Image analysis – Image compression or coding
Reexamination Certificate
1999-11-02
2004-06-01
Boudreau, Leo (Department: 2621)
Image analysis
Image compression or coding
C236S10100B
Reexamination Certificate
active
06744924
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates to error concealment in a video signal and to a method and apparatus therefore.
One of the recent targets in mobile telecommunications has been to increase the speed of the data transmission in order to enable incorporation of multimedia services to mobile networks. One of the key components of multimedia is digital video. Transmission of video comprises a continuous traffic of data representing moving pictures. As is generally known, the amount of data needed to transfer pictures is high compared to many other types of media, and so far usage of video in low bit-rate terminals has been negligible. However, significant progress has been achieved in the area of low bit-rate video compression. Acceptable video quality can be obtained at bit-rates around 20 kilo bits per second. As a result of this progressive reduction in bit-rate, it is expected that video is shortly going to become a viable service to offer over channels such as mobile channels.
A video sequence consists of a series of still images or frames. Video compression methods are based on reducing the redundancy and perceptually irrelevant parts of video sequences. The redundancy in video sequences can be categorised into spatial, temporal and spectral redundancy. Spatial redundancy means the correlation between neighbouring pixels within a frame. Temporal redundancy means the correlation between areas of successive frames. Temporal redundancy arises from the likelihood of objects appearing in a previous image appearing in the current image too. Compression can be achieved by generating motion compensation data which describes the motion (i.e. displacement) between similar areas of the current and a previous image. The current image is thus predicted from the previous one. Spectral redundancy means the correlation between the different colour components of the same image.
However, sufficient compression cannot usually be achieved by just reducing the redundancy of the sequence. Thus, video encoders try to reduce the quality of those parts of the video sequence which are subjectively the least important. In addition, the redundancy of the encoded bitstream is reduced by means of efficient lossless coding of compression parameters and coefficients. The main technique is to use variable length codes.
Video compression methods typically differentiate images which can or cannot utilise temporal redundancy reduction. Compressed images which do not utilise temporal redundancy reduction methods are usually called INTRA or I-frames whereas temporally predicted images are called INTER or P-frames. In the INTER frame case, the predicted (motion-compensated) image is rarely precise enough and therefore a spatially compressed prediction error image is also associated with each INTER frame.
Compressed video is easily corrupted by transmission errors, mainly for two reasons. Firstly, due to utilisation of temporal predictive differential coding (INTER frames), an error is propagated both spatially and temporally. In practice, this means that once an error occurs, it is easily visible to the human eye for a relatively long time. Especially susceptible are transmissions at low bit-rates where there are only a few INTRA-coded frames (the transmission of INTRA-coded frames would stop the temporal error propagation). Secondly, the use of variable length codes increases the susceptibility to errors. When a bit error alters the codeword to another one of different length, the decoder will lose codeword synchronisation and also decode subsequent error-free codewords (comprising several bits) incorrectly until the next synchronisation code. (A synchronisation code is a bit pattern which cannot be generated from any legal combination of other codewords.) Every bit in a compressed video bitstream does not have an equal importance to the decompressed images. Some bits belong to segments defining vital information such as picture type (e.g. INTRA or INTER), quantiser value and optional coding modes that have been used. In H.263, the most vital information is gathered in the picture header. A transmission error in the picture header typically causes a total misinterpretation of the subsequent bits defining the picture content. Due to utilisation of temporal predictive differential coding (INTER frames), the error is propagated both spatially and temporally. Thus, a normal approach to picture header corruption is to freeze the previous picture on the screen, to send an INTRA picture request to the transmitting terminal and to wait for the requested INTRA frame. This causes an annoying pause in the received video.
Transmission errors have a different nature depending on the underlying network. In packet-switched networks, transmission errors are typically packet losses (due to congestion in network elements). In circuit-switched networks, transmission errors are typically bit errors where a ‘1’ is corrupted to ‘0’ or vice versa.
To impede degradations in images introduced by transmission errors, retransmissions can be used, error detection and/or error correction methods can be applied, and/or effects from the received corrupted data can be concealed. Normally retransmission provides a reasonable way to protect video data streams from errors, but large round-trip delays associated with low bit-rate transmission and moderate or high error rates make it practically impossible to use retransmission, especially with real-time videophone applications. Error detection and correction methods usually require large overhead since they add some redundancy to the data. Consequently, for low bit-rate applications, error concealment can be considered as a preferred way to protect and recover images from transmission errors. Video error concealment methods are typically applicable to transmission errors occurring through packet loss and bit corruption.
H.263 is an ITU-T recommendation of video coding for low bit-rate communication which generally means data rates below 64 kbps. The recommendation specifies the bitstream syntax and the decoding of the bitstream. Currently, there are two versions of H.263. Version
1
consists of the core algorithm and four optional coding modes. H.263 version
2
is an extension of version
1
providing twelve new negotiable coding modes.
Pictures are coded as luminance (Y) and two colour difference (chrominance) components (C
B
and C
R
). The chrominance pictures are sampled at half the resolution of the luminance picture along both co-ordinate axes. Picture data is coded on a block-by-block basis, each block representing 8×8 pixels of luminance or chrominance.
Each coded picture, as well as the corresponding coded bitstream, is arranged in a hierarchical structure with four layers, which are from bottom to top: block layer, macroblock layer, picture segment layer and picture layer. The picture segment layer can either be arranged as a group of blocks or a slice.
Block layer data consists of uniformly quantised discrete cosine transform coefficients, which are scanned in zigzag order, processed with a run-length encoder and coded with variable length codes.
Each macroblock relates to 16×16 pixels of luminance and the spatially corresponding 8×8 pixels of chrominance components. In other words, a macroblock consists of four 8×8 luminance blocks and the two spatially corresponding 8×8 colour difference blocks. Each INTER macroblock is associated with a motion vector which defines the position of a corresponding area in the reference frame which resembles the pixels of the INTER macroblock. The INTER macroblock data comprises coded prediction error data for the pixels of the macroblock.
Usually, each picture is divided into groups of blocks (GOBs). A group of blocks (GOB) typically comprises 33 macroblocks (arranged as 3 rows of 11 macroblocks). Data for each GOB consists of an optional GOB header followed by data for the macroblocks within the GOB.
If the optional slice structured mode is used, each picture is divided into slices instead of GOBs. A
Hannuksela Miska
Hourunranta Ari Johannes
Antonelli Terry Stout & Kraus LLP
Boudreau Leo
Dang Duy M.
Nokia Mobile Phones Limited
LandOfFree
Error concealment in a video signal does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Error concealment in a video signal, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Error concealment in a video signal will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3363896