17 , wherein if only one reference frame is needed in which said reference frame requires a memory equal to said size of frame buffer memory divided by two, the quantization of the compression technique is adjusted to increase picture quality and avoid compression artifacts.

19. The system of claim 16 , wherein said incoming video stream is stored in a buffer at an input to said video decoder.

20. The system of claim 16 , wherein said input video stream comprises an analog input video stream.

Assignments (23)