Classifications

• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
  • H04N19/89—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving methods or arrangements for detection of transmission errors at the decoder
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
  • H04N19/89—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving methods or arrangements for detection of transmission errors at the decoder
  • H04N19/895—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving methods or arrangements for detection of transmission errors at the decoder in combination with error concealment
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
  • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
  • H04N19/51—Motion estimation or motion compensation
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
  • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
  • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
  • H04N19/51—Motion estimation or motion compensation
  • H04N19/56—Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N5/00—Details of television systems
  • H04N5/14—Picture signal circuitry for video frequency region
  • H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo

Definitions

• the present invention relates generally to video encoding and decoding and, in particular, to methods and apparatus for error concealment in video encoding and decoding.
• MPEG1 was designed for storing and distributing audio and motion video, with emphasis on video quality. Its features include random access, fast forward and reverse playback. MPEG1 serves as the basis for video compact disks and for many video games. The original channel bandwidth and image resolution for MPEG1 were established based upon the recording media then available. The goal of MPEG1 was the reproduction of recorded digital audio and video using a 1 2 centimeter diameter optical disc with a bit rate of 1 .41 6 Mbps, 1 .1 5 Mbps of which are allocated to video.
• the compressed bit streams generated under the MPEG1 standard implicitly define the decompression algorithms to be used for such bit streams. The compression algorithms, however, can vary within the specifications of the MPEG 1 standard, thereby allowing the possibility of a proprietary advantage in regard to the generation of compressed bit streams.
• MPEG2 A later developed standard known as "MPEG2" extends the basic concepts of MPEG 1 to cover a wider range of applications.
• MPEG2 may also be useful for other applications, such as the storage of full length motion pictures on Digital Video Disk (“DVD”) optical discs, with resolution at least as good as that presently provided by 1 2 inch diameter laser discs.
• the MPEG2 standard relies upon three types of coded pictures. I (“intra”) pictures are fields or frames coded as a standalone still image. Such I pictures allow random access points within a video stream. As such, I pictures should occur about two times per second. I pictures should also be used where scene cuts (such as in a motion picture) occur.
• P predicted pictures are fields or frames coded relative to the nearest previous I or P picture, resulting in forward prediction processing. P pictures allow more compression than I pictures through the use of motion compensation, and also serve as a reference for B pictures and future P pictures.
• B bidirectional pictures are fields or frames that use the most closest (with respect to display order) past and future I or P picture as a reference, resulting in bidirectional prediction. B pictures provide the most compression and increase signal to noise ratio by averaging two pictures.
• a group of pictures (“GOP") is a series of one or more coded pictures which assist in random accessing and editing.
• a GOP value is configurable during the encoding process. Since the I pictures are closer together, the smaller the GOP value, the better the response to movement. The level of compression is, however, lower.
• a GOP In a coded bitstream, a GOP must start with an I picture and may be followed by any number of I, P or B pictures in any order.
• a GOP In display order, a GOP must start with an I or B picture and end with an I or P picture. Thus, the smallest GOP size is a single I picture, with the largest size being unlimited.
• Figure 1 illustrates a simplified block diagram of an MPEG2 encoder 100.
• a video stream consisting of macroblock information and motion compensation information is provided to both a discrete cosine transformer 102 and a motion vector generator 104.
• Each 8 x 8 block (of pixels or error terms) is processed by the discrete cosine transformer 102 to generate an 8 x 8 block of horizontal and vertical frequency coefficients.
• the quantizer 106 quantizes the 8 x 8 block of frequency-domain error coefficients, thereby limiting the number of allowed values.
• the output of quantizer 106 is processed by a zigzag scanner 108, which, starting with DC components, generates a linear stream of quantized frequency coefficients arranged in order of increasing frequency. This produces long runs of consecutive zero coefficients, which are sent to the variable length encoder 1 10.
• the linear stream of quantized frequency-domain error coefficients is first run-length encoded by the variable length encoder 1 10.
• the linear stream of quantized frequency-domain error coefficients is converted into a series of run-amplitude (or run-level) pairs. Each pair indicates the number of zero coefficients and the amplitude of the non-zero coefficient which ends the run.
• variable length encoder 1 10 After the variable length encoder 1 10 encodes the run-level pairs, it then Huffman encodes the run-level pairs.
