EP1500246A1

EP1500246A1 - Automated adaptation and transformation of media flows

Info

Publication number: EP1500246A1
Application number: EP03722289A
Authority: EP
Inventors: Christian Timmerer; Harald Kosch; Hermann Hellwagner; Andreas Hutter; Jörg Heuer
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 2002-04-26
Filing date: 2003-04-17
Publication date: 2005-01-26
Also published as: AU2003229522A1; DE10218813A1; WO2003092238A1; DE10218813B4

Abstract

According to the invention, when a first data flow is transformed into a second data flow, a first data flow description for the first data flow is simultaneously transformed into a first data flow description for the second data flow.

Description

description

Automated adaptation and transformation of media flows

The invention relates to the transformation of data streams.

Devillers S .: "Bitstream Syntax Definition Language: An Input to MPEG-21 Content Representation" describes a method with which adaptation processes for given media streams can be carried out on the basis of a BSDL (Bitstream Description Language) language for describing data streams. For this purpose, transformation descriptions are used, which can, for example, be written in the language XSL (Extensible Stylesheet Language). XSL transformations are explained in "XSL Transformations (XSLT)" Version 1.0, W3C Communication dated November 16, 1999. The transformation descriptions are used to incorporate an XML document that conforms to a media stream-specific BSDL version, that is, a schema according to the "XML Schema Language" 3C Recommendation of May 2, 2001 or a DTD (Document Type Definition) transform adapted XML document. This document is then used in turn to generate a transformed data stream from a first data stream with the help of the BSDL execution.

This sequence is explained in more detail with reference to FIG. 1. A first data stream a_l in the form of a binary bit stream is fed to a first bit stream parser BP_1. Using a BSDL scheme BSDL_a, the bitstream parser BP_1 generates a first data stream description BSD_1 / 1 for the first data stream in the form of an XML document from the first data stream a_l. An XSL processor XSLP then generates a second data stream description BSD_l / 2 for the first data stream from the first data stream description BSD_1 / 1 for the first data stream using an XSLT style sheet XSL. A second bitstream parser BP_2 finally transforms the first data stream a_l using the BSDL scheme BSDL_a and the second data stream description BSD_l / 2 for the first data stream in the second data stream a_2.

This method can be used, for example, to adapt an instance of a media stream, for example a video stream, present on a media server to the requirements of a requesting client such that, after the adaptation, for example, the image format of the video stream corresponds to the screen resolution available at the terminal or, for example, that on the Transmission path from the server to the client's preferred bit rate is optimally used.

However, the method described has some disadvantages. For the adaptation, a parser must be available that knows the bit stream syntax of the codec used to encode the data streams a_l or a_2.

Furthermore, a specific BSDL version, e.g. as a schema or as a DTD, exist and be known to the system.

Finally, the second data stream description BSD_l / 2 for the first data stream is not a correct description of the second data stream a_2, because the data stream addresses contained in the second data stream description BSD_l / 2 for the first data stream for the elements in the second data stream description BSD_l / 2 for the first data stream were copied directly from the first data stream description BSD_1 / 1 for the first data stream without, for example address changes caused by discarding parts of the data stream.

The object is therefore to provide a new, improved transformation which also fulfills the functions of the method according to the prior art, but avoids the disadvantages described. This object is achieved by the inventions specified in the independent claims. Advantageous configurations result from the subordinate claims.

In a method for transforming a first data stream into a second data stream, the first data stream contains partial data stream areas that can be referenced and / or classified. The second data stream also contains partial data stream areas that can be referenced and / or classified. A data stream description is provided for the first data stream, in which at least some, preferably at least almost all, of the data stream subareas in the first data stream are referenced and / or classified. For this purpose, the data stream description can contain a series of address details, each of which references a sub-area of the data stream. Classification details are assigned to the address details, by means of which the respective sub-areas referenced by the address details are classified. The first data stream is transformed into the second data stream by a data stream transformation. For example, as in the case of the methods described in the prior art, contents can be filtered out, an image resolution reduced and / or the order of the data in the data stream can be rearranged. A first data stream description for the second data stream is generated, in which at least some, preferably at least almost all, of the data stream subareas in the second data stream are referenced and / or classified. In the first data stream description for the second data stream, in particular the references regarding the second data stream are correct.

