USRE43649E1 - Progressive block-based coding for image compression - Google Patents

Progressive block-based coding for image compression Download PDF

Info

Publication number
USRE43649E1
USRE43649E1 US11/119,499 US11949905A USRE43649E US RE43649 E1 USRE43649 E1 US RE43649E1 US 11949905 A US11949905 A US 11949905A US RE43649 E USRE43649 E US RE43649E
Authority
US
United States
Prior art keywords
coefficients
mask
block
group
bits
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
US11/119,499
Inventor
Donald Martin Monro
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zarbana Digital Fund LLC
Hanger Solutions LLC
Original Assignee
Ayscough Visuals LLC
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Ayscough Visuals LLC filed Critical Ayscough Visuals LLC
Priority to US11/119,499 priority Critical patent/USRE43649E1/en
Assigned to AYSCOUGH VISUALS LLC reassignment AYSCOUGH VISUALS LLC ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: MONRO, DONALD M.
Application granted granted Critical
Publication of USRE43649E1 publication Critical patent/USRE43649E1/en
Assigned to ZARBAÑA DIGITAL FUND LLC reassignment ZARBAÑA DIGITAL FUND LLC MERGER (SEE DOCUMENT FOR DETAILS). Assignors: AYSCOUGH VISUALS LLC
Anticipated expiration legal-status Critical
Assigned to HANGER SOLUTIONS, LLC reassignment HANGER SOLUTIONS, LLC ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: INTELLECTUAL VENTURES ASSETS 161 LLC
Expired - Lifetime legal-status Critical Current

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/30Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using hierarchical techniques, e.g. scalability

Definitions

  • the present invention relates to image compression and particularly, although not exclusively, to a progressive block-based embedded DCT coder, and to a method of encoding.
  • the JPEG baseline method for still image coding uses the Discrete Cosine Transform (DCT) in a fixed 8 ⁇ 8 pixel partition.
  • DCT Discrete Cosine Transform
  • zig-zag scanning of DCT coefficients the redundancy and band width characteristics of the DCT are exploited over a range of compressions.
  • the JPEG coder is not particularly efficient at higher compression ratios, and other methods such as wavelets have produced better results while having the advantage of being fully embedded.
  • Some researchers have also attempted to combine DCT with zerotree quantization, usually associated with wavelet transforms: see Xiong, Guleryuz and Orchard, ‘A DCT-Based Image Coder’, IEEE Sig. Proc. Lett., Vol 3, No 11, November 1996, p289.
  • a method of image compression comprising:
  • Such a method could also be applied using the one-dimensional DCT to audio recording.
  • a coder for encoding images comprising:
  • the encoder provides for significance switching of DCT coefficients in block-based embedded DCT image compression.
  • the encoder provides output on one or more data streams that may be terminated within a few bits of any point.
  • the invention also extends to a video coder/decoder including a coder as claimed in claim 17 and an associated decoder, the decoder being arranged to maintain a running record, as transmission between the coder and the decoder proceeds, of the coefficients which are currently significant.
  • the preferred two-dimensional block transform of the present invention is the Discrete Cosine Transform, although other transforms such as the Fast Fourier Transform (FFT) or the Lapped Orthogonal Transform could be used.
  • FFT Fast Fourier Transform
  • Lapped Orthogonal Transform could be used.
  • the order of transmission is not significant. It is therefore to be understood, except where logic requires, that the various parts of the method are not necessarily carried out sequentially in the order specified in section (d) of claim 1 .
  • the bits from the newly selected coefficients could be transmitted either before or after the bits of those coefficients which were previously flagged on an earlier path.
  • the bits of those coefficients which were previously flagged on an earlier path could be transmitted either before or after the new coefficients are selected for flagging.
  • bit planes are swept consecutively from the most significant bit plane to the least significant bit plane. This may either be repeated separately for each image block, or alternatively all blocks may be dealt with at the first bit plane, then all blocks dealt with at the second bit plane, and so on.
  • the philosophy of significance switching, as used in the present invention, is that the overheads introduced will be compensated for by the savings in not transmitting bits for small coefficients until they are switched on. Good performance might naturally be expected at high compression ratios, but what is surprising is the excellent performance both for lossless compression and for compression at low ratios.
  • the coder of the present invention is preferably embedded, in other words the bit stream can be stopped within a few bits of any point while still guaranteeing the least possible distortion overall. When used with an appropriate decoder, either coder or decoder can terminate the bit stream as needed, dependent upon the available bandwidth or the available bit budget.
  • the coder of the present invention has been found to out-perform the base-line JPEG method in peak signal to noise ratio (PSNR) at any compression ratio, and is similar to state-of-the-art wavelet coders.
