FR2911233A1 - Data flow decoding method for e.g. video on demand service, involves determining capacity to decode each data packets using information representative of scalability class and without reading slice header of data packets - Google Patents

Abstract

The method involves reading an information representative of a scalability class applied to data packets of a data layer using a reading unit, where the information can be an information representative of a spatial scalability category such as spatial scalability category with determined radio. The information is read in each of a set of signaling packets. A capacity to decode each of the data packets is determined by a determining unit using the information representative of the scalability class and without reading the slice header of the data packets. Independent claims are also included for the following: (1) a device for decoding a data flow representing an image or a video sequence of images (2) a downloadable computer program product comprising a set of instructions for performing a data flow decoding method (3) a method for coding a data flow representing an image or a video sequence of images (4) a device for coding a data flow representing an image or a video sequence of images (5) a downloadable computer program product comprising a set of instructions for performing a data flow coding method (6) a signal representative of a data flow representing an image or video sequence of images (7) an information medium comprising a signal representative of a data flow representing an image or video sequence of images.

Description

Methods and devices for encoding and decoding a data stream

  scalable taking into account a scalability class, computer program products, signal and corresponding data medium. FIELD OF THE INVENTION The field of the invention is that of encoding and decoding images or video sequences of images. More specifically, the invention relates to a technique for encoding images or video sequences of images generating a stream having a data layer organization, and decoding of such scalable data stream (or scalable), quality adaptable and variable spatio-temporal resolution. 2. Prior art 2.1 General principle of scalable video coding Many data transmission systems are today heterogeneous, in that they serve a plurality of clients having very different types of data access. Thus, the global Internet network, for example, is accessible both from a terminal of the personal computer (PC) type and from a radiotelephone. More generally, the bandwidth for access to the network, the processing capabilities of the client terminals, the size of their screens vary greatly from one user to another. Thus, a first customer can for example access the Internet from a powerful PC, and have an ADSL (Asymmetric Digital Subscriber Line) for 1024 kbit / s while digital subscriber line a second client seeks to access the same data at the same time from a PDA terminal (Personal Digital Assistant) connected to a low-speed modem. Most video coders generate a single compressed stream corresponding to the entire coded sequence. Thus, if several clients wish to use the compressed file for decoding and visualization, they will have to download (or stream) the complete compressed file.

  It is therefore necessary to provide these various users with a data stream that is adapted both in terms of bit rate and resolution of the images to their different needs. This need is more widely needed for all applications accessible to customers with a wide range of access and processing capabilities, including video-on-demand (VOD) applications for video-on-demand. the card), accessible to universal mobile telecommunication service (UMTS) terminals, PCs or television terminals with ADSL access, etc. ; session mobility (for example resumption on a PDA of a video session started on a television, or, on a UMTS mobile of a session started on GPRS (General Packet Radio Service)); continuity of session (in a context of bandwidth sharing with a new application); high-definition television, in which a single video encoding must be able to serve both clients having a standard definition SE) and customers having an HD terminal; - videoconferencing, in which a single encoding must meet the needs of customers with UMTS access and Internet access; etc.

  To meet these different needs, scalable or scalable image coding algorithms have been developed, allowing for adaptable quality and variable spatio-temporal resolution. According to these techniques, the encoder generates a compressed stream having a hierarchical layered structure, in which each of the layers is nested in a top layer. For example, a first data layer conveys a 256 kbit / s stream, which can be decoded by a PDA type terminal, and a second complementary data layer conveys a resolution stream greater than 256 kbit / s which can be decoded, complement the first, by a more powerful terminal type PC. The rate required for the transport of these two nested layers is in this example 512 kbit / s. Such coding algorithms are thus very useful for all applications for which the generation of a single compressed stream, organized in several scalability layers, can serve several clients of different characteristics.

  Some of these scalable video coding algorithms are now being adopted by AVC (Moving Video Expert Group) Advanced Video C'oding (MPEG-4) as part of the JVT Working Group. Joint Video Team between the International Telecommunication Union (ITU) and the International Organization for Standardization (ISO). Amendment 3 of the aforementioned AVC H264 / MPEG-4 standard (part 10) describes in particular the scalable video streams, denoted SVC (Scalable video coding).

  A stream generated by an SVC type encoder comprises a first layer and optionally upper layers. Such a stream includes a base level, also called base level, compatible with the AVC standard, and one or more enhancements, distributed in one or more layers. These enhancements can be enhancement type quality, temporal enhancement, or spatial enhancement. A spatial enhancement involves a layer change. For example, Figure 1 schematically illustrates the structure of a quality layer 11 of an SVC type data stream. Such a layer 11 comprises, for example, a level of quality in basic quantization 121, and two levels of quantization enhancement quality 122 and 123. For example, the basic quality level 121 corresponds to a stream of the AVC type. This layer 11 can also be divided into time levels, for example the three levels 131, 132 and 133.