• the run-level pairs are coded differently depending upon whether the run-level pair is included in a list of commonly-occurring run-level pairs. If the run-level pair being Huffman encoded is on the list of commonly-occurring pairs, then it will be encoded into a predetermined variable length code word which corresponds to the run-level pair. If, on the other hand, the run-level pair is not on the list, then the run-level pair is encoded as a predetermined symbol (such as an escape symbol) followed by a fixed length codes to avoid long code words and to reduce the cost of implementation.
• a predetermined symbol such as an escape symbol
• the run-length encoded and Huffman encoded output of the variable length encoder 1 10 provides a coded video bitstream.
• Picture type determination circuit 1 1 2 determines whether the frame being encoded is a P picture, an I picture or a B picture. In the case of a P or I picture, picture type determination circuit 1 10 causes the motion vector generator 104 to generate an appropriate motion vector which is then provided to variable length encoder 1 10. Such motion vector is then coded and combined with the output of variable length encoder 1 10.
• Motion compensation improves compression of P and B pictures by removing temporal redundancies between pictures.
• MPEG 2 it operates at the macroblock level.
• a previous frame 200 contains, among other macroblocks, a macroblock 202 consisting of 1 6 pixels (also referred to as "pels") by 1 6 lines.
• Motion compensation relies on the fact that, except for scene cuts, most images remain in the same location from frame to frame, whereas others move only a short distance.
• a macroblock 300 of a current frame 302 can be represented by the macroblock 202 (of Figure 2) as modified by a two dimensional motion vector 304. It is to be understood that the macroblock 300 may or may not be within the same boundaries surrounding macroblock 202 in the previous frame 200.
• a macroblock After a macroblock has been compressed using motion compensation, it contains both the prediction (commonly referred to as “motion vectors”) and temporal difference (commonly referred to as “error terms") between the reference macroblock and the macroblock being coded.
• the decoded (coded) video bit stream is, generally, sufficiently error free so as to not require additional techniques to compensate for errors in the decoded video bit stream.
• a coded video bit stream is typically referred to as a "program stream.”
• program stream When the coded video bitstream output from variable length encoder 1 10 is transported by, for example, satellite or cable transmission systems, either directly from variable length encoder 1 10 or from a recording medium onto which the coded video bitstream has been recorded, the probability of errors in the decoded video bitstream increases.
• a coded bitstream is typically referred to as a "transport stream. "
• error concealment aims to generate data which can be substituted for the lost or corrupt data, where any discrepancies in image created by the generated data (generally at the macroblock level) are not likely to be perceived by a viewer of a video image which relies upon such error concealment.
• an apparatus for concealing errors includes a detector for detecting the presence of an error in data representing the current macroblock, a system for estimating the at least one motion vector based upon a difference between a forward reference frame at the current macroblock and a decoded motion vector for the forward reference frame at the current macroblock, and a system for estimating the current macroblock based upon the estimated at least one motion vector.
• a method for concealing errors includes the steps of detecting the presence of an error in data representing the current macroblock, estimating the at least one motion vector based upon a difference between a forward reference frame at the current macroblock and a decoded motion vector for the forward reference frame at the current macroblock, and estimating the current macroblock based upon the estimated at least one motion vector.
• Figure 1 is a simplified block diagram of a MPEG 2 video encoder.
• Figure 2 is an illustration of a macroblock within a previous frame.
• Figure 3 is an illustration of a macroblock within a current frame.
• Figure 4 is simplified block diagram of a MPEG 2 video decoder of the present invention.
• Figure 5 is a block diagram of a motion compensation system of the present invention.
• Figure 6 is a state diagram which illustrates reference block fetch control of the address generation and control unit of Figure 5.
• Figure 7 is a flow chart of a method for estimating macroblocks in accordance with the present invention.
• Figure 8 is a flow chart of a method for estimating motion vectors in the temporal domain in accordance with the present invention.
• Figure 9 is a flow chart of a method for estimating motion vectors in the spatial domain in accordance with the present invention.
• Figure 10 is a flow chart of a method for macroblock estimation utilizing estimated motion vectors.
• Figure 1 1 is a flow chart of a method for macroblock estimation without the use of estimated motion vectors.
• Decoder 400 utilizes two internal busses, a GBUS 402 and an RBUS 404.
• GBUS 402 is a 64 bit bus which is utilized for data transfer between DRAM 406 and specific blocks of decoder 400 which are described below.
• DRAM 406 is a static dynamic random access memory, although other types of memories may be utilized.
• RBUS 404 is an 8 bit but used primarily for control of specific blocks through reduced instruction set computing ("RISC”) CPU 408.
• RISC CPU reduced instruction set computing
• Decoder 400 which is coupled to both GBUS 402 and RBUS 404, operates to control the functionality of specific blocks, as more particularly described below, as well as performing a portion of video bitstream decoding.
• Decoder 400 includes a demultiplexer 410 which is coupled to both GBUS 402 and RBUS 404.
