Image analysis – Image compression or coding – Including details of decompression
Reexamination Certificate
1998-08-14
2002-05-14
Tran, Phuoc (Department: 2621)
Image analysis
Image compression or coding
Including details of decompression
C382S246000, C382S250000
Reexamination Certificate
active
06389171
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to decoding of Digital Video Cassette video images and more particularly to decoding variable length coded data to provide a data stream of decoded data.
2. Description of the Related Arts
As computers become more and more powerful, the fascination of consumers and professionals alike with digital graphics becomes more and more acute. Digital graphics enable users to manipulate, transfer and store the digital graphics as data files on computers. Digital cameras are one of the first devices to take advantage of digital capture without demanding an intermediate step of first scanning the particular graphic depiction. On the heels since the introduction of the digital cameras are digital video recorders. Albeit the current prices for digital video recorders are not for the average consumer or even professional, there is still an outpouring of next generation digital video camcorders that are available. However, as the digital video camcorders become more widely accepted, the prices of the digital camcorders will drop allowing many consumers to afford the digital camcorders.
One inherent impasse of the digital video camcorders or even digital cameras is converting and reconverting the mass amount of data that represents the recorded digital images to a computer system where the user of the computer system can manipulate, transfer, or store the digital images. Thus, sophisticated encoding techniques have been developed to encode the ever increasing digital information into an ever smaller space in efforts to make digital cameras and digital camcorders more attractive for the users. Some of the digital image encoding techniques include JPEG, MPEG I and II. DVC or digital video (DV) is another encoding technique. Given the goal of digital image encoding is to encode as much data into as little of space as possible without losing detailed information, the DVC encoding technique produces variable length coding to produce more efficient coding. The variable length coding distributes coded data throughout a fixed encoded data structure. The hierarchy of the DVC coded building blocks is as follows: Video Frame (720×480 NTSC, 720×576 PAL)
DIF (Digital Interchange Format) Sequence (10 DIF Sequences per frame for NTSC, 12 for PAL)
Super Block (5 per DIF Sequence)
Video Segment (consists of 5 macro blocks or DIF blocks, 27 per DIF Sequence)
Macro or DIF Block (typically represents an 8×32 pixel area for NTSC, and a 16×16 pixel area for PAL)
Luminance and Chrominance Difference DCTs (4 luma (Y) and 2 chroma (Cr, Cb) per DIF block).
The variable length coding process for DVC is similar to other DCT based compression algorithms such as JPEG or MPEG. After quantization the AC coefficients are run length encoded which results in a series of run length-amplitude pairs. Run length refers to the number of consecutive zero AC coefficients, and amplitude refers to the amplitude of the AC coefficient at the end of the run of zero coefficients (e.g. run
3
, amplitude
12
represents 3 zero amplitude coefficients followed by a coefficient amplitude equal to
12
). The variable length code word associated with each run-amplitude pair is determined by a fixed Huffman table (Table 25, Helical-scan digital videocassette recording system using 6.35 mm magnetic tape for consumer use, IEC 61834-2 Part 2) page 169. For each DCT the dc coefficient, the variable length codewords, and an end of block (EOB) codeword are concatenated together to form the core of a variable length data stream.
Once the DCT data has been coded as an encoded data stream consisting of the dc coefficient, variable length codewords, and an EOB, the encoded data stream is stored into the fixed encoded data structure based on the hierarchy of the DVC encoded building blocks. The basic element of the fixed data structure is a DIF block that is shown in FIG.
9
. The DIF block consists of a compressed macro block and three bytes, ID
0
-ID
2
. The three bytes ID
0
-ID
2
identify the position of the compressed macro block in the data stream. Each compressed macro block includes data associated with 4 luminance (Y
0
-3) and 2 chroma difference (Cr, Cb) DCTs. Each DCT component starts with a 1.5 byte header consisting of a dc coefficient value, class number, and a DCT m
0
bit. The DCT m
0
bit indicates whether the DCT mode is the standard 8×8 DCT or a dual 4×8 (2-4×8) DCT.
FIG. 9
is deceiving because it implies that all of the data associated with a particular DCT, such as Y
0
, is stored in the area marked Y
0
. The actual data distribution of the DCT components is significantly more complex. A three pass encoding of the DCT components distributes the variable length coded data associated with a particular DCT component. In some cases the variable length coded data associated with a particular DCT component can be distributed with other DCT components.
The first pass attempts to place the variable length coded data associated with a particular DCT in an area assigned to that DCT (e.g. luminance DCT Y
0
's data would go into the DCT area labeled Y
0
). The luminance areas of the DIF block are allocated 12.5 bytes and the chrominance areas of the DIF block are allocated 8.5 bytes. The variable length coded data which is not stored in the allocated areas for the first pass is concatenated into individual DIF block overflow buffers (e.g. overflows from Y
0
through Cb for DIF block
0
is stored in a DIF block
0
overflow buffer, Y
0
through Cb for DIF block
1
is stored in a DIF block
1
overflow buffer, etc.).
For the second pass the data in the overflow buffers is distributed back into any free area in the associated DIF block (e.g. a Y
0
overflow for DIF block
0
could go into any empty area left in Y
1
through Cb in DIF block
0
). Any coded data which cannot be placed back into the DIF block by this pass is concatenated into a single global overflow buffer, referred to as the video segment buffer (VSB).
For the third and final pass the coded data contained within the global buffer is distributed into any remaining unused area within the video segment.
FIG. 10
provides an example of the three pass encoding of the DCT components for the first two DIF blocks (macro blocks) of a video segment. The variable length coded AC coefficients for each DCT start as a variable length structure with an end of block (EOB) code concatenated to the end of the data. For DIF Block A, any code data exceeding 12.5 bytes for each of Y
0
, Y
1
, and Y
2
and exceeding 8.5 bytes for Cr and Cb is placed into a buffer labeled DBA. For DIF Block B, the excess coded data for Y
1
and Y
2
is placed into a buffer labeled DBB. In pass 2 of the variable length coding, part of the coded data from DBA is placed back into DIF Block A. Because there is not enough space to contain all of the coded data, the excess coded data is stored into the video segment buffer (VSB). For DIF Block B, all of the coded data temporarily stored in DBB is absorbed back into the DIF block B, hence no additional data is added to the VSB buffer. During pass 3 the coded data left in VSB is placed into the open area that remains in DIF Block B.
To insure that the coded data fits in the allocated area during encoding, adjustment of the quantization levels for the AC coefficients controls the variable length coded data size. For example, as the quantization of the upper frequency AC coefficients gets more coarse (less granular) more of the AC coefficient values will drop to zero. Thus, the variable length coding process becomes more efficient as more AC coefficients drop to zero which results in a reduced storage requirement. However, some fine details for the original picture may be lost if too many AC coefficients values drop to zero.
The audio encoding for the DVC process is fairly straightforward providing for a 2's complement representation of each audio sample for the 48 k, 44.1 k, and 32 k one channel modes (where, for example, 48 k represents a 48 kHz
Apple Computer Inc.
Blakely , Sokoloff, Taylor & Zafman LLP
Tran Phuoc
LandOfFree
Method and apparatus for a digital video cassette (DVC)... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for a digital video cassette (DVC)..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for a digital video cassette (DVC)... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2872298