Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2001-09-26
2003-12-23
Mizrahi, Diane D. (Department: 2175)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000, C707S793000
Reexamination Certificate
active
06668258
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to the scaling of the indexing data of a multimedia document.
The field of the invention is that of the storage of and/or searching for digital multimedia documents in a database.
Prior to being stored in a database, a multimedia document is indexed. These indexing data pertaining to details such as the author, type, date of creation, a summary, descriptions, etc. are associated with the document. They are used especially to sort out the document by author, descriptor, etc. and/or classify the documents and/or of course search for a document, generally with a view to accessing the found document.
The volume of documents to be stored and therefore the associated indexing data is increasing every year. According to certain sources, it is thought that the storage requirements of certain companies are more than doubled every year.
2. Description of the Prior Art
Encoding methods have been developed, designed to reduce the volume of data to be stored.
Certain encoding methods relate to documents: they reduce the volume of the document itself while providing for its total or almost total restitution by a decoding method corresponding to the encoding method used.
Other methods relate to the scaling of the indexing data of a multimedia document. This scaling is aimed of course at reducing the volume of these indexing data but it is not indispensable for it to restitute these indexing data. It should be possible, with the scaled data, for example to sort out multimedia documents and/or to classify and/or search for a multimedia document by making comparisons between these indexing data and the indexing data of the documents of the database.
The reduction of the volume of the indexing data obtained by scaling at a given time may prove to be subsequently insufficient, especially when the volume of initial data increases and/or it is sought to obtain a higher scale ratio.
It is then possible to rescale the initial data. This is a painstaking and even impossible operation when the data are inaccessible. It is also possible to redo a scaling operation on these already scaled data and another scaling operation on the new (as yet non-scaled) data. These data, thus scaled according to different scaling operations, generally result in a disparity between the scale ratios and between the scaled data. The sorting or searching operations performed on these indexing data thus scaled cannot be performed on all the scaled data but on each category of scaled data.
It is an aim of the invention to avoid these disparities.
SUMMARY OF THE INVENTION
An object of the invention is a method for the scaling of the indexing data of a multimedia document such that the scale ratio of the scaled data may be subsequently increased to make these data more compact, with a guarantee that the data thus obtained will be equivalent to the data obtained by applying a higher scale ratio to the initial data at the very outset.
When, for example, the initial data are initially scaled in groups of size N′ and the scaled data are scaled once again in groups of size N″, the resulting data are equivalent to the data obtained by the scaling of groups of initial data of size N′N″.
More generally, if the initial data are initially scaled in groups sized n′
J
, with j=1 to J to obtain scaled data d′
J
and if these scaled data are rescaled in groups D′
k
sized J
k
with k=1 to K, the resulting scaled data d″
K
are the same as they would be if the scaling were to be done directly on groups of initial data sized n″
k
, n″
k
being the sum of the n′
j
values of the group D′
k
.
An object of the invention is a method for the scaling of indexing data D=(d
n
, n=1 to N) of a multimedia document wherein mainly the method comprises the following steps which consist:
a) at the time t, in grouping the data D in distinct and consecutive groups D
j
respectively sized n′
j
, j varying from 1 to J and respectively scaling each group D
j
to a value d′
j
according to at least one determined scaling method C, and in storing the data D′=(d′
j
, j=1 to J) thus obtained.
b) subsequently, at the time t′>t, when the number of data resulting from the previous scaling operation is too great, in grouping the data D′ in distinct and consecutive groups D′
k
respectively sized J
k
, k varying from 1 to K and scaling each group D′
k
by a value d″
k
according to a rescaling method C′ compatible with the scaling method C in such a way that the data d″
k
are equivalent to those obtained by applying the scaling method C directly to distinct and consecutive groups of data of D sized n″
k
, n″
k
being the sum of the n′
j
values of the group D′
k
and in storing the scaled data D″=(d″
k
, k=1 to K).
According to one characteristic of the invention, the step b) is reproduced using, for D′, the data resulting from the last rescaling operation.
According to another characteristic of the invention, with the data d′
j
, there are associated the sizes n′
J
and/or with the data d″
K
, there are associated the sizes n″
k
.
According to another additional characteristic:
each datum d
n
is weighted by a weight w
n
,
each datum d′
j
is weighted by a weight w′
j
, each of these weights being equal to the sum of the weights of the corresponding data of the groups D
j
,
each datum d″
k
is weighted by a weight w″
k
, each of these weights being equal to the sum of the weights of the corresponding data of the groups D′
K
,
and the weight of each datum is associated with said datum.
Prior to the storage of the scaled data, a header comprising at least one label specifying the scaling method may be associated with the scaled data. The header furthermore advantageously comprises the number of data before and/or after the encoding.
According to one embodiment of the invention, the determined scaling method C is the method C
7
based on the histogram of the data groups D
j
according to predefined categories and the rescaling method C′ is the method C
7
′ based on the computation of the sum, term by term, of groups of histograms of D′.
According to one characteristic of the invention, the data D are series of scalar values or vectors.
According to another embodiment of the invention, the determined scaling method C is the method C
4
which consists of the random choice of a data from each group of data D
j
and the determined rescaling method C′ is then the method C
4
′ which consists of the random choice of one datum among each group of data D′
k
.
According to another characteristic of the invention, the data d″
k
are equal to those obtained by applying the scaling method C directly to distinct and consecutive groups of data D respectively sized n″
k
.
According to various embodiments of the invention:
the determined scaling method C is the method C
3
based on the computation of the mean of the groups of data D
j
and the rescaling method C′ is the method C
3
′ based on the computation of the mean of each group of data D′
k
, or
the determined scaling method C is the method C
1
based on the computation of the minimum of each group of data D
j
and the rescaling method C′ is the method C
1
′ based on the computation of the minimum of each group of data D′
k
, or
the determined scaling method C is the method C
2
based on the computation of the maximum of each group of data D
j
and the rescaling method C′ is the method C
2
′ based on the computation of the maximum of each group of data D′
k
, or
the determined scaling method C is the method C
5
based on the choice of the first datum from each group of data D
J
and the rescaling method C′ is the method C
5
′ based on the choice of the first d
Mizrahi Diane D.
Patterson, Thuente, Skaar & Christensen, L.L.C.
Punit Prakash C.
LandOfFree
Method for the scaling of the indexing data of a multimedia... 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 for the scaling of the indexing data of a multimedia..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for the scaling of the indexing data of a multimedia... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3140068