Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
SYSTEM AND METHOD FOR USING PARALLELLY DECODABLE SLICES FOR MULTI-VIEW VIDEO CODING
Document Type and Number:
WIPO Patent Application WO/2008/047300
Kind Code:
A3
Abstract:
A system and method for enabling parallel decoder implementation for different views, even when there are existing dependencies between views. In various embodiments of the present invention, information is signaled to a decoder that slices are coded using certain constraints, so that parallel decoding of slices is possible. This signaling can be performed at a sequence parameter set level, or it could be performed at picture parameter set level, slice header level or macroblock header level. Additionally, the delay between various views is also signaled to the decoder, at the same location as the constraint information in various embodiments. Various algorithms can also be used to improve the coding efficiency of the system.

Inventors:
UGUR KEMAL (FI)
LAINEMA JANI (FI)
HANNUKSELA MISKA (FI)
Application Number:
PCT/IB2007/054197
Publication Date:
July 03, 2008
Filing Date:
October 15, 2007
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
NOKIA CORP (FI)
NOKIA INC (US)
UGUR KEMAL (FI)
LAINEMA JANI (FI)
HANNUKSELA MISKA (FI)
International Classes:
H04N7/26; H04N13/00
Domestic Patent References:
WO2006073116A12006-07-13
Foreign References:
EP1835747A12007-09-19
EP1578131A12005-09-21
Other References:
WIEGAND T. ET AL.: "Overview of the H.264/AVC video coding standard", IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, vol. 13, no. 7, July 2003 (2003-07-01), pages 560 - 576, XP001169882
See also references of EP 2080379A4
Attorney, Agent or Firm:
ALBERT, G.Peter (11250 El Camino Real Suite 20, San Diego CA, US)
Download PDF:
Claims:

WHAT IS CLAIMED IS:

L A method of decoding frames from multiple views having an inter- view prediction relationship, comprising: receiving a particular macroblock of a first frame of a picture taken from a first view for decoding; decoding information regarding which macroblocks in a corresponding second frame of the picture taken from a second view can be referenced by the particular macroblock; if the macroblocks permitted to be referenced by the particular macroblock have been reconstructed to a reference memory, decoding the particular macroblock.

2. The method of claim 1 , wherein a notification is received from a different decoder that the macroblocks permitted to be referenced by the particular macroblock have been reconstructed to the reference memory.

3. The method of claim 1, further comprising, after the decoding of the particular macroblock has been completed notifying the different decoder,

4. The method of claim 1, further comprising, before decoding the particular macroblock, an adaptive deblocking filter is used to remove blocking artifacts in a macroblock of the second frame that corresponds to the particular macroblock before the particular macroblock is decoded.

5. The method of claim 4, wherein the right and bottom borders of the filtered macroblock in the second frame are not filtered.

6. The method of claim 4, wherein the corresponding macroblock of the second frame is used, unfiltered, for inter- view prediction, and wherein the corresponding macroblock of the second frame is used, post-filtering, for temporal prediction.

7, The method of claim 4, wherein the area of the corresponding macroblock of the second frame that can be referenced by the particular macroblock is limited.

8. The method of claim 1 , wherein the bottom and right edges of a macroblock of the second frame that corresponds to the particular macroblock is padded by a predetermined number of pixels after the corresponding macroblock has been reconstructed.

9. The method of claim 1, wherein the macroblocks in the corresponding second frame which can be referenced by the particular macroblock are limited for fractional pixels.

10. The method of claim 1 , wherein the decoding order of macroblocks in the first and second frames moves from a top-left corner of each frame towards a bottom-right comer.

11. The method of claim 1 , wherein the decoding order of macroblocks in the first and second frames moves from a top-right corner of each frame towards a bottom-left corner.

12. The method of claim 1, wherein the decoding order of macroblocks in the first and second frames follows an arbitrary direction.

13. The method of claim 12, wherein the decoding order of macroblocks in the first and second frames are signaled in the bistream.

14. The method of claim 1, wherein the first and second frames are partitioned into slice groups based upon the arrangement of first and second cameras responsible for the first and second views.

15. The method of claim 1, wherein the macroblocks in the corresponding second frame which can be referenced by the particular macroblock are based at least

in part upon the arrangement of first and second cameras responsible for the first and second views.

16. The method of claim 1, wherein at least one of the first and second frames is taken from a picture that was previously modified by an encoder.

17. The method of claim 1, wherein the information includes the number of macroblocks that are reconstructed to the reference memory before receiving a notification from the different decoder.

18. The method of claim 1 , wherein the information includes a number of notifications to wait before beginning the decoding process.

19. The method of claim 1 , wherein the received information is signaled at the sequence parameter set level.

20. The method of claim 1, wherein the received information is signaled at the picture parameter set level.

21. The method of claim 1, wherein the received information is signaled at the slice header level.

22. The method of claim 1, wherein the received information is signaled at the macroblock header level.

23. The method of claim 1, wherein coded motion vector information is also received and decoded, and wherein the coding of motion vectors between the first view and the second view are modified when the statistical properties of motion vectors change as a reference area is restricted.

24. The method of claim 23. wherein a single codeword is used for coding both components of motion vectors between the first view and the second view, and wherein a mapping from motion vectors to codewords is adapted at each macroblock depending on a reference area.

25. The method of claim 23, wherein horizontal and vertical components of each motion vector is separately coded, and wherein an available reference area for the horizontal and vertical components is considered when mapping each motion vector component to a binary codeword.

26. A computer program product, embodied in a computer-readable medium, for decoding frames from multiple views having an inter-view prediction relationship, comprising: computer code for receiving a particular macroblock of a first frame of a picture taken from a first view for decoding; computer code for decoding information regarding which macroblocks in a corresponding second frame of the picture taken from a second view can be referenced by the particular macroblock; computer code for, if the macroblocks permitted to be referenced by the particular macroblock have been reconstructed to a reference memory, decoding the particular macroblock.

27. An apparatus, comprising: a processor; and a memory unit communicatively connected to the processor and including: computer code for receiving a particular macroblock of a first frame of a picture taken from a first view for decoding; computer code for decoding infoπnation regarding which macroblocks in a corresponding second frame of the picture taken from a second view can be referenced by the particular macroblock; computer code for, if the macroblocks permitted to be referenced by the particular macroblock have been reconstructed to a reference memory, decoding the particular macroblock.