  It is thus recalled that a level of enhancement in quality of quantization, also called SNR (in English signal to noise ratio) corresponds to an enhancement of the quality of the images inside a spatio layer. -temporelle. A spatial enhancement corresponds to a change in spatial resolution between two layers. Finally, a temporal enhancement corresponds to a set of information associated with one or more higher temporal levels, this information being able to be located in the current layer or even in a higher layer. It is also recalled that an enhancement in quantization quality can be coded according to two distinct types of coding: a first type of quantization quality coding implementing a progressive coding of the data, denoted FGS for Fine Grain Scalability; a second type of quantization quality coding implementing a non-progressive coding of the data, denoted CGS for Coarse Grain Scalability, where the FGS coding, more complex than the CGS coding, allows a greater flexibility in the adaptation of the data. flux. On the other hand, a spatial enhancement can be coded only by a CGS-type coding via a new spatio-temporal layer.

  For example, we can define different categories of spatial resolution changes, resulting from the complexity of filtering implied by the ratio, in increasing complexity: spatial scalability without the use of inter-layer prediction, spatial scalability (with inter-layer prediction) with a ratio 2, corresponding to a change in dyadic resolution; spatial scalability (with inter-layer prediction) with a 3/2 ratio; spatial scalability (with inter-layer prediction) with any other ratio.

  Finally, a temporal enhancement can be created by adding intermediate images between the previously coded images within the current layer, or even the previous layer. This scalability is inherent to the temporal prediction management system in SVC. 2.2 Concept of profiles and levels of AVC In most standards of video compression, it is defined a notion of profile (or profile in English) and level (or level in English) attached to a video stream. Specifically, a profile defines a subset of standard tools that can be used to encode a video stream. A decoder said to be compatible with this profile must be able to decode this video stream. Thus, the activation of tools according to a profile makes it possible to control the implementation complexity of an encoder / decoder according to the targeted architecture, and to adapt the tools necessary for a given application. For example, stroke appendix A as presented in document JVT-T201 by T. Wiegand, G. Sullivan, J. Reichel, H. Schwarz and M. Wien (July 15-21, 2006, Klagenfurt, Austria). The meeting, JD Joint Video Team of ISO / IEC MPEG and ITU-T VCEG (ISO / IEC JTC1 / SC29 / WG11 and ITU-T SG16 Q.6)), defines different profiles including: a basic profile, noted baseline, mainly intended for videophone, videoconferencing or mobile use applications, using tools of low complexity, as well as some tools allowing a better control of the errors; an average profile, noted hand, and a high profile, rated high, mainly intended for broadcast applications of TV type or full high definition signals, using more complex tools, or more suitable for higher resolutions. In addition, the activation of the FGS coding tool for the coding of quality enhancement levels depends on the definition of the profiles in the SVC standard. Indeed, not all profiles allow FGS coding. The concept of level of complexity allows for itself, inside a profile, to refine the measure of complexity (or "level of complexity") necessary for a decoder to decode a stream of a given profile. . Typically, each level of complexity can be attached to a number of decodable macroblocks per second, to a maximum image size, or to a maximum size of the decoding buffers. This level of complexity also makes it possible to activate certain tools of the standard, dependent on the level of complexity. For example, as shown in Table A-1 (Table A-1 to Main profile level limits) of the JVT-T201 cited above, it is possible to activate the interleave tool in the main profile from a level of complexity greater than 2.1 and less than 4.2; in this case, an indicator (for example the frame_mbs_only_flag flag) is 1 for complexity levels 2.1 to 4.2. 2.3 Signaling of profiles and levels of complexity It is recalled that a stream originating from an AVC encoder is generated for a given spatial resolution and for a given quality. Such a stream therefore comprises a single layer with a single level of quality in quantification. It is known to insert a profile and complexity level information in such a stream, before transmission, to inform a decoder receiving the stream of the complexity of this stream. Thus, the decoder knows whether it is capable of decoding the stream, that is to say if the decoder is compatible with the same profile and the same level of complexity as those indicated in the AVC stream. This information is conventionally inserted in a data unit called Sequence Parameter Set (SPS).