  • PSNR peak signal to noise ratio
  • the results show that block transform (for example DCT) coding is competitive across the whole range of compression ratios, including lossless, so that a significance-switched block coder would be capable of meeting the requirements of future image compression standards in an evolutionary manner.
  • FIG. 1 illustrates the DCT transformation and digitization of image blocks
  • FIG. 2 illustrates the significance sweeps made through the individual bit planes in coefficient space
  • FIG. 3 illustrates the zig-zag ordering of coefficients, and use of the binary mask.
  • the image to be encoded is first partitioned into a plurality of square image blocks.
  • the partitioning may either be by way of a regular tiling, for example of 8 ⁇ 8 pixel blocks, or alternatively some more complex tiling using blocks of differing sizes.
  • One convenient method of tiling is to vary the block size across the image according to the power in the image (measured by the sum of the squares of pixel intensities).
  • each image block is then transformed using the 2D spatial DCT (Discrete Cosine Transform) to produce a corresponding DCT coefficient block of the same size as the original image block.
  • Each of the coefficients within the coefficient block is then digitized and bitwise encoded as shown in FIG.
  • Each coefficient 10 within the coefficient block 20 is therefore represented as a series of individual bits 30 , ranging from the most significant bit (MSB) to the least significant bit (LSB).
  • MSB most significant bit
  • LSB least significant bit
  • the DCT transformation and digitization preferably follow the scaling used within the JPEG standard: see Wallace, G. K, “The JPEG Still Picture Compression Standard”, Comm. ACCM, Vol. 34, No. 4, pp.30-44, 1991. This scaling represents a compromise which accommodates the full range of possible DCT coefficients with reasonably safe integer roundings.
  • the encoded integer coefficients are now required to be manipulated in a progressive fashion, and transmitted within a data-stream that can be rapidly terminated at any point.
  • the integer coefficients are first rearranged into an ordered array using the zig-zag sequence of the JPEG standard.
  • FIG. 3 shows a typical zig-zag sequence for an 8 ⁇ 8 DCT coefficient block. Each coefficient thus has an associated linear zig-zag address which may be used to identify it.
  • the reordering places the coefficients in ascending order of Manhattan distance from the DC term, that is from the term in the top left hand corner of the block which is representative of zero frequency in both the x and the y directions.
  • FIG. 2 there are shown three separate coefficient blocks, numbered 1 to 3 .
  • Each block is represented by a plurality of bit planes 1 to 4 , with plane 1 containing the most significant bits for each of the digitized coefficients, and plane No. 4 the corresponding least significant bits.
  • plane 1 containing the most significant bits for each of the digitized coefficients
  • plane No. 4 the corresponding least significant bits.
  • bit planes 1 to 4 there will normally be many more than three coefficient blocks, and many more than four bit planes: in the preferred embodiment, in fact, there may be twelve bit planes, with each bit plane having 32 ⁇ 32 bits. Other sizes could also be used, for example 8 ⁇ 8 or 16 ⁇ 16.
  • This algorithm may perhaps best be described by way of example, with reference to FIG. 2 ; in this example, it will be assumed for the sake of simplicity that all the values are positive integers.
  • a sweep is made across bit plane No. 1 of each of the three blocks 1 to 3 , to determine which are the significant coefficients. These will be the largest coefficients, and as the sweep is initially being made across the most significant bits, they may be determined by selecting those entries in which there is a one rather than a zero in bit plane 1 .
  • those are the coefficients labelled A, B, C and D; coefficients E and F are not yet significant, since their bits in bit plane 1 are zero.
  • bit plane 2 Bits from coefficients which have already been switched on in the previous sweep are automatically transmitted, so in this example bits 1001 are transmitted, these representing the second most significant bits for the coefficients A to D. Any coefficients which newly become significant at this level, are switched on, as illustrated by the crossed squares representing the second, third and fourth bits for coefficient E. The bits for all such newly significant coefficients on bit plane 2 are also transmitted.
  • the process continues for as many bit planes as were initially required to digitize and bitwise encode each individual coefficient, although the very last bit plane may need to be dealt with as a special case, to be discussed below.
  • the process is progressive, in the sense that the most important information is sent first, so that the transmission may be stopped part-way through if transmission time is limited and/or limited bandwidth is available.
  • the significant bits within each bit plane may be sent in any convenient order.
  • the first bits to be send may be those for each of the coefficients which have already been “switched on” (in other words those coefficients which are already significant); the addresses of any newly-significant coefficients are then transmitted, to switch them on, followed by a stop symbol. Finally, the next bit is sent for all of the newly-significant coefficients.
  • the switches may be sent first followed by all the data: this has the advantage of improving the run length coding of the data.
  • the DC coefficient within each block is preferably sent separately, prior to the significance sweeps of the AC terms.