28. The apparatus of claim 27, wherein a notification is received from a different decoder that the macroblocks permitted to be referenced by the particular macroblock have been reconstructed to the reference memory.

29. The apparatus of claim 27, further comprising, after the decoding of the particular macroblock has been completed notifying the different decoder,

30. The apparatus of claim 27, further comprising, before decoding the particular macroblock, an adaptive deblocking filter is used to remove blocking artifacts in a macroblock of the second frame that corresponds to the particular macroblock before the particular macroblock is decoded.

31. The apparatus of claim 30, wherein the right and bottom borders of the filtered macroblock in the second frame are not filtered.

32. The apparatus of claim 30, wherein the corresponding macroblock of the second frame is used, unfiltered, for inter-view prediction, and wherein the corresponding macroblock of the second frame is used, post- filtering, for temporal prediction.

33. The apparatus of claim 30, wherein the area of the corresponding macroblock of the second frame that can be referenced by the particular macroblock is limited.

34. The apparatus of claim 27, wherein the bottom and right edges of a macroblock of the second frame that corresponds to the particular macroblock is padded by a predetermined number of pixels after the corresponding macroblock has been reconstructed.

35, The apparatus of claim 27, wherein the macroblocks in the corresponding second frame which can be referenced by the particular macroblock are limited for fractional pixels.

36. The apparatus of claim 27, wherein the decoding order of macroblocks in the first and second frames moves from a top-left corner of each frame towards a bottom-right corner.

37. The apparatus of claim 27, wherein the decoding order of macroblocks in the first and second frames moves from a top-right corner of each frame towards a bottom-left corner.

38. The apparatus of claim 27, wherein the decoding order of macroblocks in the first and second frames follows an arbitrary direction.

39. The apparatus of claim 38, wherein the decoding order of macroblocks in the first and second frames are signaled in the bistream.

40. The apparatus of claim 27, wherein the first and second frames are partitioned into slice groups based upon the arrangement of first and second cameras responsible for the first and second views.

41. The apparatus of claim 27, wherein the macroblocks in the corresponding second frame which can be referenced by the particular macroblock are based at least in part upon the arrangement of first and second cameras responsible for the first and second views.

42. The apparatus of claim 27, wherein at least one of the first and second frames is taken from a picture that was previously modified by an encoder.

43. The apparatus of claim 27, wherein the information includes, the number of macroblocks that are reconstructed to the reference memory before receiving a notification from the different decoder.

44« The apparatus of claim 27, wherein the information includes a number of notifications to wait before beginning the decoding process.

45. The apparatus of claim 27, wherein the received information is signaled at the sequence parameter set level.

46. The apparatus of claim 27, wherein the received information is signaled at the picture parameter set level.

47. The apparatus of claim 27, wherein the received information is signaled at the slice header level.

48. The apparatus of claim 27, wherein the received information is signaled at the macroblock header level.

49. The apparatus of claim 27, wherein coded motion vector information is also received and decoded, and wherein the coding of motion vectors between the first view and the second view are modified when the statistical properties of motion vectors change as a reference area is restricted.

50. The apparatus of claim 49, wherein a single codeword is used for coding both components of motion vectors between the first view and the second view, and wherein a mapping from motion vectors to codewords is adapted at each macroblock, depending on a reference area.

51. The apparatus of claim 49, wherein horizontal and vertical components of each motion vector is separately coded, and wherein an available reference area for the horizontal and vertical components is considered when mapping each motion vector component to a binary codeword.

52. A method of encoding frames from multiple views having an inter- view prediction relationship, comprising: encoding into a bitstream a first plurality of macroblocks of a first frame of a picture taken from a first view; encoding into a bitstream a second plurality of macroblocks of a second frame of the picture taken from a second view, at least a portion of the second frame referencing at least a portion of the first frame; and

signaling in the bitstream an indication of, for each macroblock in the second frame, which of the plurality of macroblocks in the first frame may be referenced.

53. The method of claim 52, wherein the signaling occurs at the sequence parameter set level.

54. The method of claim 52, wherein the signaling occurs at the picture parameter set level.

55. The method of claim 52, wherein the signaling occurs at the slice header level.

56. The method of claim 52, wherein the signaling occurs at the macroblock header level.

57. The method of claim 52, wherein the macroblocks in the first frame which can be referenced by a particular macroblock in the second frame is based at least in part upon the arrangement of first and second cameras responsible for the first and second views.

58. The method of claim 52, wherein the area of macroblocks in the first frame that can be referenced by a particular macroblock in the second frame is limited for fractional pixels.

59. The method of claim 52, wherein the encoding order of macroblocks in the first and second frames moves from a top-left corner of each frame towards a bottom-right corner.

60. The method of claim 52, wherein the encoding order of macroblocks in the first and second frames moves from a top-right comer of each frame towards a bottom-left comer.

61. The method of claim 52, wherein the indication includes a number of macroblocks that are to be reconstructed to the reference memory before a notification

3 is transmitted indicating that the decoding of at least one of the second plurality of

4 macroblocks can begin.

1 62. The method of claim 52, wherein the indication includes a number of

2 notifications to wait before the decoding of at least one of the second plurality of

3 macroblocks can begin.

1 63. The method of claim 52, wherein a decoding order of macroblocks in

2 the first and second frames is signaled in the bistream.

1 64. The method of claim 52, wherein coded motion vector information is

2 also encoded into the bitstream, and wherein the coding of motion vectors between

3 the first view and the second view are modified when the statistical properties of

4 motion vectors change as a reference area is restricted.

1 65. The method of claim 64, wherein a single codeword is used for coding

2 both components of motion vectors between the first view and the second view, and

3 wherein a mapping from motion vectors to codewords is adapted at each macroblock,

4 depending on a reference area.

1 66. The method of claim 64, wherein horizontal and vertical components

2 of each motion vector is separately coded, and wherein an available reference area for

3 the horizontal and vertical components is considered when mapping each motion

4 vector component to a binary codeword.