• a video decoder 41 2, an audio decoder 414, a host interface 41 6, a letter box unit 41 8, and a sub picture/vertical blanking interval decoder 420 are each coupled to both GBUS 402 and RBUS 404.
• Audio clock generator 428 outputs a clock signal ACLK.
• a memory controller 430 is coupled to GBUS 402.
• a clock generator 432 which provides a clock signal SCLK, is coupled to host interface 41 6.
• An output of letter box unit 41 8 is provided to video post filter on screen display system 426.
• Sub picture/vertical blanking interval decoder 420 is coupled to video post filter on screen display system 426, which system provides its output to NTSC/PAL encoder 424.
• Sub picture/vertical blanking interval decoder 420 is coupled to video post filter on screen display system 426.
• a host processor 434 interfaces with host interface 41 6.
• sub picture/vertical blanking interval decoder 420 and letter box unit 41 8 are hardwired units.
• Letter box unit 41 8 performs a 4-tap vertical filtering and sub-sampling of a video bit stream provided through GBUS 402 and operates to control the video post filter/on screen display system
• Sub picture/vertical blanking interval decoder 420 operates to decode sub picture (“SP") and vertical blanking interval ("VBI") information in the video bit stream.
• SP sub picture
• VBI vertical blanking interval
• a sub picture bitstream consists of subtitles or menu items. For example, this would include karaoke and menu highlighting.
• the functionality for decoding both types of bitstreams is incorporated into a single sub picture/vertical blanking interval decoder 420.
• decoding of the VBI bit stream occurs during the vertical blanking period, while SP bitstream decoding occurs during active display periods.
• the sub picture/vertical blanking interval decoder 420 decodes and displays on screen display ("OSD") bitstreams.
• OSD bitstreams are instead decoded by video post filter on screen display system 426.
• RISC CPU 408 operates to parse the video bitstream in order to control the decoder 400. RISC CPU 408 also partially decodes the video bitstream (for example, decoding of top-level data such as headers) and also controls various of the other units within decoder
• sub picture/video blanking interval decoder 420 sub picture/video blanking interval decoder 400 through RBUS 404. A portion of the parsing is also performed by sub picture/video blanking interval decoder 420.
• RISC CPU 408 can be utilized to change the position of an SP window through RBUS 404.
• a user can move the SP window up or down through a command to CPU 404 with a Y coordinate as a parameter.
• Letter box unit 41 8 is essentially a vertical decimation filter with downloadable coefficients. Letter box unit 41 8 operates to decimate an active area of a frame which has a ratio of 4:3.
• letter box unit 41 8 converts a
• letter box unit 41 8 converts a 720 x 480 frame to a 720 x 360 frame.
• the active picture area is centered with respect to a display area.
• Host processor 434 and RISC CPU 408 utilize DRAM 406 to exchange messages, commands and status information.
• processor 434 and CPU 408 have the capability to interrupt each other.
• CPU 408 provides a host command parser to execute such commands from host processor 434.
• a typical sequence of events during execution of a command by host processor 434 is:
• Host processor 434 writes a command to DRAM 406 and interrupts CPU 408. 2.
• CPU 408 reads the command and parameters from DRAM 406.
• CPU 408 acknowledges the command by writing a status variable to DRAM 406. 4.
• Command parser of CPU 408 parses the command and executes it. 5.
• CPU 408 interrupts host processor 434 upon completion of the command to report status.
• CPU 408 polls a DRAM command buffer (not shown) for every field sync.
• This buffer is a ring buffer where a write pointer is maintained by host processor 434 while a read pointer is maintained by CPU 408.
• Video decoder 41 2 contains an inverse cosine discrete transformer, a variable length decoder 436, a motion compensation unit 438 and an inverse discrete cosine transformer 440.
• Video decoder 41 2 decodes a coded video data stream received through GBUS 402 and provides a decoded stream to NTSC/PAL encoder 424 through RBUS 404.
• NTSC/PAL encoder converts the decoded stream into an analog signal suitable for display on a television monitor having NTSC and/or PAL signal inputs.
• Demultiplexer 410 operates on data entering decoder 400.
• data is in the form of packets, and includes audio, video and other streams of multiplexed packets.
• Demultiplexer 41 0 selects desired audio packets, video packets and other desired information packets, but rejects the other packets within the video bitstream. For example, audio packets representing audio in several different languages may be present in the video bitstream.
• demultiplexer 41 0 selects only those audio packets corresponding to that language which is selected for presentation with the corresponding video packets.
• Host interface 41 6 provides a glueless interface for host processor 434.
• RBUS controller 422 sends out messages on RBUS 404 and acts as an arbitrator for RBUS 404.
• the motion compensation unit 500 includes an address generation and control unit 502.
• the address generation and control unit 502 corresponds to the memory controller 430 of Figure 4.
• the address generation and control unit 502 accepts motion vectors from variable length decoder 436 and calculates a starting address of a reference macroblock.
• the address generation and control unit 502 issues a data transfer request to the memory controller unit 430.
• data transfer occurs in 64 bit (8 byte) segments at addresses aligned at 8-byte boundaries.
• this data returns from the DRAM 406, the data are latched within the motion compensation unit 500.
• Each 8 bit element of these latched data is then run through horizontal and vertical half-pel filters 504, and the resulting data is stored in the prediction RAM (random access memory) 506.