  • the system needs to transmit addressing information representative of the positions of those coefficients which are significant. While this could be done simply by transmitting a list of addresses sent in zig-zag order, the applicants have determined that the sending of a binary mask, also in zig-zag order, can further improve efficiency. For example, referring to FIG. 3 , let us assume that coefficients 2 , 3 , 6 , 7 , 13 , 14 and 17 have become newly significant at the bit plane level of the current sweep. Those coefficients then have to be “turned on”, so that all the bits corresponding to those coefficients in the lower bit planes will in future automatically be transmitted. That requires the transmission of the addresses of those coefficients.
  • the addresses may be transmitted not directly but by way of a bit mask, in zig-zag order.
  • the mask 001100110000011001 may be sent.
  • a ‘1’ in the mask indicates that that particular coefficient has newly become significant. This may be run length coded as 2020502 STOP.
  • the coefficients within each DCT block may be negative as well as positive, and if appropriate negative values may be suitably bitwise encoded using a 2 's compliment representation. If the coefficient is positive, the “1” bits are significant; if negative the “0” bits are significant. The first data bit sent when a coefficient becomes significant determines the sign.
  • the final bit plane may need to be dealt with as a special case. Any coefficient not previously switched on will be either 0 or ⁇ 1.
  • One way of dealing with the final phase is to send a mask only for the ⁇ 1 coefficients; there is no need to send any data, as all non-significant coefficients are 0 and all those newly switched are ⁇ 1.
  • the mask may be sent sequentially, followed by a special stop symbol to indicate its end.
  • a length symbol representative of the zig-zag length may be sent before the mask to obviate the need for the special stop symbol.
  • the mask could be preceded by the Manhattan depth of its highest order coefficient; for example in FIG. 3 , the highest order coefficient has zig-zag address 17 and a Manhattan depth of 5 .
  • Using Manhattan depth will result in three extra mask bits being set (with the mask being assumed to terminate at the end of the zig-zag line that includes the given Manhattan depth).
  • the Manhattan depth requires two fewer bits at all block sizes than would a zig-zag length. Recalling that the DCT packs most of the energy into a small subset of coefficients, leaving a large number of smaller coefficients, it will be understood that the Manhattan depth is more efficient for small values.
  • the switching mask will contain mostly “Off” symbols (zeros), which can be efficiently compressed using an arithmetic coder. It is particularly useful to encode the mask data using first order predictive adaptive arithmetic coding, since much of the data comprises long runs of low entropy off symbols. Other methods such as Huffman could also be used to compact the mask data prior to transmission. Run length coding may also be used.
  • the output from the coder will take the form of one or more bit streams.
  • the mask and symbol may be sent as one stream with the DCT data being sent as another stream.
  • the synchronization may be controlled so that each stream is maintained within a few bytes of synchrony with the others. This allows a decoder to interrupt the transmission at any time.
  • a decoder maintains a record of the mask for each image block, giving the current status of each of its DCT coefficients.
  • the mask is updated at each significance pass.
  • an “off” mask may be maintained and sent as well as the “on” mask discussed above. This allows the coder to avoid the sending of bits which are so far down the bit planes as to represent noise rather than real data. In practice, once a coefficient has been significant for several bit planes it may well be sufficiently-well defined for visual purposes, and it could then be turned off.
  • the significance testing need not be carried out for each consecutive bit plane. In many circumstances, it may be more efficient, and may given an acceptable result, to mask several planes at once. This reduces the overhead of masking each bit plane individually.
  • each bit plane is mathematically equivalent to making comparisons against a decreasing threshold value which goes as 2 n .
  • Other threshold values could be used instead, providing for either wider or narrower significance steps.
  • the bit planes are divided up into groups, with each group being masked separately. Depending on the application, each group might consist of the same number or alternatively of a different number of bit planes. Some of the groups might comprise only a single bit plane, while others are made up of several.
  • the mask switching algorithm described above performs slightly better than JPEG at low compression ratios, and substantially better at high compression ratios. Indeed, using a 16 ⁇ 16 block size, performance over the whole range of compression ratios is very similar to that achieved by the wavelet coder of Said and Pearlman: see SAID, A. and PEARLMAN, W. A,: ‘Image Compression Using the Spatial-Orientation Tree’, IEEE International Symposium on Circuits and Systems, 1993, (694), pp.279-282.
  • the present invention therefore provides state of the art performance while having the advantage of being usable with DCT, a transformation which is widely used and understood as a result of its adoption at the core technology of JPEG and all current versions of MPEG.
  • a video codec comprises a hardware or software based coder, and a hardware or software based decoder. Bits are transmitted progressively from the coder to the decoder, with the coder being instructed to keep sending bits until a certain compression target has been reached, or a certain distortion achieved.
  • the decoder can progressively reconstruct the image. As the mask data are received, the decoder updates a record, held in memory, of which coefficients are currently switched on. As data transmission proceeds, more and more coefficients are switched on.