1 67. A computer program product, embodied in a computer-readable

2 medium, for encoding frames from multiple views having an inter-view prediction

3 relationship, comprising:

4 computer code for encoding into a bitstream a first plurality of

5 macroblocks of a first frame of a picture taken from a first view;

6 computer code for encoding into a bitstream a second plurality of ? macroblocks of a second frame of the picture taken from a second view, at least a e portion of the second frame referencing at least a portion of the first frame; and

9 computer code for signaling in the bitstream an indication of, for each

1 0 macroblock in the second frame, which of the plurality of macroblocks in the first

11 frame may be referenced.

1 68. An encoding apparatus, comprising: z a processor; and

3 a memory unit communicatively connected to the processor and

4 including:

5 computer code for encoding into a bitstream a first plurality of

6 macroblocks of a first frame of a picture taken from a first view;

7 computer code for encoding into a bitstream a second plurality

8 of macroblocks of a second frame of the picture taken from a second view, at least a

9 portion of the second frame referencing at least a portion of the first frame; and

10 computer code for signaling in the bitstream an indication of,

11 for each macroblock in the second frame, which of the plurality of macroblocks in the 2 first frame may be referenced.

1 69. The apparatus of claim 68, wherein the signaling occurs at the

2 sequence parameter set level.

1 70. The apparatus of claim 68, wherein the signaling occurs at the picture

2 parameter set level.

1 71. The apparatus of claim 68, wherein the signaling occurs at the slice

2 header level.

1 72. The apparatus of claim 68, wherein the signaling occurs at the

2 macroblock header level.

1 73. The apparatus of claim 68, wherein the macroblocks in the first frame

? which can be referenced by a particular macroblock in the second frame is based ai

3 least in part upon the arrangement of first and second cameras responsible for the first

4 and second views.

74. The apparatus of claim 68, wherein the area of macroblocks in the first frame that can be referenced by a particular macroblock in the second frame is limited for fractional pixels.

75. The apparatus of claim 68, wherein the encoding order of macroblocks in the first and second frames moves from a top-left comer of each frame towards a bottom-right corner.

76. The apparatus of claim 68, wherein the encoding order of macroblocks in the first and second frames moves from a top-right corner of each frame towards a bottom-left corner.

77. The method of claim 68, wherein the indication includes a number of macroblocks that are to be reconstructed to the reference memory before a notification is transmitted indicating that the decoding of at least one of the second plurality of macroblocks can begin.

78. The method of claim 68, wherein the indication includes a number of notifications to wait before the decoding of at least one of the second plurality of macroblocks can begin.

79. The method of claim 68, wherein a decoding order of macroblocks in the first and second frames is signaled in the bistream.

80. The method of claim 68, wherein coded motion vector information is also encoded into the bitstream, and wherein the coding of motion vectors between the first view and the second view are modified when the statistical properties of motion vectors change as a reference area is restricted.

81. The method of claim 80, wherein a single codeword is used for coding both components of motion vectors between the first view and the second view, and wherein a mapping from motion vectors to codewords is adapted at each macroblock, depending on a reference area.

82. The method of claim 80, wherein horizontal and vertical components of each motion vector is separately coded, and wherein an available reference area for the horizontal and vertical components is considered when mapping each motion vector component to a binary codeword.

Description:

SYSTEM AND METHOD FOR USING PARALLELLY DECODABLE SLICES FOR MULTI- VIEW VIDEO CODING

FIELD OF THE INVENTION

[0001] The present invention relates generally to multi-view video coding and decoding. More particularly, the present invention relates to the implementation of parallel decoder implementation for different views in multi-view systems, even when dependencies exist between views.

BACKGROUND OF THE INVENTION

[0002] This section is intended to provide a background or context to the invention that is recited in the claims. The description herein may include concepts that could be pursued, but are not necessarily ones that have been previously conceived or pursued. Therefore, unless otherwise indicated herein, what is described in this section is not prior art to the description and claims in this application and is not admitted to be prior art by inclusion in this section.

[0003] With recent advances in video capture and display technologies, three- dimensional (3D) video communication and entertainment services are quickly becoming a reality and will enter into consumer domain in the near future. 3D video communication and entertainment services will likely revolutionize the way users enjoy and interact with content and will open the door for many new and exciting services. Such services may include stereoscopic video broadcasting, 3D-TV, and a wide variety of others.

[0004] 3D-TV applications require special 3D displays whose characteristics can vary. For example, if a 3D display is auto-stereoscopic, then users do not need to wear any special lenses or glasses to see different views for each eye. Moreover, a 3D display could support several views to be displayed simultaneously so that user can see different views depending on the user's location with respect to the screen. This feature is referred to as head motion-parallax and is characterized as one of the most real and immersive experience available for users.

[0005] In order to enable these types of applications, new paradigms are required for the representation, transmission and rendering of 3D-video content, which could be very different from typical 2D-video use-cases. Recently, the Joint Video Team (JVT) of the ITU-T/VCEG and ISO/MPEG standardization groups undertook the effort for a Multi-View Video Coding (MVC) standard. MVC is an encoding framework of multi-view sequences which is produced either by a camera system comprising multiple cameras capturing the same event from different locations or by a single capable of capturing a 3D scene. In applications exploiting MVC, the user can enjoy real and immersive experiences, as the multi-view video represents a three dimensional scene in real space. The principle difference between a MVC coder and a traditional 2D-video coder is that MVC exploits inter-view redundancy, in addition to temporal and spatial redundancies. Therefore, in multi-view video coding structures, inter-view dependencies exist between pictures. [0006] Unfortunately, inter-view dependencies between pictures pose serious complexity problems and parallelism issues to a video system. These issues occur because two pictures at different views need to be decoded sequentially. This is especially problematic for 3D-TV use-cases, where many views need to be displayed simultaneously. To understand these issues, it is helpful to consider a 3D-TV system that is simultaneously displaying two views, and where views are coded using the coding structure illustrated in Figure 1. In order to decode a picture in "View-1" at any temporal instant, the picture in "View-0" at the same temporal instant should be decoded first. With this structure, the only way to display two views at the same time is to therefore have the decoder running two times faster than a regular video decoder. Even though two independent decoders running on different platforms might be available, both decoders need to run twice as fast as a regular 2D video decoder. The situation gets worse as the number of views supported by the 3D display increases. Currently, there are commercial displays which can display 100 views simultaneously, and if all of the views depend on each other, then the decoder must ran 1 GO times faster,

{0007] In light of the above, it is clear that parallel decoding of separate views is a crucial requirement for 3D-TV systems. One solution for increasing the parallelism is

to code each view independently. However, it has been learned that this kind of simulcasting approach results in a significant penalty in coding efficiency which is not desirable.