  In an SVC stream from an SVC type encoder, profile and complexity level information is conventionally inserted into each of the stream layers in an SPS data unit. In other words, an SPS unit is generated for each spatial layer of the stream. An SVC stream is organized in AUs ("Access Units" for Access Units) each corresponding to a given time instant and comprising one or more NALUs ("Network Abstraction Layer Units" for Network Access Units), or packets of data. Each NALU is associated with an image resulting from spatio-temporal decomposition, at a spatial resolution level, and at a quantization quality level. This structuring in NALUs enables the SVC decoder to be able to adapt in bit rate and / or space-time resolution by eliminating the NALUs of spatial resolutions, time frequency or encoding quality that are too high. Currently, in SVC, two sets of NALUs coexist: NALUs AVC (for the base level of an SVC layer): each AVC-specific NALU has a single header octet; N NALUs SVC (for SVC layer enhancement levels or SVC layer base level above the first layer): each SVC specific NALU has one AVC header byte and three SVC header bytes , including in particular the fields (P, D, T, Q) ("Priority_id", "Dependency_id", "Temporal_level", "Quality_level") necessary for filtering the information contained in the NALU. Specifically: • the field "Priority_id" indicates a priority level of a NALU that can be used to guide a quality adaptation; • The "Dependency_id" field allows to know the spatial resolution of a hierarchical coding layer. This field can also control a level of enhancement in quantization quality or time enhancement in the context of a layered coding; The "Temporal_level" field makes it possible to indicate the temporal level indicating the image frequency; • the "Quality_level" field is used to indicate the progressive quantization level, and thus to control the flow rate / quality and / or complexity; • the base ("layer_base_flag") field indicates a data syntax (payload, or payload) compatible with AVC (no inter-layer prediction, no refinement in quality). Thus, the enhancements are indicated in the SVC stream using the parameters T, D and Q (and the indicator "layer_base_flag", which indicates whether or not it is an enhancement) included in the header of each data packet (NALU). More precisely, by noting Tmin, Dmin and Qmin the parameters of the basic level of the first layer: - the parameter T indicates a temporal enhancement, with T> Tmin; the Q parameter indicates an enhancement in quantization quality, with Q> Qmin; parameter D indicates a spatial enhancement, with D> Dmin. In particular, we recall that if the spatial enhancement D is a spatial enhancement of ratio 1, then it is equivalent to an enhancement in CGS type quality. 2.4 Disadvantages of the prior art It is thus noted that the different types of scalability of the standard (temporal scalability, in quality, spatial yes) use different tools: for example the temporal scalability uses (or not) images B, bidirectionally predicted , scalability in quality uses (or not) FGS coding, etc. As previously stated, the definition of a profile of the SVC standard indicates which tools are supported by that profile, and the level defines the decoding complexity associated with the level. A decoder compatible with a profile P and a level L must therefore support all types of scalability defined in the profile P.

  However, we note that certain types of applications do not require knowledge of all types of scalability. By way of example, for a given level L indicating 1000 macroblocks to be decoded, the SVC stream may correspond to: an SNR scalability stream such as, for example, a CIF base level (22 × 18 macroblocks), in English Common Intermediate Format, corresponding to a digital image format defined by the ITU, followed by two levels of enhancement in quantization quality; or a spatial scalability stream such as, for example, a base level in QCIF (9x 11 macroblocks), or quarter of CIF, followed by a QVGA spatial enhancement (20x15 macroblocks) giving 399 macroblocks. Although these two streams are compatible with the L level, it is impossible to know, on the decoding side, what type of scalability was used on the coding side: spatial scalability with a dyadic ratio, a 3/2 ratio, or a predetermined ratio, or SNR scalability with FGS, or CGS type coding, etc. A first possible solution to remedy this problem is to define new profiles, restricting the type of scalability supported.

  However, this solution increases the number of defined profiles by adding additional complexity in the understanding of the profiles. A second solution is based on the use of a notion of level within the definition of a profile, to specify what type of scalability is supported by a level.

  However, this solution is not allowed by the current definition of levels. Indeed, in AVC, a level of complexity corresponds to a number of macroblocks to be decoded per second, and this number is increasing with the levels. This definition can also be extended to the SVC standard by specifying rules for setting the number of decodable macroblocks per second in an SVC stream according to the number of enhancements. For example, the following steps are considered: considering the complexity N of decoding the base level of the first layer of the SVC stream; a spatial enhancement of ratio rx, ry (increase of the horizontal resolution of a factor rx, and of the vertical resolution of a factor ry) adds a decoding complexity of the order of rx * ry * N at the level of base of the first layer; an enhancement in quantization quality adds a decoding complexity of the order of 0.5 * N ', with N' the decoding complexity of the QO quality level of the spatio-temporal layer to which the enhancement in quality belongs. Thus, if we consider the following flow: - DO Q0, corresponding to the basic level of complexity N; - DO Q1, adding a complexity of 0.5 * N; - DO Q2, adding a complexity of 0.5 * N; Dl Q0, of ratio 2 (change of dyadic resolution) and adding a complexity of 4 * N; - D1 Q1, adding a complexity of 0.5 * (4 * N); - D1 Q2, adding a complexity of 0.5 * (4 * N); where Q1 and Q2 correspond to enhancement levels in quantization quality, and D1 corresponds to a spatial enhancement (QO and DO corresponding to the base level), then the total decoding complexity of the stream is 10 * N.