  • the decoder can reconstruct a lossless image, with the exception of any small rounding errors that may have occurred during the DCT digitization process.
  • a decoder in a multi-media system which uses a progressive, embedded coder as described above can begin to reveal an image as soon as transmission commences. This is an advantage often claimed for wavelets, but significance-switched block transforms also have this capability.

Abstract

A method of image compression includes significance switching of DCT coefficients in block-based embedded DCT procedures. Bitwise digitized DCT coefficients are passed through successive significance sweeps of the whole image from the most significant down to the least significant coefficient bit planes. With each new sweep, newly significant coefficients may appear within a block, and block-masking is used to transmit the addresses of those newly significant coefficients. An off-mask may also be used. The invention further relates to a hardware or software-based image encoder.

Description

This Application is a continuation of International Application No. PCT/GB98/00360, filed Feb. 5, 1998, now pending (which is hereby incorporated by reference).
TECHNICAL FIELD
The present invention relates to image compression and particularly, although not exclusively, to a progressive block-based embedded DCT coder, and to a method of encoding.
BACKGROUND OF THE INVENTION
The JPEG baseline method for still image coding uses the Discrete Cosine Transform (DCT) in a fixed 8×8 pixel partition. Through a linear quantization table and zig-zag scanning of DCT coefficients, the redundancy and band width characteristics of the DCT are exploited over a range of compressions. Recently, however, it has become clear that the JPEG coder is not particularly efficient at higher compression ratios, and other methods such as wavelets have produced better results while having the advantage of being fully embedded. Some researchers have also attempted to combine DCT with zerotree quantization, usually associated with wavelet transforms: see Xiong, Guleryuz and Orchard, ‘A DCT-Based Image Coder’, IEEE Sig. Proc. Lett., Vol 3, No 11, November 1996, p289.
It is an object of the present invention to advance the field of image compression generally, and in particular to provide an improved method of image compression which is capable of use with well understood transforms such as the DCT.
SUMMARY OF THE INVENTION
According to a first aspect of the present invention this is provided a method of image compression comprising:
    • (a) dividing an image to be compressed into a plurality of image blocks;
    • (b) carrying out a two-dimensional block transform on each block to produce a corresponding plurality of coefficient blocks;
    • (c) bitwise digitizing the coefficients within each coefficient block to define a plurality of bit planes for each coefficient block;
    • (d) defining a group of one or more consecutive bit planes starting with the most significant bit plane;
    • (e) selecting those coefficients which first become significant within the group;
    • (f) flagging the said selected coefficients;
    • (g) transmitting information representative of the positions of the said selected coefficients; and transmitting the bits within the group of the said coefficients; and
    • (h) repeating (d) to (g) one or more times, with each new group starting with the most significant bit plane not previously dealt with; and, at each repeated pass, also transmitting the bits within the current group of those coefficients which were previously flagged on an earlier pass.
Such a method could also be applied using the one-dimensional DCT to audio recording.
According to a second aspect of the invention there is provided a coder for encoding images, comprising:
    • (a) means for dividing an image to be compressed into a plurality of image blocks;
    • (b) means for carrying out a two-dimensional block transform on each block to produce a corresponding plurality of coefficient blocks;
    • (c) means for bitwise digitizing the coefficients within each coefficient block to define a plurality of bit planes for each coefficient block;
    • (d) means for defining a group of one or more consecutive bit planes starting with the most significant bit plane;
    • (e) means for selecting those coefficients which first become significant within the group;
    • (f) means for flagging the said selected coefficients;
    • (g) means for transmitting information representative of the positions of the said selected coefficients, and for transmitting the bits within the group of the said coefficients; and
    • (h) means for repeating (d) to (g) one or more times, with each new group starting with the most significant bit plane not previously dealt with; and means for transmitting, at each repeated pass, the bits within the current group of those coefficients which were previously flagged on an earlier pass.
Preferably, the encoder provides for significance switching of DCT coefficients in block-based embedded DCT image compression. The encoder provides output on one or more data streams that may be terminated within a few bits of any point.
The invention also extends to a video coder/decoder including a coder as claimed in claim 17 and an associated decoder, the decoder being arranged to maintain a running record, as transmission between the coder and the decoder proceeds, of the coefficients which are currently significant.
The preferred two-dimensional block transform of the present invention is the Discrete Cosine Transform, although other transforms such as the Fast Fourier Transform (FFT) or the Lapped Orthogonal Transform could be used.
It will be appreciated that in the method of the present invention the order of transmission is not significant. It is therefore to be understood, except where logic requires, that the various parts of the method are not necessarily carried out sequentially in the order specified in section (d) of claim 1. For example, the bits from the newly selected coefficients could be transmitted either before or after the bits of those coefficients which were previously flagged on an earlier path. Similarly, the bits of those coefficients which were previously flagged on an earlier path could be transmitted either before or after the new coefficients are selected for flagging.