[0008] It would therefore be desirable to develop a system and method for improving the parallel decoding of separate views for 3D-TV systems and other systems while not suffering from the coding efficiency issues described above.

SUMMARY OF THE INVENTION

[0009] Various embodiments of the present invention provide a system and method for enabling parallel decoder implementation for different views, even with existing dependencies between views. This is achieved by encoding views with certain constraints so that any macroblock in a certain view is allowed to depend only on reconstruction values of a subset of macroblocks in other views. The encoding constraints that include information about which MBs that could be referenced in another view are known both by encoder and decoder, either because the constraints are pre-defined or signaled in the bitstream. The macroblock delay between two views can also be signaled to the decoder. The delay can be signaled at the same time as the other constraint information. With this information, the decoding of macroblocks in different views can happen simultaneously, as they are not referencing each other. If an encoder signals the encoding constraints, certain trade-offs between coding efficiency, parallelism and complexity can be realized. With the embodiments of the present inventions, encoders and decoders for different views can run in parallel so that the delay involved in outputting pictures is reduced. [0010] Various embodiments also provide for more efficient entropy coders for coding the motion vectors among views using the parallelly decodable slice (PDS) concept discussed herein. This is accomplished taking advantage of the fact that (1) motion vectors in a PDS arrangement cannot take all possible values due to the restriction of the reference area, and (2) the statistical properties of horizontal and vertical components of motion vectors arc inter-related in PDS. [0011} These and other advantages and features of the invention, together with the organization and manner of operation thereof, will become apparent from the

following detailed description when taken in conjunction with the accompanying drawings, wherein like elements have like numerals throughout the several drawings described below.

BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a representation of a conventional sample prediction chain for two views of a picture in a 3D-TV use case;

[0013] Figure 2 is an overview diagram of a system within which various embodiments of the present invention may be implemented;

[0014] Figure 3 is a representation showing how an embodiment of the present invention is implemented, with a one macroblock delay between two views;

[0015] Figure 4 is a representation showing a sample decoding process for the implementation depicted in Figure 3;

[0016] Figure 5 is a representation of a sliding deblocking operation in accordance with a particular embodiment of the present invention;

[0017] Figure 6 is a representation showing a sample decoding process with the sliding deblocking approach depicted in Figure 5

[0018] Figure 7 is an illustration showing the effect of sub-pixel interpolation during the implementation of various embodiments of the present invention;

[0019] Figure 8 is an illustration showing how padding may be used to address the sub-pixel interpolation issues depicted in Figure 7;

[0020] Figure 9 is an illustration of how coding efficiency can be affected through the implementation of various embodiments of the present invention;

[0021] Figure 10 is a perspective view of an electronic device that can be used in the implementation of the present invention;

[0022] Figure 11 is a schematic representation of the circuitry of the electronic device of Figure 10; and

[0023] Figure 12 is a representation of a reference area showing the ranges for horizontal and vertical components of motion vectors when the parallel decoding of views is permitted.

DETAILED DESCRIPTION OF VARIOUS EMBODIMENTS

[0024] Various embodiments of the present invention provide a system and method for enabling parallel decoder implementation for different views, even with existing dependencies between views. This is achieved by encoding views with certain constraints so that any macroblock in a certain view is allowed to depend only on reconstruction values of a subset of macroblocks in other views. The encoding constraints that include information about which MBs that could be referenced in another view are known both by encoder and decoder, either because the constraints are pre-defined or signaled in the bitstream. With this information, the decoding of macroblocks in different views can happen simultaneously, as they are not referencing each other. If an encoder signals the encoding constraints, certain trade-offs between coding efficiency, parallelism and complexity can be realized. [0025] Figure 2 shows a generic multimedia communications system for use with the present invention. It should be noted that, as discussed herein, a bitstream to be decoded can be received from a remote device located within virtually any type of network. Additionally, the bitstream can be received from local hardware or software. It should also be understood that, although text and examples contained herein may specifically describe an encoding process, one skilled in the art would readily understand that the same concepts and principles may also apply to the corresponding decoding process and vice versa.

[0026] As shown in Figure 2, a data source 100 provides a source signal in an analog, uncompressed digital, or compressed digital format, or any combination of these formats. An encoder 110 encodes the source signal into a coded media bitstream. The encoder 1 10 may be capable of encoding more than one media type, such as audio and video, or more than one encoder 110 may be required to code different media types of the source signal. The encoder 1 10 may also get synthetically produced input, such as graphics and text, or it may be capable of producing coded bitstreams of synthetic media. In the following, only processing cf one coded media bitstream of one media type is considered to simplify the description. It should be noted, however, that typically real-time broadcast services comprise several streams (typically at least one audio, video and text sub-titling

stream). It should also be noted that the system may include many encoders, but in the following only one encoder 110 is considered to simplify the description without a lack of generality.

[0027] The coded media bitstream is transferred to a storage 120. The storage 120 may comprise any type of mass memory to store the coded media bitstream. The format of the coded media bitstream in the storage 120 may be an elementary self- contained bitstream format, or one or more coded media bitstreams may be encapsulated into a container file. Some systems operate "live", i.e. omit storage and transfer coded media bitstream from the encoder 1 10 directly to the sender 130. The coded media bitstream is then transferred to the sender 130, also referred to as the server, on a need basis. The format used in the transmission may be an elementary self-contained bitstream format, a packet stream format, or one or more coded media bitstreams may be encapsulated into a container file. The encoder 110, the storage 120, and the sender 130 may reside in the same physical device or they may be included in separate devices. The encoder 110 and sender 130 may operate with live real-time content, in which case the coded media bitstream is typically not stored permanently, but rather buffered for small periods of time in the content encoder 110 and/ or in the sender 130 to smooth out variations in processing delay, transfer delay, and coded media bitrate.