By generating a first data stream description for the second data stream, a parser can be omitted in subsequent transformations. Insofar as a parser is not necessary for the provision of the first data stream description for the first data stream, for example, if it is provided by an encoder, the Procedures completely without the use of a parser for the data stream. This is an enormous advantage, especially in networks and when using data streams in different formats, since it is often not possible to provide a suitable parser here.

The first data stream description for the first data stream is preferably transformed into the first data stream description for the second data stream by a data stream description transformation. The first data stream description for the second data stream is thus generated from the first data stream description for the first data stream.

The generation of the first data stream description for the second data stream can even take place parallel to the data stream transformation and also in the same process step. In the case of the data stream description transformation described above, the data stream description transformation then takes place parallel to the data stream transformation. The parallel success stands in contrast to a sequential success, i.e. that the generation of the first data stream description for the second data stream can be carried out independently of the data stream transformation and vice versa.

The first data stream is preferably transformed into the second data stream using the first data stream description for the first data stream.

It is also preferred that the first data stream is transformed into the second data stream using a data stream transformation description.

Analogously, the first data stream description for the first data stream can be transformed into the first data stream description for the second data stream using a data stream description transformation description. Then in particular the data stream transformation can be described. exercise is the same as the data stream description transformation description.

At least one of the data streams is preferably a data record, a bit, media, audio, still picture and / or video stream. The data stream is encoded in particular in the MPEG-4 or JPEG2000 standard.

The data stream description is especially in XML (Extensible Markup Language). The data stream transformation description and / or the data stream description transformation description is preferably a style sheet in XSL and / or in Javascript.

The method is particularly suitable for use cases in which further transformations follow the first transformation in order to further adapt the data stream in terms of content and / or scope. Accordingly, the second data stream is preferably transformed into a third data stream using the first data stream description for the second data stream.

A method in which the first data stream description for the first data stream functions as a look-up table for the data stream transformation is based on the same inventive idea as the previously described methods. In this case, the first data stream description for the first data stream is used directly, ie without the conversion known from the prior art, into a second data stream description for the first data stream for the data stream transformation. This saves the effort for converting the first data stream description for the first data stream into the second data stream description for the first data stream and the storage space for the second data stream description for the first data stream. The first data stream description for the first data stream is used completely and unchanged in that the data stream transforma- the references or classes of interest are looked up directly in the first data stream description for the first data stream.

A device which is set up to carry out a method of the type described above can be implemented, for example, by providing corresponding means for each of the method steps mentioned, which means carry out the method steps. Advantageous configurations of the device result analogously to the advantageous configurations of the method.

A program product for a data processing system, which contains software code sections with which one of the described methods can be carried out on the data processing system, can be implemented by suitable implementation of the method in a programming language and translation into code that can be executed by the data processing system. The software code sections are stored for this. A program product is understood to mean the program as a tradable product. It can be in any form, for example on paper, a computer-readable data medium or distributed over a network.

Further essential advantages and features of the invention result from the description of an embodiment with reference to the drawing. It shows:

FIG. 2 shows a method for transforming a first data stream into a second data stream.

FIG. 2 shows an encoder Enc which supplies a first data stream a_l and a first data stream description BSD 1/1 for the first data stream. The first data stream description BSD_1 / 1 for the first data stream, which describes the first data stream a_l and is compliant with a structural specification BSDL, is used in two ways: First, uses the first data stream description BSD_1 / 1 for the first data stream together with a data stream transformation description XSL_a in order to transform the first data stream a_l in the form of a media stream into a second data stream a_2 in the form of a media stream. The first data stream description BSD_1 / 1 is then transformed into a first data stream description BSD_2 / 1 for the second data stream, even under the guidance of a data stream description transformation description XSL_b. The data stream description transformation description XSL_b can advantageously be designed such that the data stream addresses in the first data stream description BSD_2 / 1 for the second data stream are correct with respect to the second data stream a_2. In this way, a further transformation, which accordingly contains further transformation descriptions in the form of data stream transformation descriptions and data stream description transformation descriptions, can be carried out again directly and according to exactly the same procedure without additional intermediate processing steps. Direct cascading of transformations is thus possible in a uniform manner. The structure specification for the BSDL schemes can correspond to the definition according to the prior art described at the outset or else advantageously correspond to the general data stream description described below.