• the motion compensation unit sits idle.
• prediction data is required for reconstruction by the reconstruction unit 508 of decoded picture data.
• the predicted data is obtained by averaging two such predictions, that is, the output of the half-pel filters at the time and a value from a prediction RAM 506 that was stored after a forward prediction.
• the reconstruction unit 508 supports this averaging of the half-pel filters 504.
• An estimation RAM 510 holds coefficient data transformed in the inverse discrete cosine transformer 440. Reconstruction of each picture starts once the estimation RAM 510 is full.
• the motion compensation unit 500 issues a data transfer request and begins reconstruction.
• the reconstruction basically consists of adding signed numbers from the output of the inverse discrete cosine transformer stored in the estimation RAM 510 to the outputs (stored in the prediction RAM 506) of the half-pel filters 504 for non-intra blocks. For intra-blocks however, the addition is not required. In such a case, the adder output is clipped before it is latched at the output of the reconstruction unit 508 when reconstruction of the picture occurs.
• a state machine 600 which represents the functionality of the address generation and control unit 502 in regard to the transfer of reference picture data from DRAM 406 and the construction of a macroblock is now explained. From a start state 602, a state machine 600 proceeds to a get address state 604. If no motion compensation is to be used to construct the macroblock, state machine 600 proceeds to a yO wait state 606. If only backward motion compensation is to be utilized, then state machine 600 proceeds to state 608 to get or fetch a previous macroblock b which will serve as a reference macroblock. If however forward motion compensation is to be utilized, then state machine 600 proceeds to state 610 to get or fetch a forward macroblock f which will serve as a reference macroblock.
• State machine 600 then proceeds to the yO wait state 606. If the macroblock to be constructed is to be based upon both the forward macroblock f and the previous macroblock b, then state machine 600 proceeds from state 610 to state 608 to also get or fetch a previous macroblock. In such an instance, both the forward and the previous macroblock will serve as reference macroblocks.
• state machine 600 waits for luminance data to be received in regard to the reference macroblock or macroblocks.
• state 61 2 the luminance portion of the macroblock to be constructed is reconstructed.
• state machine waits for chrominance data to be received in regard to the reference macroblock or macroblocks.
• state 61 8 reconstruction of the chrominance portion of the macroblock to be constructed occurs.
• state 620 proceeds to state 620 to await an instruction to construct a new macroblock.
• state machine 600 Similar to the case of the previously constructed macroblock, state machine 600 then proceeds to a get address 1 state 622. if no motion compensation is to be used to construct the macroblock, state machine 600 proceeds to a yl wait state 624. If only backward motion compensation is to be utilized, then state machine 600 proceeds to state 626 to get or fetch a previous macroblock b1 which will serve as a reference macroblock. If, however, forward motion compensation is to be utilized, then state machine 600 proceeds to state 628 to get or fetch a forward macroblock f1 which will serve as a reference macroblock. State machine 600 then proceeds to the y1_wait state 624. If the new macroblock to be constructed is to be based upon both the forward macroblock f 1 and the previous macroblock b1 , state machine 600 proceeds from state
• state machine 600 waits for luminance data to be received in regard to the reference macroblock or macroblocks.
• the luminance portion of the macroblock to be constructed is reconstructed.
• state machine waits for chrominance data to be received in regard to the reference macroblock or macroblocks.
• state machine 600 proceeds to back to start state 602.
• state machine 600 waits until the motion vector FIFO memory of the variable length decoder 436 is not empty.
• the address generation and control unit 502 then generates a request for a motion vector. Two consecutive requests, one for X (horizontal) and one for Y (vertical) components of the motion vectors are made.
• the address generation and control unit 502 obtains both components of the motion vector, the address of the reference block is calculated. The address generation and control unit 502 then sends a request for data transfer to the memory controller unit.
• a motion vector points to a sub-pixel location instead of to an exact pixel location, in order to more accurately represent a P or B picture, it is necessary to generate half-pixel (half-pel) data.
• the smallest unit of concealment is a slice.
• a slice consists of a series of sequential macroblocks.
• motion vectors are estimated using either temporal prediction or spatial prediction.
• spatial prediction pixels from a successfully decoded macroblock are copied for use in decoding the macroblock having a data error.
• motion vectors from a successfully decoded macroblock are utilized to predict a new motion vector field in order to decode the macroblock having a data error.
• the basic concept is that if there is a motion of an object from a frame K-2 (that is, two frames prior to frame K), one can assume that this motion will most likely continue from frame K-2 up through frame K. Therefore, the assumption is that the motion will be basically linear. Based upon that assumption, the present invention estimates pixels and motion vectors, the estimation method depending upon the data available for such estimation.
• the motion compensation unit 438 first attempts to estimate motion vectors in the temporal domain at step 702.