[0028] The sender 130 sends the coded media bitstream using a communication protocol stack. The stack may include but is not limited to Real-Time Transport Protocol (RTP), User Datagram Protocol (UDP), and Internet Protocol (IP). When the communication protocol stack is packet-oriented, the sender 130 encapsulates the coded media bitstream into packets. For example, when RTP is used, the sender 130 encapsulates the coded media bitstream into RTP packets according to an RTP payload format. Typically, each media type has a dedicated RTP payload format. It should be again noted that a system may contain more than one sender 130. but for the sake of simplicity, the following description only considers one sender 130, f§029] The sender 130 may or may not be connected to a gateway 140 through a communication network. The gateway 140 may perform different types of functions. such as translation of a packet stream according to one communication protocol stack

to another communication protocol stack, merging and forking of data streams, and manipulation of data stream according to the downlink and/ or receiver capabilities, such as controlling the bit rate of the forwarded stream according to prevailing downlink network conditions. Examples of gateways 140 include multipoint conference control units (MCUs), gateways between circuit-switched and packet- switched video telephony, Push-to-talk over Cellular (PoC) servers, IP encapsulators in digital video broadcasting-handheld (DVB-H) systems, or set-top boxes that forward broadcast transmissions locally to home wireless networks. When RTP is used, the gateway 140 is called an RTP mixer and acts as an endpoint of an RTP connection.

[0030] The system includes one or more receivers 150, typically capable of receiving, de-modulating, and de-capsulating the transmitted signal into a coded media bitstream. The coded media bitstream is typically processed further by a decoder 160, whose output is one or more uncompressed media streams. Finally, a renderer 170 may reproduce the uncompressed media streams with a loudspeaker or a display, for example. The receiver 150, decoder 160, and renderer 170 may reside in the same physical device or they may be included in separate devices. [0031] Scalability in terms of bitrate, decoding complexity, and picture size is a desirable property for heterogeneous and error prone environments. This property is desirable in order to counter limitations such as constraints on bit rate, display resolution, network throughput, and computational power in a receiving device. [0032] Communication devices of the present invention may communicate using various transmission technologies including, but not limited to, Code Division Multiple Access (CDMA), Global System for Mobile Communications (GSM), Universal Mobile Telecommunications System (UMTS), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Transmission Control Protocol/ Internet Protocol (TCP'IP), Short Messaging Service (SMS). Multimedia Messaging Service (MMS), e-mail, Instant Messaging Service (IMS), Bluetooth, IEEE 802.11, etc. A communication device may communicate using various media including, but not limited to, radio, infrared, laser, cable connection, and the like. The communication devices may be located in a mode of transportation

including, but not limited to, an automobile, a truck, a taxi, a bus, a boat, an airplane, a bicycle, a motorcycle, etc.

[0033] For the implementation of various embodiments of the present invention, it is helpful to consider the representation depicted in Figure 3. In Figure 3, two frames in View-0 and View-1 are going to be decoded, and View-1 uses View-0 frame as a reference. Conventionally, the decoding of View-1 could start only after the decoding of View-0 has been completed and its reconstruction is available. If the decoding time for View-0 and View-1 are to and Ti respectively, then the total decoding time would be To + Ti because of the sequential process, even if the resources exist to have two decoders.

[0034] Parallel decoding of views is achieved by limiting the reference area that each macroblock in View-1 can use. In this example, macroblocks in View-1 can only reference an area in View-0 that is covered by the co-located macroblock and macroblocks occurring prior to the co-located macroblock. For example, the macroblock at location (2,2) in View-1 cannot reference the shaded area in View-0. In this particular example, two decoders for each view can run in parallel, with a one macroblock delay in between the decoders.

[0035] Figure 4 presents in detail the decoding process for the two views depicted in Figure 3. The decoding process for the first and second views starts simultaneously, as depicted at 400 (for View-0) and 405 (for View-1). The decoding of MB(O 1 O) at the View-1 decoder does not start before the View-0 decoder notifies Vicw-1 decoder that available reference area for MB(0,0) of View-1 is available in the reference memory. While the decoding of MB(O 5 O) of View-0 is being performed, the View-1 decoder can either enter into WAIT state (as depicted at 415), or perform a decoding operation that does not require to access the reference area, such as parsing the bitstream. Once the View-0 decoder decodes the MB(0,0) at 410 and makes the reconstructed macroblock ready in the memory at 412. the View-1 decoder is notified at 425. Once notified, the View-1 decoder can decode IViB(O 3 O), as it does not reference any area in view-0 that is not yet available. (It should be noted that, if the MB(O 5 O) of View-1 is intra coded, then the decoding operation could start before the Vicw-1 decoder receives a notification that MB(O 3 O) of View-0 is reconstructed and

available). With this system, MB(O 3 I) in View-0 and MB(O 5 O) in View-1 could be decoded in parallel, as indicated at 430 in Figure 4. The View-0 decoder updates the memory with the reconstruction values from MB(O 5 I) only after the View-1 decoder completes the decoding of MB(0,0). The View-0 decoder can either wait to be notified by the View-1 decoder, as depicted at 435, or it can continue to decode subsequent macroblocks and store the reference in a temporary buffer, without updating the reference area. The View-1 decoder's notification that the decoding process for MB(O 5 O) has been completed is represented at 440. This process is then repeated for the next macroblock and subsequent macroblocks as necessary. [0036] Using the process depicted in Figure 4, the parallel implementation of two decoders is possible. In such a parallel implementation, the total decoding time equals:

To + delay_overhead + synchronization_overhead

[0037] In this equation, delay _overhead is the time spent waiting for the initial macroblock(s) to be decoded, synchronization ^ overhead is the time spent in the WAIT states for both of the decoders. If the decoding of each macroblock at different views takes different amounts of time, then both decoders need to wait for each other quite often, resulting in a significant amount of synchronization overhead. In addition, although synchronizing between decoders may be simple for certain architectures, it could be quite problematic for certain architectures. It would therefore be advantageous to have fewer synchronization points per frame, but this could also reduce the parallelism. On another note, if the initial delay between views is large (delay ^ overhead is large), coding efficiency of the system would increase because of the increased area available for referencing.