The bit planes are swept consecutively from the most significant bit plane to the least significant bit plane. This may either be repeated separately for each image block, or alternatively all blocks may be dealt with at the first bit plane, then all blocks dealt with at the second bit plane, and so on.
The philosophy of significance switching, as used in the present invention, is that the overheads introduced will be compensated for by the savings in not transmitting bits for small coefficients until they are switched on. Good performance might naturally be expected at high compression ratios, but what is surprising is the excellent performance both for lossless compression and for compression at low ratios. The coder of the present invention is preferably embedded, in other words the bit stream can be stopped within a few bits of any point while still guaranteeing the least possible distortion overall. When used with an appropriate decoder, either coder or decoder can terminate the bit stream as needed, dependent upon the available bandwidth or the available bit budget.
The coder of the present invention has been found to out-perform the base-line JPEG method in peak signal to noise ratio (PSNR) at any compression ratio, and is similar to state-of-the-art wavelet coders. The results show that block transform (for example DCT) coding is competitive across the whole range of compression ratios, including lossless, so that a significance-switched block coder would be capable of meeting the requirements of future image compression standards in an evolutionary manner.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention may be carried into practice a number of ways and one specific embodiment will now be described, by way of example, with reference to the figures in which:
FIG. 1 illustrates the DCT transformation and digitization of image blocks;
FIG. 2 illustrates the significance sweeps made through the individual bit planes in coefficient space; and
FIG. 3 illustrates the zig-zag ordering of coefficients, and use of the binary mask.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
In the preferred the preferred method of the present invention, the image to be encoded is first partitioned into a plurality of square image blocks. The partitioning may either be by way of a regular tiling, for example of 8×8 pixel blocks, or alternatively some more complex tiling using blocks of differing sizes. One convenient method of tiling is to vary the block size across the image according to the power in the image (measured by the sum of the squares of pixel intensities). As shown in FIG. 1, each image block is then transformed using the 2D spatial DCT (Discrete Cosine Transform) to produce a corresponding DCT coefficient block of the same size as the original image block. Each of the coefficients within the coefficient block is then digitized and bitwise encoded as shown in FIG. 1. Each coefficient 10 within the coefficient block 20 is therefore represented as a series of individual bits 30, ranging from the most significant bit (MSB) to the least significant bit (LSB). The DCT transformation and digitization preferably follow the scaling used within the JPEG standard: see Wallace, G. K, “The JPEG Still Picture Compression Standard”, Comm. ACCM, Vol. 34, No. 4, pp.30-44, 1991. This scaling represents a compromise which accommodates the full range of possible DCT coefficients with reasonably safe integer roundings.
The encoded integer coefficients are now required to be manipulated in a progressive fashion, and transmitted within a data-stream that can be rapidly terminated at any point. In order to achieve this, the integer coefficients are first rearranged into an ordered array using the zig-zag sequence of the JPEG standard. FIG. 3 shows a typical zig-zag sequence for an 8×8 DCT coefficient block. Each coefficient thus has an associated linear zig-zag address which may be used to identify it. The reordering places the coefficients in ascending order of Manhattan distance from the DC term, that is from the term in the top left hand corner of the block which is representative of zero frequency in both the x and the y directions.
Turning now to FIG. 2, there are shown three separate coefficient blocks, numbered 1 to 3. Each block is represented by a plurality of bit planes 1 to 4, with plane 1 containing the most significant bits for each of the digitized coefficients, and plane No. 4 the corresponding least significant bits. It will be understood of course, that in a typical implementation there will normally be many more than three coefficient blocks, and many more than four bit planes: in the preferred embodiment, in fact, there may be twelve bit planes, with each bit plane having 32×32 bits. Other sizes could also be used, for example 8×8 or 16×16.
In order to determine the order in which the individual bits will be transmitted, the following algorithm is used:
Sweep DCT bit planes from MSB to LSB
    • Sweep all image blocks
      • Switch on significant coefficients
      • Send data bits for significant coefficients
    • Next block
Next bit plane
This algorithm may perhaps best be described by way of example, with reference to FIG. 2; in this example, it will be assumed for the sake of simplicity that all the values are positive integers. First, a sweep is made across bit plane No. 1 of each of the three blocks 1 to 3, to determine which are the significant coefficients. These will be the largest coefficients, and as the sweep is initially being made across the most significant bits, they may be determined by selecting those entries in which there is a one rather than a zero in bit plane 1. In FIG. 2, those are the coefficients labelled A, B, C and D; coefficients E and F are not yet significant, since their bits in bit plane 1 are zero.