• Figure 8 illustrates such a method.
• the algorithm starts at step 800.
• the motion compensation unit 438 determines whether a decoded motion vector for a forward reference frame at the macroblock positioned by a vector p is available. This motion vector is designated as
• step 804 indicates a failed attempt. If a decoded motion vector for a forward reference frame at the macroblock positioned by the vector p is available, the algorithm proceeds to step 806, which determines whether a decoded motion vector is available for the difference between ( 1 ) a forward reference frame at the macroblock positioned by the vector p; and (2) a decoded motion vector for a forward reference frame at the macroblock positioned by the vector p, where such decoded motion vector is designated by MV(k-m, p).
• step 804 the algorithm proceeds to step 804 to indicate a failed attempt. If available, the algorithm proceeds to step 808 at which an estimated motion vector for a current frame, the k-th frame, at the macroblock positioned by the vector p is determined. Such estimated motion vector is taken to be equal to the difference between ( 1 ) a forward reference frame at the macroblock positioned by the vector p; and (2) a decoded motion vector for a forward reference frame at the macroblock positioned by the vector p. The algorithm then proceeds to step 810 which indicates a successful motion vector estimation in the temporal domain.
• step 704 it is determined whether motion vector estimation in the temporal domain was successful. If so, the algorithm proceeds to step 706, where based upon the estimated motion vector, the motion vector to be used for estimating the subject macroblock is updated. If the motion vector estimation in the temporal domain was not successful, the algorithm proceeds to step 708, where motion vector estimation is performed in the spatial domain. The algorithm for such estimation is shown in
• FIG. 9 The algorithm starts at step 900 of Figure 9 and proceeds to step 902, where it is determined whether a decoded motion vector for the macroblock located immediately above the estimating macroblock is available. Such a motion vector is designated by MV(k,p-( 1 ,0)). If not, a failure is indicated at step 904.
• step 906 the motion vector for the current frame, the k-th frame at the macroblock positioned by the vector p, ⁇ MV(k, p)), is estimated to be equal to the decoded motion vector, the macroblock located immediately above the estimating macroblock, MV(k,j ( 1 ,0)), where ( 1 ,0) is a vector indicating a row index as 1 and a column index as 0.
• step 908 indicates a successive motion vector estimation in the spatial domain.
• step 908 the motion vector for the current macroblock is updated at step 706. Then, at step 71 2, the current macroblock is estimated using the just estimated motion vector, whether that motion vector is estimated in the temporal domain at step 702 or, in the spatial domain, at step 708.
• step 1000 macroblock estimation with the estimated motion vector is started.
• step 1 002 the estimated macroblock for the current frame, the k-th frame, at the macroblock positioned by the estimated motion vector p, ⁇ MB(k, p " ), is estimated to be equal to the decoded macroblock of the difference of ( 1 ) the forward reference frame at the macroblock positioned by the vector p; and (2) the estimated motion vector for the current frame, the k-th frame, at the macroblock positioned by the vector p.
• This decoded macroblock is designated as MB(k-m, p — MV(k, f3)), where m is the frame index difference between the current frame and a forward reference frame.
• Step 71 6 the current macroblock is estimated without the use of an estimated motion vector.
• Step 71 6 is detailed in Figure 1 1 .
• macroblock estimation without use of an estimated motion vector starts at step 1 100.
• step 1 102 it is determined whether the macroblock for the frame preceding the current frame (the k-th frame being the current frame) positioned by the vector p, MB(k-1 , p is available. If such macroblock is available, then, at step 1 104, the current macroblock positioned by the vector p is estimated to be equal to the macroblock for the frame preceding the current frame positioned by the vector p.
• the algorithm is then completed as indicated at step 714.
• step 1 106 it is determined whether the macroblock for the current frame positioned by the vector p but indexed by minus 1 row and in the same column, MB(k, f ( 1 ,0) is available, where ( 1 ,0) is a vector indicating a row index as 1 and a column index as 0.
• the current macroblock (for the current frame, the k-th frame, positioned by the vector p) is estimated to be equal to the macroblock for the current frame positioned by the vector p but indexed by minus 1 row and in the same column, MP(k,j ( 1 ,0)).
• the algorithm is then completed as indicated at step 714.
• step 1 1 10 it is then determined whether the decoded macroblock for the macroblock located immediately above the macroblock to be estimated, MB(k,p + (1 ,0)), is available, where (1 ,0) is a vector indicating a row index as 1 and a column index as 0.
• the estimated macroblock for the current frame, the k-th frame, at the macroblock positioned by the vector p is estimated to be equal to such decoded macroblock for the macroblock located immediately above the macroblock to be estimated, MB(k, p ( 1 ,0)).
• the algorithm is then completed as indicated at step 714.
• the macroblock estimation without an estimated motion vector fails, as indicated at step 1 1 14. In this case, the macroblock can be left blank.
• the present invention has been described in relation to decoding of a coded video bit stream, the present invention is also applicable to the coding of a video bit stream, where an error is detected during or after coding and the error is concealed prior to recording or transport.