[0038] In the process depicted in Figure 4, only a one-MB delay between two views is discussed. It should be noted that it is possible to have a larger delay, although this would hinder the parallelism while, at the same time, increasing the coding efficiency, as there would be a larger reference area available for each macroblock in View-1. Thus, the macroblock delay between two views could be signaled in the bitstrcam rather than having it pre-defined.

[0039] In light of the design constraints discussed above, syntax elements can be defined so as to define the available reference area for each macroblock, In particular, pds_block_size and pds initial_delay can be used to indicate a block size and initial delay, respectively. The available reference area contains reconstruction samples from a list of macroblocks that are given using the following equation: (-1. J], where j = (pds_initial_delay -t- ( mbλddr / pds_block__size) ) *pds_block_size 1 [0040] A parallel decoder implementation can use pds_block_size information to decide how frequently it needs to synchronize between views. For example, the sample decoding process illustrated on Figure 4 synchronizes two views at every macroblock, which means that pds_block_size is one. The pds_initial_delay is used by a decoder to decide how many pds_blocks (pds_block is a number of macroblocks of its size given by pds_Wock_size) need to be made available by the reference view before decoding can begin. For the example illustrated in Figure 4, the pds_initial_delay is 1.

[0041] The following syntax may be used to implement one embodiment of the present invention:

[0042] In this syntax, a pds_parameters present_flag equal to 1 indicates that slices having view_id equal to i utilize pds_parameters. If a pds_parameters_present_flag equals 0, then slices belonging to viewjd equal to i are not parallelly_decodable slice. A fixed_pds_fbr_all_sequence_flag[i] equal to 1 indicates that all slices belonging to viewed equal to i share the same pds parameters. The parallelly_decodable_slice_flag indicates that the interprediction is limited such that an un-deblockcd reference is used for interview prediction for the current slice, and macroblocks contain only samples from the multiview reference that belong to their corresponding available_reference_area, where available_reference_area include samples only from a group of macroblocks that is defined for each of the macroblocks using the equation

(-1, J], where j = (pds_initial_delay + ( mbAddr / pds_block_size) ) *ρds_block_size - 1 [0043] pds_block_size_minusl_for_list_0[i] plus 1 defines the pds_block_size, in number of macroblocks, for the multiview reference i in list-O. pds_initial_dclay_forJist_0[i] defines the pds initial delay for the multiview reference i in list-O. pds_block_size_minusl for list 1 [i] plus 1 defines the pds_block_size for the multiview reference i in list-1. pds_initial_delay_for_list_l [i] defines the pds_initial_delay for the multiview reference i in list- 1. [0044] Although the basic concept of how parallel decoding operation can be achieved with certain encoding constraints, it should be noted that the system of the various embodiments of the present invention affects the operation of adaptive deblocking and sub-pixel interpolation. With regard to adaptive deblocking, it is noted that in the H.264/ AVC (advanced video coding) standard, an adaptive deblocking filter is utilized to remove the blocking artifacts caused by transform and quantization. The deblocking filter operates on edges across boundaries. If the boundary happens to be between two macroblocks,. then decoded information from neighboring macroblocks is required to filter the edges. This filtering process has to be modified when it is used in conjunction with various embodiments of the present invention. This is due to the fact that, at the time a macroblock is made available for referencing by other views, all of its neighboring macroblocks might not have yet been decoded. In such a case, the boundaries neighboring with macroblocks which are not yet decoded cannot be filtered. There are a number of alternatives to address this issue.

[0045] A first option to address the issue of not-yet decoded macroblocks involves the use of unfiltered reconstruction for inter-view prediction. In this approach, a decoder uses the unfiltered reference when performing inter-view prediction, but regular referencing for temporal prediction. This eliminates the need for accessing neighboring macroblocks. This approach is illustrated in Figure 3. |0046| A second option to address around the issue of not-yet decoded macroblocks involves the use of sliding deblocking. In this approach, before the macroblock is made available as a reference for other views, all of the boundaries except for the right and bottom macroblock boundaries are filtered. This process is generally depicted in Figure 5. Figure 6 shows the detailed implementation of this type of

decoding operation. Like the embodiment depicted in Figure 4, the decoding process for the first and second views starts simultaneously, as depicted at 600 (for View-0) and 605 (for Vievv-1). After starting, the View-1 decoder immediately enters a WAIT state at 615. While the View-1 decoder is in the WAIT state at 615, the View-0 decoder decodes the MB(0,0) at 610 and makes the reconstructed macroblock ready in the memory at 612. At 614, the View-0 decoder filters the entire MB(O 5 O) except for the bottom and right boundaries thereof. Once this is completed, the View-1 decoder is notified at 620. At 630, the View-1 decoder decodes MB(0,0), while the View-0 decoder decodes the next macroblock — MB(0,l) in this instance. Before the View-0 decoder updates the memory with the reconstruction values from MB(0, 1 ), it waits for the View-1 decoder to complete the decoding of MB(0,0). This waiting is represented at 635, and the View-1 decoder' s notification that this process has been completed is represented at 640. This process is then repeated for the next macroblock and subsequent macroblocks as necessary. In addition to the above, an encoder can utilize a modified deblocking operation where macroblock boundaries are filtered without using any decoded information from neighboring macroblocks other than reconstructed samples. This could be signaled in the bitstream to the decoder. [0047] A third option to address the issue of not-yct decoded macroblocks involves further limiting the reference area. With this approach, the area that a macroblock can reference in another view is further limited so that none of the pixels that might change after filtering are allowed for referencing. This approach can also be used in conjunction with sliding deblocking approach discussed above. [0048] Sub-pixel interpolation is performed for blocks whose motion vector has fractional pixel accuracy. In the H.264/AVC standard, motion vectors may have integer, half-pixel or quarter-pixel accuracies. The half-pixel samples are obtained by- using a 6-tap FIR filter, and quarter-pixel samples are obtained by bilinear filtering using the two nearest samples at half or integer pixel positions. The effect of sub- pixel interpolation in terms of different embodiments of the present invention can be understood in view of the depiction in Figure 3. In figure 3, the macroblock at location (2,2), might reference an area in View-1, where it lies within the available area (indicated as white), but still requires pixels from an unavailable area of the