Once the significant coefficients have been selected, all of the corresponding bits in the lower bit planes 2 to 4 are automatically switched on, as indicated by the filled in squares representing the bits of coefficients A to D. The selected bits within bit plane 1 are then transmitted.
Next, a second sweep is made over the whole image across bit plane 2. Bits from coefficients which have already been switched on in the previous sweep are automatically transmitted, so in this example bits 1001 are transmitted, these representing the second most significant bits for the coefficients A to D. Any coefficients which newly become significant at this level, are switched on, as illustrated by the crossed squares representing the second, third and fourth bits for coefficient E. The bits for all such newly significant coefficients on bit plane 2 are also transmitted.
Another sweep is then carried out on the third bit plane. Once again, all the bits representative of coefficients which have already been switched on are automatically sent: in this case, these are the third bits of coefficients A, B, C, D and E. At this level, coefficient F newly becomes significant, and accordingly the one representative of bit 3 of that coefficient is also sent. At the same time, that coefficient is switched on, as indicated by the half-shaded squares representative of the third and fourth bits of that coefficient.
Finally, a sweep is performed across the fourth bit planes. In this example, all of the illustrated coefficients have previously been switched on at a higher level, and hence all of the resultant bits 111000 are transmitted.
The process continues for as many bit planes as were initially required to digitize and bitwise encode each individual coefficient, although the very last bit plane may need to be dealt with as a special case, to be discussed below. The process is progressive, in the sense that the most important information is sent first, so that the transmission may be stopped part-way through if transmission time is limited and/or limited bandwidth is available.
It has been found in practice to be more efficient to exclude the DC component of each block from the above scheme, and to send that separately. Accordingly, in the preferred embodiment switching applied only to the AC terms of the coefficient block. Rather than sweeping across each of the coefficient blocks for a particular bit plane, it would in an alternative embodiment be possible to sweep through all the bit planes within one block before proceeding to the next block. The corresponding algorithm for this would be:
Sweep all image blocks
    • Sweep DCT bit planes from MSB to LSB
      • Switch on significant coefficients
      • Send data bits for significant coefficients
    • Next bit plane
Next block
During each significance sweep, the significant bits within each bit plane may be sent in any convenient order. For example, within each bit plane, the first bits to be send may be those for each of the coefficients which have already been “switched on” (in other words those coefficients which are already significant); the addresses of any newly-significant coefficients are then transmitted, to switch them on, followed by a stop symbol. Finally, the next bit is sent for all of the newly-significant coefficients. Alternatively, the switches may be sent first followed by all the data: this has the advantage of improving the run length coding of the data.
For improved efficiency at higher compressions, the DC coefficient within each block is preferably sent separately, prior to the significance sweeps of the AC terms.
It should be understood that the system needs to transmit addressing information representative of the positions of those coefficients which are significant. While this could be done simply by transmitting a list of addresses sent in zig-zag order, the applicants have determined that the sending of a binary mask, also in zig-zag order, can further improve efficiency. For example, referring to FIG. 3, let us assume that coefficients 2, 3, 6, 7, 13, 14 and 17 have become newly significant at the bit plane level of the current sweep. Those coefficients then have to be “turned on”, so that all the bits corresponding to those coefficients in the lower bit planes will in future automatically be transmitted. That requires the transmission of the addresses of those coefficients.
The applicant has realized that the addresses may be transmitted not directly but by way of a bit mask, in zig-zag order. Thus, in order to switch on the mentioned coefficients (assuming none have previously been switched on), the mask 001100110000011001 may be sent. A ‘1’ in the mask indicates that that particular coefficient has newly become significant. This may be run length coded as 2020502 STOP.
The coefficients within each DCT block may be negative as well as positive, and if appropriate negative values may be suitably bitwise encoded using a 2's compliment representation. If the coefficient is positive, the “1” bits are significant; if negative the “0” bits are significant. The first data bit sent when a coefficient becomes significant determines the sign.
The final bit plane may need to be dealt with as a special case. Any coefficient not previously switched on will be either 0 or −1. One way of dealing with the final phase is to send a mask only for the −1 coefficients; there is no need to send any data, as all non-significant coefficients are 0 and all those newly switched are −1.
It should be recalled that only new coefficients have to be masked at each pass, since once a coefficient has been switched on it remains switched on until the end of the procedure. The masking method is efficient, with typically fewer bits needed to transmit the switching information than a direct list of coefficient addresses.