  Nevertheless, a level of complexity still can not define the type of scalability supported. Thus, if one considers the techniques of the prior art, a spatial scalability supported by a stream SVC can be deduced by the decoder from the reading of information present in the stream, but this reading is complex and n ' is not immediate. In effect, the decoder must first read the header of the video data units (slice header) in order to know which SPS these data units refer to. Then, the decoder must read in the corresponding SPS the spatial resolution of the images. In addition, in order to know the type of spatial scalability, the decoder must read this information for each space-time layer. However, as indicated above, it is not easy to know immediately which spatio-temporal layer refers to an SPS without reading the video data units (slice header). Scalability in quantization quality can be deduced by reading the headers of the video data packets, but again, the relationship with a given SPS is not immediate and the deduction of the scalability path used to decode some spatio-temporal layer is complex. 3. DISCLOSURE OF THE INVENTION The invention proposes a new solution that does not have all of these disadvantages of the prior art, in the form of a method of decoding a data flow representative of an image. or a sequence of images, having a data layer organization comprising at least a first layer, each layer comprising a base level and optionally at least one enhancement level (in quantization quality). The data is organized into data packets (NALUs), each data packet comprising a header, and said stream is associated with a set of signaling packets (SPS) characterizing it. According to the invention, such a decoding method comprises the following steps: reading, in each signaling packet, at least one piece of information representative of a scalability class applied to the data packets of at least one layer; deciding, for each of the data packets of said layer, an ability to decode (or not) said packet, on the basis of said information and without reading the header of said data packets.

  Thus, the invention is based on a new and inventive approach to decoding data in a scalable stream, according to which the scalability class that a decoder is able to decode is determined. The concept of scalability class refers to the notion of profile and level detailed above. The class makes it possible in particular to decide the decoding capacity in the case where the decoder does not support all the tools of the profile. According to one particular aspect of the invention, each signaling packet comprises at least one indicator, specifying to which layer (s) of said stream said at least one piece of information representative of a scalability class applies.

  Thus, the decoders can have directly and simply information to validate their capacity, or their inability to process the data stream considered. The one or more information representative of a scalability class may in particular comprise: an information representative of a category of spatial scalability; and / or - information representative of a type of quality coding (SNR). In this case, the information representative of a category of spatial scalability can, for example, designate one of the categories belonging to the group comprising: spatial scalability not using inter-layer prediction or inter-layer prediction with conservation of the spatial resolution between two layers; - spatial scalability with a dyadic ratio; - spatial scalability with a 3/2 ratio; at least one spatial scalability with a determined ratio: at least one filtering and / or spatial processing tool. Of course, it is not mandatory for all of these categories to be available: it is possible for a standard to retain only a few (two or more) of these categories, and / or add other categories.

  The information representative of a type of quality coding may for example be a binary element, corresponding, for a first value, to a non-progressive coding in quality (CGS), and for a second value, to a progressive coding in quality. (FGS).

  According to a particular aspect of the invention, provision is furthermore made for each signaling packet to include information specifying the necessity of FGS type decoding in at least one lower layer of said stream. The invention also relates to devices for decoding a data stream representative of an image or an image sequence, comprising means for implementing the method described above, and in particular: reading means in each signaling packet, at least one piece of information representative of a scalability class applied to the data packets of at least one layer; decision means, for each of the data packets of said layer, of a capacity to decode (or not) said packet, taking into account said information representative of a scalability class and without reading the header of said data packets. Such a device may especially be a multimedia terminal, a microcomputer, a radio-telephone, a television, and more generally any apparatus or element of apparatus capable of rendering audiovisual streams. The invention also relates to computer program products downloadable from a communication network and / or recorded on a computer readable medium and / or executable by a processor, and comprising program code instructions for the implementation of the method decoding. The invention furthermore relates to a method of encoding a data stream representative of an image or an image sequence as described above, and which can be processed by the decoding method of the invention. Such a coding method comprises a step of inserting, in each signaling packet, at least one piece of information representative of a scalability class applied to the data packets of at least one layer, so as to allow a decoder to control its ability to decode (or not) at least one of the data packets of said layer, taking into account said information representative of a scalability class and without reading the header of said data packets. The invention also relates to a coding device implementing this method, and comprising means for inserting into each signaling packet at least one piece of information representative of a scalability class applied to data packets of at least one data packet. layer, so as to allow a decoder to control its ability to decode (or not) at least one of the data packets of said layer, taking into account said information representative of a scalability class and without reading the header said data packets. Such a device can in particular be a server, for example a video on demand (VOD) server, or an intermediate network element, such as a network gateway. The invention also relates to computer program products downloadable from a communication network and / or recorded on a computer-readable and / or executable medium by a processor, comprising program code instructions for implementing the encoding method described. above. The invention finally relates to the signals representative of a data flow representative of an image or sequence of images and produced by this coding method, as well as the data carriers carrying at least one of these signals. Each signaling packet of such a signal comprises at least one piece of information representative of a scalability class applied to the data packets of at least one layer, so as to allow a decoder to control its ability to decode (or not) for each of the data packets of said layer, said packet of said layer, taking into account said information representative of a scalability class and without reading the header of said data packets.

