Image analysis – Image compression or coding
Reexamination Certificate
1999-06-30
2003-02-25
Patel, Jayanti K. (Department: 2721)
Image analysis
Image compression or coding
C707S793000
Reexamination Certificate
active
06526171
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a technology to encoding images such as mobile pictures, and in particular, to an image object managing method of managing objects configuring images by use of identifiers, an image processing apparatus using the method, an image stream produced by the method and a recording media for the stream, and a recording media on which programs achieving the method is recorded.
2. Description of the Related Art
Heretofore, there have been proposed a large number of methods of encoding or coding mobile pictures. One of these methods which is standardized by a moving picture coding experts group (MPEG) has been broadly employed in the present stage of art. The MPEG standard specifications include MPEG1 and MPEG2 at present. While MPEG1 is adopted for video compact disks (CD) and MPEG cameras, MPEG2 is utilized for digital video disks (VD) and digital satellite broadcasting. As general features of the MPEG standards, there is included data compression according to a correlation with respect to time, which leads to high coding efficiency. In this compression method, differences between a plurality of frames (screen images constituting a mobile picture) which are continuous with respect to time are recorded to compress data. In general, adjacent frames of a mobile picture are quite similar in image features to each other and hence a high compression ratio can be obtained by coding the difference therebetween. This is why the compression efficiency is improved.
New specifications of MPEG4 are being prepared or discussed at present. This is developed inherently for communication and is different in features from MPEG1 and MPEG2. Description will be, however, given of only sections of MPEG4 which relate to the present invention. MPEG4 is conspicuously different from MPEG1 and MPEG2 in an object encoding operation. According to MPEG4, each object appearing on a screen can be encoded. Consequently, it is possible that a person (an assassin)
1
and a background
2
of one film scene are respectively encoded to be also respectively transmitted, and it is possible, on a side having received the respectively transmitted signals, to combine the images into one scene. In this operation, each object is encoded, for example, as follows. Although a little change appears between successive scenes in actual mobile pictures, a large change takes place between scenes in this example for easily understanding in the description of this application.
First, an image of a person is obtained before a single-color background in a studio using, for example, a method called “blue back”. Thereafter, a rectangle in which the person's image is completely contained is defined in the overall image and is trimmed as shown in FIG.
8
. Using as a key the single color of the background, a mask is produced to separate the person from the background (FIG.
9
).
FIG. 8
is compressed by a discrete cosine transform (DCT) which is similar to ordinary MPEG2. The mask, i.e.,
FIG. 9
is similarly compressed. Compressed data of
FIGS. 8 and 9
is transmitted to a partner (on the receiving side). The receiver decodes the data of
FIGS. 8 and 9
and then recognizes that a white area of
FIG. 9
is “transparent” to set the background area of the decoded
FIG. 8
to “transparent” in accordance with the mask of
FIG. 9
(reference is to be made to FIG.
10
). Thereafter, the obtained image is combined with a background image separately prepared.
This method has a feature of higher encoding efficiency when compared with a method in which the overall screen is encoded. This is because of a principle that the background screen is almost still, i.e., little change, and hence the quantity of data to be processed is small, namely, only the moving sections of the person are to be encoded.
The present invention provides, in consideration of image compression technologies such as MPEG4 having the feature above, novel characteristics to achieve an object-base encoding operation.
A video image stream of MPEG4 includes a stream of backgrounds and streams of respective objects.
FIG. 4
shows an example of MPEG4 streams corresponding to
FIGS. 1
to
3
.
A stream
10
includes individual object streams
11
to
19
and a control stream
20
describing a composite rule of these object screams (the description stipulates, for example, positions of images with respect to depth in the screen image and timing of appearance of images).
In
FIG. 1
, an assassin
1
and a background
2
are related to object streams
11
and
12
, respectively. In
FIG. 2
, a target
3
, a background
4
, and a hindrance
5
are associated with object streams
13
to
15
, respectively. In
FIG. 3
, a target
6
, a background
7
, an assassin
8
, and a hindrance
9
are related to object streams
16
to
19
, respectively. Object streams
13
and
16
,
15
and
19
, and
14
and
17
are not interrupted therebetween and are hence respectively continuous object streams, i.e., each combination forms one object stream.
Stream
10
is subdivided in a time division procedure into small packets to be transmitted. Consequently, for the receiving side of these object streams (and the control stream) to restore the original streams, there are required identifiers to identify the respective object streams. According to the stipulation of MPEG4, only the number of bits is determined for the assignment of identifiers and no other rules are stipulated for the identifiers. Therefore, for the identifiers, serial numbers are ordinarily assigned in an order of appearance of objects. To guarantee the time sequence of the objects, a time stamp is assigned to each object stream.
Even in a case in which objects which regarded as the same object by viewers, for example, objects corresponding to an identical person appear in different scenes, if the pertinent stream is once interrupted, a subsequent scream is ordinarily assigned with a new serial number. Namely, another number is assigned to the subsequent stream. In consequence, when at least the stream is simply analyzed by a computer, a correlation, i.e., the person appearing in two or more scenes cannot be appropriately identified.
Consequently, the prior art is attended with problems as follows.
An amateur can produce copies of video images only for curiosity or entertainment to distribute the copies via a network. Moreover, it may also be possible that a malicious person who aims at disgracing dignity of a particular person copies images of the person appearing in video images and combines the copied images with another background and other objects to produce original video images. There exists fear that such an act infringes the right of portraits of the person. Additionally, the produced screen images are not associated with intention of the producers or programs or films. Namely, there also exists fear of an infringement of a copyright. Particularly, when the user of the right is to be charged, there possibly occurs a matter of money.
When the video images are processed in the analog format or in MPEG2 and preceding digital formats, these actions do not easily occur in general because of difficulty in separating a person from a background thereof. However, since object streams can be separated in accordance with MPEG4 under discussion, the problem above may frequently arises. Therefore, a producer who provides video images in a format in which the encoding of images is carried out for objects as in MPEG4 is required to manage the copyright more strictly than in the job in which video images are encoded in MPEG or any preceding format. For this purpose, it is necessary to obtain streams of video images available in the market so as to determine whether or not particular objects exist therein. In the prior art, only the object identifiers can be used as information to identify associated objects. The identifiers are serial numbers assigned in a sequence of appearance of video images and are not associated with the contents of the objects. Namely
Antonelli Terry Stout & Kraus LLP
Hitachi, ltd.
Patel Jayanti K.
LandOfFree
Image object managing method, an image 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 Image object managing method, an image processing apparatus..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image object managing method, an image processing apparatus... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3172998