Various methods of packaging the mask prior to transmission may be used. The mask may be sent sequentially, followed by a special stop symbol to indicate its end. In an alternative scheme, a length symbol representative of the zig-zag length may be sent before the mask to obviate the need for the special stop symbol. Finally, the mask could be preceded by the Manhattan depth of its highest order coefficient; for example in FIG. 3, the highest order coefficient has zig-zag address 17 and a Manhattan depth of 5. Using Manhattan depth will result in three extra mask bits being set (with the mask being assumed to terminate at the end of the zig-zag line that includes the given Manhattan depth). However, the Manhattan depth requires two fewer bits at all block sizes than would a zig-zag length. Recalling that the DCT packs most of the energy into a small subset of coefficients, leaving a large number of smaller coefficients, it will be understood that the Manhattan depth is more efficient for small values.
In each case, the switching mask will contain mostly “Off” symbols (zeros), which can be efficiently compressed using an arithmetic coder. It is particularly useful to encode the mask data using first order predictive adaptive arithmetic coding, since much of the data comprises long runs of low entropy off symbols. Other methods such as Huffman could also be used to compact the mask data prior to transmission. Run length coding may also be used.
Depending upon the method chosen to package the mask, the output from the coder will take the form of one or more bit streams. In the first arrangement mentioned above, where the mask is terminated by a special stop symbol, the mask and symbol may be sent as one stream with the DCT data being sent as another stream. In the alternatives, where a separate length symbol or Manhattan depth symbol is used, it may be convenient to output three separate streams: one specifying the Manhattan depth, one for the mask data, and one for the DCT data.
Where several separate data streams are used, the synchronization may be controlled so that each stream is maintained within a few bytes of synchrony with the others. This allows a decoder to interrupt the transmission at any time.
At the far end of the transmission stream, a decoder maintains a record of the mask for each image block, giving the current status of each of its DCT coefficients. The mask is updated at each significance pass.
In a variant of the method, an “off” mask may be maintained and sent as well as the “on” mask discussed above. This allows the coder to avoid the sending of bits which are so far down the bit planes as to represent noise rather than real data. In practice, once a coefficient has been significant for several bit planes it may well be sufficiently-well defined for visual purposes, and it could then be turned off.
In a further more general variant of the method, the significance testing need not be carried out for each consecutive bit plane. In many circumstances, it may be more efficient, and may given an acceptable result, to mask several planes at once. This reduces the overhead of masking each bit plane individually.
It will be understood that masking each bit plane separately is mathematically equivalent to making comparisons against a decreasing threshold value which goes as 2n. Other threshold values could be used instead, providing for either wider or narrower significance steps. In one embodiment, the bit planes are divided up into groups, with each group being masked separately. Depending on the application, each group might consist of the same number or alternatively of a different number of bit planes. Some of the groups might comprise only a single bit plane, while others are made up of several.
It is found in practice that the mask switching algorithm described above performs slightly better than JPEG at low compression ratios, and substantially better at high compression ratios. Indeed, using a 16×16 block size, performance over the whole range of compression ratios is very similar to that achieved by the wavelet coder of Said and Pearlman: see SAID, A. and PEARLMAN, W. A,: ‘Image Compression Using the Spatial-Orientation Tree’, IEEE International Symposium on Circuits and Systems, 1993, (694), pp.279-282. The present invention therefore provides state of the art performance while having the advantage of being usable with DCT, a transformation which is widely used and understood as a result of its adoption at the core technology of JPEG and all current versions of MPEG.
In a typical embodiment of the present invention, a video codec (Coder/Decoder) comprises a hardware or software based coder, and a hardware or software based decoder. Bits are transmitted progressively from the coder to the decoder, with the coder being instructed to keep sending bits until a certain compression target has been reached, or a certain distortion achieved. Using the two or three individual streams of data previously referred to, the decoder can progressively reconstruct the image. As the mask data are received, the decoder updates a record, held in memory, of which coefficients are currently switched on. As data transmission proceeds, more and more coefficients are switched on. If the process is allowed to continue until all of the data has been transmitted, the decoder can reconstruct a lossless image, with the exception of any small rounding errors that may have occurred during the DCT digitization process. A decoder in a multi-media system which uses a progressive, embedded coder as described above can begin to reveal an image as soon as transmission commences. This is an advantage often claimed for wavelets, but significance-switched block transforms also have this capability.

Claims (22)

1. A method of image compression comprising the steps of:
(a) dividing an image to be compressed into a plurality of image blocks;
(b) carrying out a two-dimensional block transform on each block to produce a corresponding plurality of coefficient blocks;
(c) bitwise digitizing the coefficients within each coefficient block to define a plurality of bit planes for each coefficient block;
(d) defining a group of one or more consecutive bit planes starting with the most significant bit plane;
(e) selectively flagging those, by a coder device, coefficients which first become significant within the a group of one or more consecutive bit planes of corresponding coefficient blocks of image blocks resulting from a block transform, starting with the most significant bit plane; and
(f) transmitting, by the coder device, information representative of the positions of the said flagged coefficients and transmitting the bits within the a group of the said flagged coefficients; and,
(g) repeating steps (d) to (f) one or more times, with each new group starting with the most significant bit plane not previously dealt with and, at each repeated pass, also transmitting the bits within the current group of those coefficients which were previously flagged on an earlier pass.