  In such a signal, the information representative of a scalability class may include at least one of the elements belonging to the group comprising: an information representative of a category of spatial scalability, designating one of the categories belonging to the group comprising: lack of inter-layer prediction for spatial scalability; spatial scalability with a dyadic ratio; spatial scalability with a 3/2 ratio; - spatial scalability with a determined ratio; at least one filtering and / or spatial processing tool; information representing a type of quality coding (SNR), in the form of a binary element, corresponding, for a first value, to a non-coding; progressive in quality (CGS), and for a second value, progressive quality coding (FGS); an information item specifying the necessity of an FGS type decoding in at least one lower layer of said stream. 4. List of Figures Other features and advantages of the invention will appear more clearly on reading the following description of a particular embodiment, given as a simple illustrative and non-limiting example, and the accompanying drawings, among which: Figure 1, described in relation to the prior art, schematically shows the layered structure of a scalable flow; Figure 2 shows the main steps of the decoding method according to a particular embodiment of the invention; FIG. 3 illustrates the main steps of the coding method according to a particular embodiment of the invention; FIG. 4 illustrates the structure of a signal according to the invention; Figures 5 and 6 respectively illustrate the simplified structure of a coding device and a decoding device according to the invention. . DESCRIPTION OF AN EMBODIMENT OF THE INVENTION 5.1 GENERAL PRINCIPLE The general principle of the invention is based on the consideration, on the decoder side, of information representative of a scalability class applied to the data packets of FIG. least one layer of the SVC stream. This information can notably be inserted into the signaling packets characterizing the SVC stream, also called SPS, so as to allow a decoder to control its ability to decode all or part of the data packets of the layer in question, without reading the header data packets, also called slice header in English. Each signaling packet may also include one or more indicators, specifying to which layer (s) of the flow the information representative of a scalability class applies. A particular embodiment of the invention is described below, in which three classes of scalability are considered: a class of spatial scalability; a quality scalability class (also called SNR scalability); and a class of joint spatial and quality scalability. According to this particular embodiment of the invention, the compatibility of a decoder, that is to say its ability to decode data packets of an SVC stream, therefore results in a triplet of information of complexity comprising a profile P, a level L and a scalability class C. In other words, one comes to add to the notion of profile / level existing previously a notion of additional class 25. 5.2 Principle of coding Encoder side, the invention consists according to this particular embodiment to report in the SVC stream the scalability class supported by the stream, or possibly, more generally, the complexity information triplet. Thus, as illustrated in FIG. 3, the coding method comprises an insertion step 31 in the signaling packets of an SVC stream 30 of at least one piece of information representative of a scalability class applied to the data packets. at least one layer. Each spatio-temporal layer of an SVC stream is more precisely considered. These signaling packets belong for example to an SPS unit, as defined previously. It is recalled that an SPS unit is conventionally generated for each spatial layer of the stream. This representative information of a scalability class includes information representative of a category of spatial scalability and / or information representative of a type of quality coding (SNR). More specifically, the information representative of a spatial scalability category designates one of the categories belonging to the group comprising: spatial scalability not using inter-layer prediction or inter-layer prediction with preservation of the spatial resolution between two layers; - spatial scalability (with inter-layer prediction) with a dyadic ratio; spatial scalability (with inter-layer prediction) with a 3/2 ratio; spatial scalability (with inter-layer prediction) with a determined ratio, each category including the previous one.

  This spatial scalability category information can also be extended to denote an absence of inter-layer prediction, or at least one particular filtering and / or spatial processing tool (for example the use or not of a prediction technique of smoothed texture called "smoothed reference prediction", the use or not of smoothing technique at the boundaries of blocks said "deblocking filter" in the inter-layer prediction, the use or not of customizable filter for the inter-layer prediction,. ..). The information representative of a scalability class may also include information representative of a type of coding in quantization quality at the level of the current spatio-temporal layer, if it is present in the stream, that is, that is, if the number of quantization quality levels for the current spatio-temporal layer is greater than 1. For example, this information representative of a type of quantization quality encoding is in the form of a bit , corresponding, for a first value (for example a bit equal to `0 '), to a non-progressive coding in quality (CGS), and for a second value (for example a bit equal to` l'), to a coding progressive in quality (FGS). According to the particular embodiment described, each signaling packet also comprises at least one indicator, specifying to which layer (s) of the flow the information representative of a scalability class applies. In other words, this indicator specifies which spatial-temporal layer of the SVC stream is associated with the signaling packets (SPS). Finally, each signaling packet may also include information specifying the need for FGS type decoding in at least one lower layer of the stream. This information can notably be used for the decoding of the SVC stream during the decoding of the current spatio-temporal layer. The following is an example of a syntax corresponding to an SPS unit of the SVC standard embodying the invention according to this particular embodiment of the invention (the added elements appearing in italics): seq_parameter_set_svc_extension () {C Descriptor layer id 11 information field added u (3) spatial_scalability_type 11 information field added u (2) num_max_quality_level_minusl // information field added u (2) if (num_max_quality_level_minusl> 0) {// test added required_ fgs_ flag ll field information added u (1)} // end of the test requiredjgs_layerjlag // information field added u (1) interlayer_deblocking_filter_control_present_flag 0 u (1) if (spatial_scalability_type! = 0) {// modify as a test extended_spatial_scalability 0 u (2) if (chroma_format_idc> 0) {chroma_phase_x_plusl 0 u (2) chroma_phase y_plusl 0 u (2)} // end of the add test edit if (extended_spatial_scalability = = 1) {scaled_base_left_offset 0 se ( v ) scaled_base_top_offset 0 se (v) scaled_base_right_offset 0 se (v) scaled_base_bottom_offset 0 se (v)}} if (num_max_quality _level_minusl> 0) {// added a test for the presence of the following information fgs_coding_mode 2 u (1) if ( fgs_coding_mode = = 0) {grouping_size_minusl 2 ue (v)} else {numPosVector = 0 do {If (numPosVector = = 0) {scan_indexO_minusl 2 ue (v)} else {delta_scan_index_minusl [numPosVector] 2 ue (v)} numPosVector ++ } while (scanPosVectLuma [numPosVectoru 1] <15)}} // end of the information presence test} The semantics associated with this syntax are more precisely described in the aforementioned document JVT-T201. More precisely, the layer_id element corresponds to the indicator indicating the spatio-temporal layer to which the SPS relates.

