Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1997-01-27
2001-01-30
Britton, Howard (Department: 2713)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C348S715000, C348S716000
Reexamination Certificate
active
06181746
ABSTRACT:
DETAILED DESCRIPTION OF THE INVENTION
1. Field of the Invention
This invention relates to an image data decoding method and apparatus and, more particularly, to a method and apparatus for decoding encoded input image data, storing them in display units into a memory, and reading the decoded data in display units to display. This invention can be applied, for example, to an MPEG decoder.
2. Description of the Prior Art
Improving technology for compressing images into codes and storing them into various record media such as a CD-ROM and a DAT has become popular; their representative is the international encoding standards MPEG. Today home appliance manufacturers and computer manufacturers are making efforts at developing multimedia information home appliances and are intending to market goods which meet MPEG. Here is a brief description of the processing based on MPEG.
FIG. 1
is a typical diagram showing the flow of encoding and decoding images based on MPEG. As shown in
FIG. 1
, images input from an image input device
2
, such as a camera, are compressed and encoded by a video encoder
4
. Quantization processes and DCT (discrete cosine transform) are usually involved in encoding. Encoded data is written into a record medium
6
.
Decoding follows the reverse process: data is read from the record medium
6
and decoded by a video decoder
8
. Decoding is preceded by a reverse quantization process and reverse DCT. Decoded image data by the video decoder
8
are output in a displayable format at displayable timing and reproduced and displayed by a display device
10
. Decoding is performed with forward prediction from past reproduced images and reverse prediction from future reproduced images (bidirectional prediction), as necessary.
FIG. 2
is a schematic diagram of a GOP (group of pictures) of MPEG. Fifteen screens in
FIG. 2
(called pictures in MPEG) can be regarded to be taken in this order. These fifteen pictures, called a GOP in MPEG, form one predictive process unit. That is, encoding and decoding can be done through cross reference to pictures within a GOP, which section that a GOP is a unit of random access. Motion images can be encoded and decoded by continuously processing sets, each of which is comprised of one GOP and its sequence header. As shown in
FIG. 2
, there are three kinds of pictures: I, P, and B. An I picture (image encoded within its frame) is limitedly encoded within its own frame and reference to other pictures is not necessary for decoding. A P picture (image encoded with forward prediction) requires forward prediction; only past reproduced images are necessary for decoding. However, a B picture (image encoded with bidirectional prediction) requires bidirectional prediction; I and P pictures which are reproduced after the B picture will also be referred. In
FIG. 2
the direction of prediction is shown with an arrow. When a GOP is actually encoded, the data of I or P pictures input after the B picture must be known. Therefore, I
2
picture, for example, is encoded before B
0
and B
1
pictures and the encoded data is written into the record medium
6
. In
FIG. 2
, a cycle M of I or P pictures applied to bidirectional prediction is three.
FIG. 3
is a diagram showing the decoding and reproducing order based on MPEG. The upper row of
FIG. 3
indicates the decoding order; the lower row indicates the reproducing order. The decoding order is the same as the encoding order; that is, the decoding order corresponds to the order in which pictures are arranged in the record medium
6
. Therefore, on the decoding side, I
2
, B
3
, B
4
, . . . pictures are read in this order from the record medium
6
, decoded, re-ordered and then output in the original order. In
FIG. 3
, the I
2
picture must not be output before the B
0
and B
1
pictures, so the I
2
picture is held in an internal memory until the B
0
and B
1
pictures are output. Similarly, other I and P pictures must be held in a memory until their corresponding B pictures are output. As shown in
FIG. 3
, decoded B pictures are output as soon as possible in order to adopt a small capacity memory.
FIG. 4
is a diagram showing the process timing of decoding and displaying by the conventional video decoder
8
. The lower part of
FIG. 4
shows memory banks in the video decoder
8
and pictures held in each bank at each timing. This memory has
4
banks for the reason described below. Each bank corresponds to an area in which one frame (one picture) of data is stored.
As shown in
FIG. 4
, periods of frames FL
1
-
6
depend on vertical synchronizing signals. Each frame period consists of two field periods. It is assumed that the low state of a field signal indicates the first field period and the high state of the field signal indicates the second field period.
In
FIG. 4
, the time at which I
2
, B
0
, and B
1
pictures of some GOP have already been decoded is the starting point of FL
1
. Subsequently, P
5
, B
3
, B
4
, . . . pictures will be decoded as in the decoding order of FIG.
3
. However, with display operation, the starting point of FL
1
occurs only when the B
0
picture has been displayed. Subsequently, the B
1
, I
2
, . . . pictures are displayed as in the reproducing order of FIG.
3
. Decode and display operation in each frame period will be described below.
(1) FL
1
The B
1
picture is displayed. To do this, the B
1
picture is held in a bank (assumed to be a bank 1) until the end of FL
1
. The I
2
picture to be displayed in FL
2
is still held in another bank (assumed to be bank 0). Apart from display operation, the P
5
picture is decoded. To store the decoded data, still another bank (assumed to be a bank 2) is assigned. Decoding is done in macroblocks (one macroblock is
16
pixels by
16
lines) and data is stored in macroblocks in a memory. In this frame a bank 3 is an empty area (unused).
(2) FL
2
The I
2
picture is displayed. The I
2
picture is held in bank 0 until the end of this frame. The P
5
picture to be displayed later is still held in the bank 2. The data of the B
1
picture which has already been displayed is unnecessary; therefore, bank 1 is assigned to the B
3
picture to be decoded. In this frame, too, the bank 3 is an empty area.
(3) FL
3
The I
2
picture had been displayed, but it is still held in bank
0
because the reference to the I
2
picture is needed for the B
4
picture which is being decoded. The B
3
picture which is being displayed and the P
5
picture to be displayed later stay in the banks. The B
4
picture which is being decoded is stored in macroblocks in bank 3. To achieve this, the memory must have four banks.
In this way a timing chart of
FIG. 4
is obtained according to the rules: “a B picture which has been already displayed is unnecessary” and “I and P pictures are held for a certain period to decode the B pictures between them.” As shown in
FIG. 4
, all four banks are used every three frame periods (FL
3
,
6
,
9
, . . . ).
As is stated, the conventional video decoder
8
needs memory having four banks to hold a decoded picture until the subsequent B pictures are decoded and to hold the present picture until it is actually displayed.
Memory capacity necessary to reserve four banks will be described below. For a screen of an NTSC system having 352 pixels×240 lines, one frame corresponds to about 123.8 Kbytes of data. Four frames have 495.2 Kbytes of data. In addition, a 40-50 Kbyte temporary buffer called a VBV buffer should be prepared with MPEG, resulting in a total capacity of about 540-550 Kbytes. This exceeds 4 Mbits, and so one 4 Mbit DRAM will result in a shortage of capacity. For a screen of a PAL system having 352 pixels×288 lines, one frame corresponds to about 148.5 Kbytes of data. Thus, the same problem occurs.
For the above reason, the conventional video decoder
8
increases its memory capacity, for example, by adding one 1 Mbit DRAM to one 4 Mbit DRAM. To make an apparatus smaller and less expensive, it is desirable to use only one 4 Mbit DRAM. To solve this problem, the following method, for example, based on the
Britton Howard
Fish & Richardson P.C.
Rohm Co. Ltd
LandOfFree
Image data decoding method and apparatus using memory for... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image data decoding method and apparatus using memory for..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image data decoding method and apparatus using memory for... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2469141