This data stream description, that is to say the structure specification BSDL, is generic, that is to say it is independent of special coding formats, in particular independently of the coding format of the data stream described. Nevertheless, the elements can be typed in terms of specific coding formats. For this purpose, the data stream contains data stream subareas that can be referenced and classified. It also has a data stream syntax that is given by the format and / or the content of the data stream. At least some data stream subareas are referenced. This means that they are provided with a reference that clearly identifies them. bar makes. The most prominent example of such a reference is the description of the position of a data stream sub-area in the data stream by means of address data. The data stream subareas are also classified by a division into at least one class of a class division. By dividing them into a class, they are assigned a property that is predetermined by the class and characteristic of them. The classification is at least partially independent of the syntax of the data stream. This means that it does not follow the structure and syntax of the data stream to be described, but contains classes that are independent of the structure, in particular the coding format, of the data stream and are not related to the content.

A marker ms_l [x] is assigned to a data stream sub-area a_l [x] of the data stream a_l, which specifies the meaning of the data stream sub-area a_l [x] for the data stream a_l. This marker ms_l [x] does not have to be generic, but can be bitstream-specific, e.g. a VOP can be marked in an MPEG-4 video stream. This can be used to support the methods for transforming data streams described in the introduction. This has the advantage that the description of the data stream, in particular the class division, is generic and only the values of the markers ms_l [x] can vary, for example based on a classification scheme. The language can thus be parsed with a processor independent of the bit stream, regardless of the type of bit stream a_l.

The description of the data stream can have the following special properties:

- The data stream description assigns one or more markers m_l [i] to data stream subareas a_l [x] in the data stream a_l. The markers m_l [i] are used to translate the instructions of the transformation descriptions XSL onto the data stream. An interpretation of the meaning of a Markers m_l [i] and / or the data stream subarea a_l [x] for the data stream a_l are not necessary and are not necessarily present.

The same marker m_l [i] can be used for several data stream sub-areas, e.g. a_l [x] and a_l [y] are used if, for example, the different data stream subareas a_l [x] and a_l [y] belong to a semantic unit in the data stream a_l or comprise, for example, syntactic elements of the same type. A data stream subarea a_l [x] marked with the marker m_l [i] can contain data stream subareas a_l [z], which in turn are marked with markers m_l [j]. This is advantageous, for example, with regard to the speed of execution of the transformation if different transformation descriptions allow XSL to be carried out with different granularity. In one possible embodiment, these data stream subareas a_l [z] can be identified recursively on the basis of the identification of the data stream subarea a_l [x].

- A data stream sub-area a_l [x] can be assigned a value v [x] which corresponds, for example, to the value represented by a_l [x] in the bit stream. With a transformation, e.g. with XSLT, this value can be changed.

Referencing in the form of addressing the data stream subareas a_l [x] can be implemented in one of the following ways: Specification of the beginning and / or end of the data stream subarea a_l [x] by counting units from the beginning of the data stream.

Specification of the beginning and / or end of the data stream section a_l [x] by counting units from the beginning of the previous data stream section a_l [w]. - Specification of the beginning and / or end of the data stream section a_l [x] by counting units from the end of the previous data stream section a 1 [w]. Specification of the beginning and / or end of the data stream subarea a_l [x] by counting units from the start of the data stream subarea a_l [a] in which the data stream subarea a_l [x] is contained. - Specification of the end of the data stream section a_l [x] by counting units from the beginning of the data stream section a_l [x].

Without explicit specification of the beginning of the data stream subarea a_l [x], it can be specified that the beginning of the data stream subarea a_l [x] follows the end of the previous data stream subarea a_l [w] or corresponds to the beginning of the data stream subarea a_l [a], if the data stream subarea a_l [x] is first contained in the data stream subarea a_l [a]. The specification of these values can be done by counting units. One or more units, for example bits or bytes, can be used here.

Not all of the above properties have to be present in a description, but the structural elements that model these properties can also be used independently of one another.

The data stream transformation description XSL_a and the data stream description transformation description XSL_b can be generated in advance or can be generated automatically by a processing unit which, for example, takes additional transmitted information about the requesting terminal into account in a client-server scenario.