  The spatial_scalability_type element corresponds to the information representative of a scalability class indicating the spatial scalability category used. For example, the value 0 corresponds to a spatial scalability not using inter-layer prediction or inter-layer prediction with conservation of the spatial resolution between two layers, the value 1 indicates a dyadic spatial scalability (with inter-layer prediction) , the value 2 indicates a spatial scalability of ratio 3/2 (with inter-layer prediction), and the value 3 indicates a spatial scalability of determined ratio (with inter-layer prediction).

  The num_max_quality__level_minusl element corresponds to information representative of a scalability class indicating whether a type of quantization quality enhancement encoding is used, indicating the maximum number of quality levels for the current spatio-temporal layer, and The required_fgs_flag element specifies the type of quality coding used, for example, indicating whether an FGS encoding is used in the spatio-temporal layer under consideration. Finally, the required_fgs_layer_flag element corresponds to the information specifying the need for an FGS type decoding in at least one lower dependent spatio-temporal layer of the stream. In other words, this indicator specifies whether the enhancements in quality of the layers on which the current layer depends have been coded by an FGS coding. The SVC signal 32 thus constructed, as illustrated in FIG. 4, thus comprises data packets 42 and a set of SPS signaling packets characterizing the stream, each signaling packet 41 comprising at least one piece of information representative of a class of scalability applied to the data packets 42 of the corresponding layer. In particular, it is recalled that the SPS 41 associated with a layer of the stream 32 can be broadcast either with the data packets 42 of the stream 32, or through a separate transmission channel. 5.3 Principle of decoding On the decoder side (for example a broadcast server), it is thus possible to easily know the scalability classes supported in the stream. In particular, if we consider the three above-mentioned scalability classes (spatial scalability, quality scalability, joint scalability), the joint scalability class can be deduced from the presence of both spatial scalability and scalability. quality. As illustrated in connection with FIG. 2, the decoding method according to this particular embodiment of the invention comprises the following steps, upon receipt of an SVC stream 32, coded as previously described: - reading 21, in each packet of signaling, at least one piece of information representative of a scalability class applied to the data packets of at least one layer; decision of an ability to decode all or part of the data packets of the layer, based on this information and without reading the header of the data packets. Such a decoder can thus simplify decoding an SVC stream: More precisely, by using the terminology defined on the coder side, the decoder implements the following algorithm: reading of all the SPS data packets of the received stream, to be decoded; for each SPS: o read the layer_id element; o reading the spatial_scalability_type element; o reading the element num_max_quality_level_minusl; ^ if this element is nonzero: reading the required_fgs_flag element; o reading the required_fgs_layer_flag element. 5.4 Example of an SVC stream The following flows are described as examples: Stream 1: Stream 2: Stream 3: Stream 4: Scenario Scenario Spatial Scenario scenario Conjunction SNR Conjunct discardable (separable) representation CIF15Hz CIF CIF CIF base D = 0 Q = 0 D = 0 Q = 0 D = 0 Q = 0 D = 0 Q = 0 enhancement CIF15Hz SD, ratio 2 CIF CIF, 1 D = 0 Q = 1, D = 1 Q = 0 D = 0 Q = 1, FGS discardable CGS D = 0 Q = 1, FGS enhancement CIF30Hz HD, ratio SD, ratio 2 SD, ratio 2 2 D = 1 Q = 0 any D = 1 Q = 0 D = 1 Q = 0 D = 2 Q = 0 enhancement CIF30Hz SD D = 1 Q = 1, SD D = 1 Q = 1, FGS 3 D = 1 Q = 1, FGS FGS In the first three examples of flows (SNR, spatial, and spouse flows) each enhancement information depends on previous information (the basic representation does not depend on any other information).