reference frame (indicated as shaded in the reference memory). Figure 7 illustrates the case where the motion vector is (-0.25,-0.25) for the macroblock at location (2,2) for View-1. As seen in Figure 7, some pixels that are required to obtain the reference arc not yet available (i.e., the pixels in the shaded area). In order to solve this problem, a number of different approaches may be used. A first such approach involves padding. In this approach, after the macroblock is put into the reference memory, the bottom and right boundaries are extended by padding. The number of pixels that need to be padded depends on details of the interpolation filter, e.g., the filter length. For example, in H.264/AVC, a padding of three pixels would be needed. Figure 8 shows the padded area in a generic scenario.

[0049] A second approach for addressing the issue of not- yet-available pixels involves further limiting the reference area. With this approach, the area that a macroblock can reference in another view is further limited for fractional pixels. As a result, none of the pixels from unavailable area are used for sub-pixel interpolation. [0050] Because the reference area for inter-view prediction is constrained, the utilization of the various embodiments of the present invention may also result in a certain amount of degradation of coding efficiency. This penalty may become more evident in certain camera arrangements. For example, it is helpful to consider a camera system which includes three cameras that vertically aligned but horizontally spaced as illustrated in Figure 9. Assuming that View-0 is coded as the base view, View-1 frames references View-0 frames and View-2 frames references View-1 frames, if various embodiments of the present invention are used, with a one MB delay between views, then the optimal matching blocks for View-2 and View-1 usually fall in unavailable reference areas, thereby decreasing the coding efficiency. This phenomenon can be addressed in many ways. A first approach involves arranging view dependencies. In this approach, the encoder can arrange the referencing structure between views according to the camera arrangement used to capture the views. With this arrangement the best matching blocks in each view have a greater chance of being in the available reference area. In the example depicted in Figure-9, reversing the order of references (i.e., with View-2 being the base view,

View-1 referencing View-2 and View-0 referencing View-l) would increase the coding efficiency.

[0051] A second option for improving the coding efficiency involves modifying the original picture prior to encoding. In this approach, the encoder uses modified original pictures for certain views and signals the decoder the details of the modifications. For example, if View-2 in the aforementioned example is horizontally mirrored, blocks in View-2 would have a higher chance of finding their corresponding best matches in other views, thereby increasing the coding efficiency. It should be noted that, using this approach, different modifications would be required for different camera arrangements.

[0052] A third option for improving the coding efficiency involves the utilization of slice groups. In this approach, the frames in each view are partitioned into different slice groups according the camera arrangement. As a result, the chance of finding an optimal matching block in other views is maximized. For the example illustrated in Figure 9, one can build slice groups that are horizontally shifted from each other with one or more macroblocks in order to improve the coding efficiency. [0053] A fourth option for improving the coding efficiency involves utilizing a modified raster scan. With this approach, the encoding/decoding order of macroblocks is changed. For the previous example, if View-2 macroblocks start to be encoded/decoded from the top-right corner instead of the top-left corner, the chance of finding best match would increase, hence increasing coding efficiency. It should be noted that, with this approach, different scan orders would be required for different camera arrangements.

[0054] In addition to the above, signaling can also be enabled through the use of supplemental enhancement infoπnation (SEI) message syntax. The elementary unit for the output of an H.264/ AVC-MVC encoder and the input of an H.264/AVC - MVC decoder is a Network Abstraction Layer (NAL) unit. For transport over packet- oriented networks or storage into structured files, NAL units are typically encapsulated into packets or similar structures. In transmission or storage environments that do not provide framing structures, a bytestream format has been specified in the H.264/AVC standard. The bytestream format separates NAL units

from each other by attaching a start code in front of each NAL unit. To avoid false detection of NAL unit boundaries, encoders must run a byte-oriented start code emulation prevention algorithm, which adds an emulation prevention byte to the NAL unit payload if a start code would have otherwise occurred. In order to enable straightforward gateway operation between packet-oriented and stream-oriented systems, start code emulation prevention is always performed, regardless of whether the bytestream format is in use.

[0055] An SEI NAL unit contains one or more SEI messages, which are not required for the decoding of output pictures but assist in related processes, such as picture output timing, rendering, error detection, error concealment, and resource reservation. About 20 SEI messages are currently specified in the H.264/ A VC-MVC standard. User data SEI messages enable organizations and companies to specify SEI messages for their own use. The H.264/ AVC standard contains the syntax and semantics for the specified SEI messages, but no process for handling the messages in the recipient has been conventionally defined. Consequently, encoders are required to follow the H.264/ A VC standard when they create SEI messages, and decoders conforming to the H.264/ AV -MVC standard are not required to process SEI messages for output order conformance.

[0056] One of the reasons to include the syntax and semantics of SEI messages in H.264/AVC is to allow system specifications, such as 3GPP multimedia specifications and DVB specifications, to interpret the supplemental information identically and hence interoperate. It is intended that system specifications can require the use of particular SEI messages both in encoding and in decoding, and the process for handling SEI messages in the recipient may be specified for the application in a system specification. In light of the above, signaling of parallelly decodable slice parameters can be accomplished utilizing SEI messages. In such case, the encoder may want to disable the deblocking filter for pictures where inter-view prediction is performed.

[0057) In addition to the above, there are still other tools which may be used to increase coding efficiency. This is accomplished taking advantage of the fact that (I ) motion vectors in a PDS arrangement cannot take all possible values due to the

restriction of the reference area, and (2) the statistical properties of horizontal and vertical components of motion vectors are inter-related in PDS. The 11.264/ A VC video coding standard supports two different entropy coding arrangements—Context Adaptive Variable Length Coding (CAVLC) and Context Adaptive Binary Arithmetic Coding (CABAC). Exemplary implementations for CAVLC are provided as follows according to various embodiments, but, one skilled in the art would understand that the various embodiments are applicable to CABAC well.

