Image analysis – Image compression or coding – Interframe coding
Reexamination Certificate
1998-12-22
2002-01-15
Couso, Jose L. (Department: 2621)
Image analysis
Image compression or coding
Interframe coding
Reexamination Certificate
active
06339656
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus for performing moving picture encoding decoding processing, especially an apparatus which uses a motion vector for the encoding process.
2. Description of the Related Art
Recently, the technical development of the digital telecommunication system has progressed remarkably, and various technologies such as multimedia communication and the personal mobile communication are discussed and researched actively. Various types of data are used and transmitted in such multimedia communication including voice data, moving picture data, and still image data. Therefore, high speed communication technology for picture data will be a key and important technology in the forthcoming multimedia communication system. Because the data amount of image data is huge, the data encoding decoding technology is an important technology. As for the encoding decoding algorithm, there are MPEG1, MPEG2, H.261 and H.263 in ITU-T etc. Standardization of those algorithms is discussed and determined according to the usage and purpose.
In these encoding decoding algorithms, an intraframe encoding process including orthogonal transformations and an interframe encoding process including motion compensation are performed. Through those encoding decoding processes, a process that relates to the motion vector detection occupies the major part of the whole processing in the algorithm. The data encoding decoding using the motion vectors is indispensable processing for the improvement of the picture quality. Interframe encoding first calculates a data difference between former image data and the current input image and performs encoding using the calculated interframe difference (as shown in
FIG. 16
, data difference
1
is calculated based on a former image data al and an input image data b). If a data difference becomes small, encoding efficiency becomes large. However, when image data with much motion is compressed, the interframe difference becomes large. In this case, use of interframe difference processing will not contribute to improve the encoding efficiency very much. Then, the encoding efficiency can be improved by performing motion compensation preliminarily, where former image data is moved based on the forecasting of the movement of former image data, and then performing interframe data difference processing (as shown in
FIG. 16
, difference
2
is calculated based on an amended data a
2
from former image with motion compensation and an input current image data b). This movement direction with forecasting image data motion is defined as a motion vector.
Hereafter, the outline of the motion vector detection is shown below. The compressed algorithm performs encoding process by a macroblock unit which composes 16 pixels×16 pixels matrix data, and the motion vector is detected by the macroblock unit. The macroblock for calculating motion vector is moved around by a pixel length step in the search area (FIG.
17
(
a
)). Next, the correlation of the moved macroblock and the corresponding macroblock of the input image is calculated. The correlation is calculated as the absolute value harmony of the respective difference of the corresponding pixel data.
The absolute value harmony of the difference of the macroblock is calculated respectively after moving in all the search areas, and the maximum correlation is detected among the calculated correlation, and the amount of the movement of the detected moved macroblock is assumed to be a motion vector.
However, if for example macroblock
200
located on the boundary of the frame comes to the target macroblock for detecting motion vector as shown in
FIG. 17
(
b
), the search area
210
for the macroblock
200
covers the area beyond the frame. In a conventional encoding process, such a motion vector was supposed not to be handled. When an image picture moves in the direction where actual image data go outside of the frame, motion compensation cannot be used for the interframe encoding. So the encoding efficiency has dropped remarkably. Moreover, the picture quality might worsen as a result due to the noise such as block distortions. Then, unrestricted motion vector is added as an option in H.263 etc. in order to raise accuracy. The unrestricted motion vector is used for calculating the motion vector when actual image data go outside of the frame, assuming that whole data of the search area
210
exist by providing pixel data to the vacant area of the search area
210
outside of the frame. When correct image data for outside area cannot be obtained from the frame data, the data of frame boundary is assumed to be the nearest data. As for the implementation method for achieving the above mentioned unrestricted motion vector calculation processing, the following implementation methods normally can be considered. The first implementation method is the software processing without dedicated hardware such as LSI. According to the flow chart as shown in
FIG. 4
, with the pixel data moving, the address after moving is determined to be within the frame or outside the frame by adding or subtracting moving distance to the address before moving. If the address after moving is outside the frame, the pixel data difference is processed based on the nearest boundary pixel data of the frame and the input current pixel.
The second implementation method is the hardware by expanding the set frame memory area in order to cover all search area composed of the frame data area and the outer surrounding area, and all data of the search area which are the frame data and the outer surrounding area data are transmitted precedingly, and correlation processing is performed based on the data on the expanded frame memory and the input data.
However, according to the above mentioned first implementation by software means, when motion vector calculation process for the macroblock, the conditional branch processes are included in all calculation every after moving around in the search area, the processing becomes complex. In addition, if a processor uses the pipeline processing, the so called “pipeline penalty” is caused when the conditional branch process occurs and the processing efficiency deteriorates greatly.
Moreover, according to the above mentioned second implementation by hardware means of expanding memory area, the necessary memory capacity consumption becomes large.
SUMMARY OF THE INVENTION
Therefore, with the foregoing in mind, it is an object of the present invention to provide a moving picture data encoding decoding processing apparatus which can detect the unrestricted motion vector with a dedicated memory having a small capacity without increasing the software process.
In order to achieve the objects, a moving picture encoding decoding processing apparatus includes a frame memory, a prediction memory for storing the image data of a search area for a motion vector detection, an image data transmission means for transmitting image data of a first macroblock and surrounding image data existing in the frame memory from the frame memory to the prediction memory and writing the transmitted data to the corresponding area wherein the first macroblock is the macroblock whose search area extends over the frame image area, an image data extrapolation means for forming an image data to the extrapolation area of the search area where an image data had not been written onto the prediction memory by extrapolation based on the pixel data located on the boundary with the search area, and a motion vector detecting part for detecting the unrestricted motion vector based on the image data of the search area formed on the prediction memory.
This embodiment makes it possible to form the image data for the search area onto the prediction memory rapidly and efficiently and to detect unrestricted motion vector. The image data for the search area in the prediction memory is formed by transmitting and writing the first macroblock currently processed as a target for encoding and surrounding image data of th
Couso Jose L.
Matsushita Electric - Industrial Co., Ltd.
Merchant & Gould P,C,
LandOfFree
Moving picture encoding decoding processing apparatus does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Moving picture encoding decoding processing apparatus, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Moving picture encoding decoding processing apparatus will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2817850