  In the fourth example, the enhancement 2 of the discardable joint flow depends on the basic representation, and not on the enhancement 1 directly preceding the enhancement 2. By repeating the definitions proposed in paragraph 5.2, the values of the information introduced in the SPS of each space-time layer, for each of the four flows described above. Thus, if we consider the first stream comprising only scalability in quantization quality, the information representative of a scalability class can be represented as follows: SNR SPS attached SPS attached to the representation enhancement base 2 layer_id (= D) 0 1 spatial_scalability_type x 0 num_max_quality_level_minusl: 1 1 required_fgs_flag 0 1 required_fgs_layer_flag 0 0 where `x 'means no field value is specified,` 0' means inserting a bit equal to and `1 'is the insertion of a bit equal to` 1'. If we consider the second stream comprising only a spatial scalability, the information representative of a scalability class can be represented as follows: Spatial SPS attached SPS attached SPS attached to the to the representation enhancement base enhancement 1 2 layer_id (= D) 0 1 2 spatial_scalability_type x 1 3 num_max_quality_level_minusl: 0 0 0 required_fgs_flag xxx required_fgs_ayer_flag 0 0 0 If we consider the third stream comprising both scalability in quantization quality and spatial scalability, the representative information of a scalability class can be represented as follows: SPS attached SPS spouse attached to the basic enhancement representation 2 layer_id (= D) 0 1 spatial_scalability_type X 1 num_max_quality_level_minusl: 1 1 required_fgs_flag 1 1 required_fgs_layer_flag 0 1 Finally, if the we consider the fourth stream comprising the f If a quantization quality scalability and a discardable spatial scalability, the information representative of a scalability class can be represented as follows: disconnected spouse SPS attached SPS attached to the representation enhancement base 2 layer_id (= D) 0 1 spatial_scalability_type X 1 num_max_quality_level_minusl: 1 1 required_fgs_flag 1 1 required_fgs_layer_flag 0 0 On reading these examples, we see that the fourth discardable joint stream can be decoded by a decoder that only supports spatial scalability, even if the complete stream includes enhancements in FGS quality. Thus, decoders of reduced complexity can decode certain complex flows, which were considered as not compatible with the decoders of less complexity according to the prior art. 5.5 Structure of an Encoding Device and a Decoding Device Finally, in connection with FIGS. 5 and 6, the simplified structure of an encoding device and a decoding device implementing respectively a technique encoding method and a decoding technique according to one of the particular embodiments described above.

  Such a coding device comprises a memory 51, a processing unit 52, equipped for example with a microprocessor P, and driven by the computer program 53, implementing the coding method according to the invention. At initialization, the code instructions of the computer program 53 are for example loaded into a RAM memory before being executed by the processor of the processing unit 52. The processing unit 52 receives as input data to encode, for example video data. The microprocessor of the processing unit 52 implements the steps of the coding method described above, according to the instructions of the computer program 53, to generate a scalable stream 32, comprising at least one piece of information representative of a class of scalability . For this, the coding device comprises means for inserting, in the signaling packets, at least one piece of information representative of a scalability class applied to the data packets of at least one layer, so as to enable a decoder to control its ability to decode the data packets of said layer, without reading the header of the data packets. These means are controlled by the microprocessor of the processing unit 52. A decoding device comprises a memory 61, a processing unit 62, equipped for example with a microprocessor P, and driven by the computer program. 63, implementing the decoding method according to the invention. At initialization, the code instructions of the computer program 63 are for example loaded into a RAM before being executed by the processor of the processing unit 62. The processing unit 62 receives as input the scalable stream 32. The microprocessor of the processing unit 62 implements the steps of the decoding method described above, according to the instructions of the computer program 63, for decoding at least one data packet of the scalable stream 32. For this, the decoding device comprises means for reading, in each signaling packet, at least one piece of information representative of a scalability class applied to the data packets of at least one layer, and means for deciding a capacity to decode the data packets of the layer, based on this information and without reading the header of the data packets. These means are controlled by the microprocessor of the processing unit 62.10

Claims (14)

