Video coding system

• US 20050152451A1
• Filed: 01/12/2005
• Published: 07/14/2005
• Est. Priority Date: 01/13/2004
• Status: Active Grant

• Current Assignee: CS Technology
• Original Assignee: CS Technology
• Inventors: Byun, Hyun Il
• Granted Patent: US 7,620,104 B2
• Time in Patent Office: Days
• US Class Current: 375/240.30
• CPC Class Codes: H04N 19/42 characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation H04N19/635 takes precedence
• CPC Class Codes: H04N 19/44 Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
• CPC Class Codes: H04N 19/61 in combination with predictive coding
• CPC Class Codes: H04N 19/91 Entropy coding, e.g. variable length coding [VLC] or arithmetic coding

Abstract

Disclosed is a digital video coding system and more particularly technology capable of reducing the load of a CPU (Central Processing Unit) that controls a video coding system for implementing a digital video data compression/decompression standard such as MPEG-1, 2, 4, H.261 or H.263. In a conventional video coding system, the circuit configuration and control method of a variable-length encoder are becoming increasingly complex with the development of a video compression/decompression standard. In a conventional system structure, a complex control operation must be carried out, such that there is a problem in that the load of the CPU controlling the video coding system increases. Thus, the video coding technology is provided to reduce the load of the CPU necessary for controlling a VLC (Variable Length Coding) unit.

Claims

1. A video coding system for performing video coding through DCT (Discrete Cosine Transform), MC (Motion Compensation) and ME (Motion Estimation), comprising:

an MC unit for producing an MAD (Mean Absolute Difference) value of a current macroblock;

a control unit for estimating the number of bits of the current macroblock based on the MAD value and a QP (Quantization Parameter) and controlling an end of a VP (Video Packet); and

a VLC (Variable length coding) unit for performing the ME, texture coding, and a packing task associated with all macroblocks of the VP under control of the control unit.

2. The video coding system of claim 1, wherein the control unit estimates the number of bits of the current macroblock by means of an equation and decides the end of the VP when a value of adding a “

bit_est”

value to a sum of the number of bits of up to a previous macroblock is larger than a predetermined target VP length, and

wherein the equation is “

bit_est=k×

(MAD/QP)+offset”

where “

bit_est”

denotes the estimated number of bits, “

k”

denotes a constant produced by experimentation, and “

offset”

is a parameter for controlling the probability of a macroblock to be coded within the estimated number of bits.

3. The video coding system of claim 1, wherein the VLC unit comprises:

a header coding control part (header_ctrl) responsible for controlling header coding;

a header ROM (Read Only Memory) table (mb_rom) including various tables necessary for the header coding;

a header memory (hmem) for storing a result of the header coding;

a texture coding control part (texture_ctrl) responsible for controlling texture coding;

a texture ROM table (ac_rom) including various tables necessary for the texture coding;

a texture memory (tmem) for storing a result of the texture coding;

a packing execution part (pack_ctrl) responsible for packing the header and texture; and

a VLC register (vlc_reg) for storing various setup items necessary for the header and texture coding.

4. The video coding system of claim 3, wherein the packing execution part (pack_ctrl) comprises:

a packing control sub-part responsible for entirely controlling the packing execution part (pack_ctrl);

a packing operation sub-part for performing the packing task;

a texture memory control sub-part for sending data necessary for the packing task from the texture memory (tmem) to the packing operation sub-part; and

a header memory control sub-part for sending data necessary for the packing task from the header memory (hmem) to the packing operation sub-part, and writing the packed data to the header memory (hmem).

5. The video coding system of claim 4, wherein the packing control sub-part determines whether or not the header coding and the texture coding have been completed, respectively, through signals “

tend” and

“

hend”

when a signal “

pack_en”

, “

toverflow”

or “

hoverflow”

is 1, and sends a packing start command through a signal “

p_start”

if the header coding and the texture coding have been completed, respectively.

6. The video coding system of claim 4, wherein the packing operation sub-part comprises:

a shift &

merge element for appropriately shifting and merging data of two A-bit units from the texture memory (tmem) and/or the header memory (hmem), and generating data of a new A-bit unit as a result of packing, “

A”

contained in the A-bit unit being a value equal to or greater than a width value of the texture memory (tmem) and a width value of the header memory (hmem); and

a register file element for storing information (“

pre_h1_start_pos”

, “

pre_h2_start_pos” and

“

pre_t_start_pos”

) indicating start positions of Header 1, Header 2 and texture data of a first macroblock of the VP so that coded data of all macroblocks included in the VP can be packed.

7. The video coding system of any one of claims 1 to 6, wherein coded data of the current macroblock is packed and written to a header memory (hmem) in case of a general VP.

8. The video coding system of any one of claims 1 to 6, wherein coded data of all macroblocks included in the current VP is packed and written to a header memory (hmem) when VP data is partitioned.

9. The video coding system of claim 8, wherein the video coding system performs the steps of:

when the texture memory (tmem) or the header memory (hmem) overflows while texture coding on an N-th macroblock is performed, compulsorily packing data from a first macroblock up to an (N−

1)-th macroblock in the current VP;

re-performing the ME on the N-th macroblock; and

simultaneously performing the texture coding on the N-th macroblock and the ME on the (N+1)-th macroblock.