If the first data stream description BSDL_1 / 1 for the first data stream is not provided by the encoder Enc when the first data stream a_l is generated, it can be generated from the first data stream a_l using a parser in accordance with the prior art method. In the exemplary embodiment shown, the generation of second data stream descriptions for the first and / or second data stream is completely dispensed with.

The data stream transformation of the first data stream a_l into the second data stream a_2 using the first data stream description BSD_1 / 1 for the first data stream and the data stream transformation description XSL_a takes place in an XSL and data stream processor XBP.

The data stream description transformation of the first data stream description BSD_1 / 1 for the first data stream into the first data stream description BSD_2 / 1 for the second data stream takes place, in particular in parallel, in an XSL processor XSLP.

Data stream transformation and data stream description transformation can also be combined in one process.

The data stream transformation description XSL_a and the data stream description transformation description XSL_b can be combined in one document.

Claims

claims

1. Method for transforming a first data stream (a_l) into a second data stream (a_2), in which - the first data stream (a_l) contains data stream subareas which can be referenced and / or classified,

the second data stream (a_2) contains data stream subareas which can be referenced and / or classified,

a first data stream description (BSD_1 / 1) is provided for the first data stream, in which at least some of the data stream subareas in the first data stream (a_l) are referenced and / or classified,

- The first data stream (a_l) is transformed into the second data stream (a_2) by a data stream transformation, characterized in that a first data stream description (BSD_2 / 1) is generated for the second data stream, in which at least some of the data stream subareas in the second data stream (a_2 ) are referenced and / or classified.

2. The method according to claim 1, characterized in that the first data stream description (BSD_1 / 1) for the first data stream is transformed by a data stream description transformation into the first data stream description (BSD_2 / 1) for the second data stream.

3. The method according to at least one of the preceding claims, characterized in that the generation of the first data stream description (BSD_2 / 1) for the second data stream takes place parallel to the data stream transformation.

4. The method according to at least one of claims 1 and 2, characterized in that the generation of the first data stream description (BSD_2 / 1) for the second data stream and the data stream transformation take place in a common process.

5. The method according to at least one of the preceding claims, characterized in that the first data stream (a_l) using the first data stream description (BSD_1 / 1) for the first data stream is transformed into the second data stream (a_2).

6. The method according to at least one of the preceding claims, characterized in that the first data stream (a_l) is transformed into a second data stream (a_2) using a data stream transformation description (XSL_a).

7. The method according to at least claim 2, characterized in that the first data stream description (BSD_1 / 1) for the first data stream is transformed into a first data stream description (BSL_2 / 1) for the second data stream using a data stream description transformation description (XSL_b) ,

8. The method according to at least claims 6 and 7, characterized in that the data stream transformation description (XSL_a) is at least substantially the same as the data stream description transformation description (XSL_b).

9. The method according to at least one of the preceding claims, characterized in that at least one of the data streams (a_l, a_2) is a bit, media, audio, image and / or video stream.

10. The method according to at least one of the preceding claims, characterized in that at least one data stream description (BSD_1 / 1, BSD_2 / 1) is at least partially in XML.

11. The method according to at least claim 6 or 7, characterized in that the data stream transformation description (XSL_a) and / or the data stream description transformation description (XSL_b) is written in XSL and / or in a script language.

12. The method according to at least one of the preceding claims, characterized in that the second data stream (a_2) is transformed into a third data stream using the first data stream description (BSD_2 / 1) for the second data stream.

13. The method, in particular according to at least one of the preceding claims, for transforming a first data stream (a_l) into a second data stream (a_2) in the

the first data stream (a_l) contains data stream subareas that can be referenced and / or classified, - the second data stream (a_2) contains data stream subareas that can be referenced and / or classified,

a first data stream description (BSD_1 / 1) is provided for the first data stream, in which at least some of the data stream subareas are referenced and / or classified to the first data stream (a_l),

the first data stream (a_l) is transformed into the second data stream (a_2) by a data stream transformation,

the first data stream (a_l) is transformed into the second data stream (a_2) using the first data stream description (BSD_1 / 1) for the first data stream, characterized in that that the first data stream description (BSD_1 / 1) acts as a look-up table for the first data stream for the data stream transformation.

14. The device, which is set up to carry out a method according to at least one of the preceding claims.

15. Program product for a data processing system, which contains software code sections with which a method according to at least one of claims 1 to 13 can be carried out on a data processing system.