  A method of decoding a data stream representative of an image or sequence of images, having a data layer organization comprising at least a first layer, each layer comprising a base level and optionally at least one a level of enhancement, said data being organized into data packets (NALUs), each data packet comprising a header, said stream being associated with a set of signaling packets (SPS) characterizing said stream, characterized in that comprises the steps of: reading, in each signaling packet, at least one piece of information representative of a scalability class applied to the data packets of at least one layer; Deciding, for each of the data packets of said layer, an ability to decode said packet, taking into account said information representative of a scalability class and without reading the header of said data packets.   2. Decoding method according to claim 1, characterized in that each signaling packet comprises at least one indicator, specifying to which layer (s) of said flow said at least one piece of information representative of a scalability class. 'applied.   3. Decoding method according to any one of claims 1 to 2, characterized in that the one or more information representative of a scalability class comprises: information representative of a category of spatial scalability; and / or information representative of a type of quality coding (SNR).   4. Decoding method according to claim 3, characterized in that said information representative of a category of spatial scalability designates a descategories belonging to the group comprising: spatial scalability not using inter-layer prediction or inter-layer prediction with conservation of the spatial resolution between two layers; - spatial scalability with a dyadic ratio; spatial scalability with a 3/2 ratio; at least one spatial scalability with a determined ratio: at least one filtering and / or spatial processing tool.   5. decoding method according to any one of claims 3 and 4, characterized in that said information representative of a quality coding type is a binary element, corresponding, for a first value, to a non-progressive coding quality (CGS), and for a second value, a progressive quality coding (FGS).   6. decoding method according to any one of claims 1 to 5, characterized in that each signaling packet comprises information specifying the need for a type of decoding FGS in at least one lower layer of said stream.   A device for decoding a data stream representative of an image or sequence of images, having a data layer organization comprising at least a first layer, each layer comprising a base level and optionally least one enhancement level, said data being organized into data packets (NALUs), each data packet comprising a header, said stream being associated with a set of signaling packets (SPS) characterizing said stream, characterized in that it comprises: means for reading, in each signaling packet, at least one piece of information representative of a scalability class applied to the data packets of at least one layer; - decision means, for each of the packets; data of said layer, of an ability to decode said packet, taking into account said information representative of a class of scalability and without reading the header of said quets of data.   8. Computer program product downloadable from a communication network and / or recorded on a computer readable medium and / or executable by a processor, characterized in that it comprises program code instructions for the implementation of the decoding method according to any one of claims 1 to 6.   A method of encoding a data stream representative of an image or sequence of images, having a data layer organization, comprising at least a first layer, each layer comprising a base level and optionally least one enhancement level, said data being organized into data packets (NALUs), each data packet comprising a header, said stream being associated with a set of signaling packets (SPS) characterizing said stream, characterized in that comprises an insertion step, in each signaling packet of at least one piece of information representative of a scalability class applied to the data packets of at least one layer, so as to allow a decoder to control its capacity to decode for each of the data packets of said layer, said packet of said layer, taking into account said information representative of a class of scalability and without lectu re of the header of said data packets.   Apparatus for encoding a data stream representative of an image or sequence of images, having a data layer organization, comprising at least a first layer, each layer comprising a base level and optionally least one enhancement level, said data being organized into data packets (NALUs), each data packet comprising a header, said stream being associated with a set of signaling packets (SPS) characterizing said stream, characterized in that it comprises means for inserting in each signaling packet at least one piece of information representative of a scalability class applied to the data packets of at least one layer, so as to allow a decoder to control its capacity to decode for each of the data packets of said layer, said packet of said layer, taking into account said information representative of a scalability class and without the of the header of said data packets.   11. Computer program product downloadable from a communication network and / or recorded on a computer readable medium and / or executable by a processor, characterized in that it comprises program code instructions for the implementation of the coding method according to claim 9.   Signal representative of a data stream representative of an image or sequence of images, having a data layer organization, comprising at least a first layer, each layer comprising a base level and optionally at least one a level of enhancement, said data being organized into data packets (NALUs), each data packet comprising a header, said stream being associated with a set of signaling packets (SPS) characterizing said stream, characterized in that each packet of data signaling comprises at least one piece of information representative of a scalability class applied to the data packets of at least one layer, so as to enable a decoder to control its ability to decode, for each of the data packets of said layer, said packet of said layer, taking into account said information representing a scalability class and without reading the header of said packets of data.   13. Signal representative of a data stream according to claim 12 characterized in that said at least one piece of information representative of a scalability class comprises at least one of the elements belonging to the group comprising: a piece of information representative of a category of spatial scalability, designating one of the categories belonging to the group comprising: - absence of inter-layer prediction for spatial scalability; - spatial scalability with a dyadic ratio; - spatial scalability with a 3/2 ratio; - spatial scalability with a determined ratio; at least one filtering and / or spatial processing tool. an information representative of a type of quality coding (SNR), in the form of a binary element, corresponding, for a first value, to a non-progressive coding in quality (CGS), and for a second value, to progressive quality coding (FGS); an information item specifying the necessity of an FGS type decoding in at least one lower layer of said stream.   A data carrier carrying at least one signal representative of a data stream representative of an image or sequence of images, having a data layer organization, comprising at least a first layer, each layer comprising a base level and optionally at least one enhancement level, said data being organized into data packets (NALUs), each data packet comprising a header, said stream being associated with a set of signaling packets (SPS) characterizing said stream characterized in that each signaling packet comprises at least one information representative of a scalability class applied to the data packets of at least one layer, so as to enable a decoder to control its ability to decode for each of the data packets of said layer, said packet of said layer, taking into account said information representative of a class e of scalability and without reading the header of said data packets.