[0058] In H.264/ AVC, motion vectors have two components— horizontal (MV x ) and vertical (MVy) components. Each component of a motion vector is coded separately using a prediction motion vector. In other words, a prediction is first formed, and then the difference between the prediction and the original motion vector is coded for each of the horizontal and vertical components of the motion vector. This difference is mapped to binary Exp-Golomb codewords according to their probability of occurrence, which is defined by the standard. For example, the following table presents the motion vector difference (mvd), and the associated Exp-Golomb codewords.

Codeword MVD

1 n

010 0.25

011 -0.25

00100 0.5

00101 -0.5

001 10 1.0

001 1 1 -1 .0

[0059] The arrangement presented above has two drawbacks when used with the PDS system. First, this arrangement assumes that motion vectors can point anywhere in the reference area (within the limits defined in the H.264/AVC standard). In PDS, however, the reference area is restricted. Second, the H.264/AVC standard assumes that statistical properties of horizontal and vertical components of motion vectors are independent, as they are coded separately. However, in the case of the motion vectors in the PDS system, these two components are inter-related. This point is demonstrated in Figure 12, which shows the ranges for horizontal and vertical components of motion vectors in view of the representation in Figure 12. In Figure

12, it is shown how the vertical component of a motion vector has different ranges depending on the horizontal component of the motion vector. In particular, Figure 12 shows a first vertical motion vector range 1200, a second vertical motion vector range 1210, a first horizontal motion vector range 1220 and a second horizontal motion vector range 1230. When the vertical motion vector is applied after the horizontal motion vector, the first vertical motion vector range 1200 applies when the horizontal motion vector range is less than or equal to zero, and the second vertical motion vector range 1210 applies when the horizontal motion vector range is greater than zero. Similarly, when the vertical motion vector is coded horizontal motion vector, then the first horizontal motion vector range 1220 applies when the vertical motion vector range is less than zero but greater than -16, and the second horizontal motion vector range applies when the vertical motion vector range is less than or equal to -16. [0060] Various arrangements may be used to address the above shortcomings of the H.264/AVC standard. In a first arrangement, a single codeword can be used to code both of the components of the respective motion vector. In addition, the mapping from motion vectors to codewords may be adapted at each macroblock, depending upon the reference area. The coding of motion vectors between views (or disparity vectors) when the statistical properties of motion vectors change as the reference area is restricted. This provides an optimal coding efficiency, but also possesses a high complexity because of the updating process performed for each macroblock. [0061] In another example, the horizontal and vertical components of motion vectors can be separately coded, while also considering the available reference area for each component when mapping the motion vector component to the binary codeword. For example, it is helpful to consider the situation where the motion vector for a designated macroblock 1250 in Figure 12 is to be coded. In this situation, one can assume that the horizontal component of the motion vector is coded first, after which the vertical component is coded (although it may desirable to reverse this order in certain situations). Because MVy follows MVx In the bitstream, MV Y information cannot be utilized to code MVx. For this reason, the largest range that is possible for MVx should be utilized to map MVx to the binary codeword. After MVx is coded, the range that the vertical component vector can take is known. For

49-

example, if MVx is a positive value, then MVy can not take values between [-16, O]. Using this range, the vertical component is mapped to the binary codeword. The mapping of motion vector components to binary codewords essentially involves removing the values that are unavailable because of the reference area constraint, and shifting the remaining possible values up in the codeword list. [0062] The following is an example demonstrating how variable length codes may be adapted in accordance with the above embodiments in a CABAC environment. In this example, it is assumed that the following variable length codes represent a set of motion vector differences, with the following being the only candidate variable length codes:

[0063] In this example, it is also assumed that, because of the parallel decoding restriction, the motion vector (i.e., the motion vector prediction plus the motion vector difference) can only take certain values. The prediction is known to the encoder and the decoder, so both the encoder and the decoder know what values cannot constitute the motion vector difference. If this restriction means that motion vectors A, B and C cannot be used, then the variable length codes could be altered as follows:

[0064] As a result of the above change, the coding values of D, E, F and G comprise fewer bits.

[0065] Figures 10 and 1 1 show one representative electronic device 12 within which the present invention may be implemented. It should be understood, however, that the present invention is not intended to be limited to one particular type of electronic device. The electronic device 12 of Figures 2 and 3 includes a housing 30, a display 32 in the form of a liquid crystal display, a keypad 34, a microphone 36, an ear-piece 38, a battery 40, an infrared port 42, an antenna 44, a smart card 46 in the form of a UICC according to one embodiment of the invention, a card reader 48, radio interface circuitry 52, codec circuitry 54, a controller 56 and a memory 58. Individual circuits and elements are all of a type well known in the art. These various components can be incorporated into devices on virtually any of the devices discussed herein. [0066] The present invention is described in the general context of method steps, which may be implemented in one embodiment by a program product including computer-executable instructions, such as program code, embodied on a computer- readable medium and executed by computers in networked environments. Examples of computer readable mediums can include various types of storage mediums including, but not limited to, electronic device memory units, random access memory (RAM), read only memory (ROM), compact discs (CDs), digital versatile discs (DVDs) and other internal or external storage devices. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Computer-executable instructions, associated data structures, and program modules represent examples of program code for executing steps of the methods disclosed herein. The particular sequence of such executable instructions or associated data structures represents examples of corresponding acts for implementing the functions described in such steps. fθ§67] Software and web implementations of the present invention could be accomplished with standard programming techniques with rale based logic and other logic to accomplish the various database searching steps, correlation steps, comparison steps and decision steps. It should also be noted that the words

"component" and "module," as used herein and in the claims, is intended to encompass implementations using one or more lines of software code, and/or hardware implementations, and/or equipment for receiving manual inputs. [0068J The foregoing description of embodiments of the present invention have been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the present invention to the precise form disclosed, and modifications and variations are possible in light of the above teachings or may be acquired from practice of the present invention. The embodiments were chosen and described in order to explain the principles of the present invention and its practical application to enable one skilled in the art to utilize the present invention in various embodiments and with various modifications as are suited to the particular use contemplated. The features of the embodiments described herein may be combined in all possible combinations of methods, apparatus, modules, systems, and computer program products.