Landscapes

• Engineering & Computer Science (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Compression Or Coding Systems Of Tv Signals (AREA)
• Picture Signal Circuits (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Publications (1)

Family

ID=25221814

Family Applications (1)

Country Status (7)

Cited By (3)

Families Citing this family (51)

Citations (4)

Family Cites Families (20)

• 1997
  • 1997-03-13 US US08/816,867 patent/US6078616A/en not_active Expired - Fee Related
• 1998
  • 1998-03-06 DE DE19882177T patent/DE19882177T1/de not_active Ceased
  • 1998-03-06 JP JP53964698A patent/JP2001514830A/ja active Pending
  • 1998-03-06 AU AU63478/98A patent/AU6347898A/en not_active Abandoned
  • 1998-03-06 CN CN98805118A patent/CN1256048A/zh active Pending
  • 1998-03-06 WO PCT/US1998/004497 patent/WO1998041028A1/en not_active Ceased
  • 1998-03-06 KR KR1019997007831A patent/KR100547095B1/ko not_active Expired - Fee Related
• 2000
  • 2000-04-11 US US09/546,782 patent/US6449311B1/en not_active Expired - Fee Related
  • 2000-04-11 US US09/546,955 patent/US6175597B1/en not_active Expired - Fee Related
  • 2000-04-11 US US09/546,954 patent/US6285715B1/en not_active Expired - Fee Related

Patent Citations (4)

Non-Patent Citations (2)

Also Published As

Similar Documents

Legal Events