2. A method as claimed in claim 1 in which step (g) is carried out across the entire image to be compressed further comprising:
repeating the selectively flagging and the transmitting one or more times, each with a new group starting with a most significant bit plane not previously dealt with and, at each repeated pass, also transmitting bits within a current group of those coefficients which were previously flagged on an earlier pass.
3. A method as claimed in claim 1 2 in which step (g) the repeating is separately repeated performed for each image block.
4. A method as claimed in claim 1 2 in which the block transform is the a two-dimensional Discrete Cosine Transform.
5. A method as claimed in claim 1 2 in which the block transform is the a Lapped Orthogonal Transform.
6. A method as claimed in claim 1 2 in which the block transform is the a Fast Fourier Transform.
7. A method as claimed in claim 1 2 further including, at step (f) transmitting, by the coder device, mask information representative of a binary mask which defines the positions of the said selected flagged coefficients.
8. A method as claimed in claim 7 in which the binary mask defines the positions of the selected said flagged coefficients within each coefficient block in JPEG zig-zag order.
9. A method as claimed in claim 7 in which the binary mask is associated with the a mask length code to define the mask end point.
10. A method as claimed in claim 7 in which the binary mask is associated with a stop-code to define the a mask end point.
11. A method as claimed in claim 7 in which transmitted mask information is an entropy-coded version of the mask.
12. A method as claimed in claim 11 in which the transmitted mask information is an arithmetic coded version of the mask.
13. A method as claimed in claim 11 in which the transmitted mask information is a Huffman coded version of the mask.
14. A method as claimed in claim 7 in which the transmitted mask information is run length coded.
15. A method as claimed in claim 1 2 in which the a binary mask defines the positions of the selected said flagged coefficients within each coefficient block in JPEG zig-zag order, the binary mask is associated with a mask length code to define the a mask end point, and the mask length code defines the a mask end point zig-zag address.
16. A method as claimed in claim 1 2 in which the a binary mask defines the positions of the selected said flagged coefficients within each coefficient block in JPEG zig-zag order, the binary mask is associated with a mask length code to define the a mask end point, and the mask length code defines the a Manhattan distance from a DC term to the mask end point.
17. A method as claimed in claim 1 2 further including the step of transmitting information representative of a binary off-mask for defining the positions of coefficients whose bits are no longer required to be sent.
18. A coder for encoding images device, comprising the steps of:
(a) means for dividing an image to be compressed into a plurality of image blocks;
(b) means for carrying out a two-dimensional block transform on each block to produce a corresponding plurality of coefficient blocks;
(c) means for bitwise digitizing the coefficients within each coefficient block to define a plurality of bit planes for each coefficient block;
(d) means for defining a group of one or more consecutive bit planes starting with the most significant bit plane;
(e) means fora mechanism configured to selectively flagging thoseflag coefficients which first become significant within the a group of one or more consecutive bit planes of corresponding coefficient blocks of image blocks resulting from a block transform, starting with the most significant bit plane; and
(f) means fora mechanism configured to transmittingtransmit information representative of the positions of the said flagged coefficients and transmitting the bits within the a group of the said flagged coefficients; and,
(g) means for repeating steps (d) to (f) one or more times, with each new group starting with the most significant bit plane not previously dealt with, and means for transmitting, at each repeated pass, the bits within the current group of those coefficients which were previously flagged on an earlier pass.
19. A coder device as claimed in claim 18 in which the means for transmitting information representative of the positions of the said selected coefficients mechanism configured to transmit comprise comprises a binary mask means mechanism.
20. A coder device as claimed in claim 19 including means for transmitting a transmitting mechanism configured to transmit, as synchronized data streams, the coefficient bits and mask information.
21. A video coder/decoder comprising:
a coder and an associated decoder, wherein
(1) the coder encoding images and comprising the steps of comprises:
(a) means for dividinga mechanism configured to divide an image to be compressed into a plurality of image blocks;
(b) means fora mechanism configured to carryingcarry out a two-dimensional block transform on each block to produce a corresponding plurality of coefficient blocks;
(c) means fora mechanism configured to bitwise digitizing thedigitize coefficients withwithin each coefficient block to define a plurality of bit planes for each coefficient block;
(d) means for defininga mechanism configured to define a group of one or more consecutive bit planes starting with thea most significant bit plane;
(e) means fora mechanism configured to selectively flaggingflag those coefficients which first become significant within the group;
(f) means fora mechanism configured to transmittingtransmit information representative of the positions of the said flagged coefficients and for transmitting the bits within the a group of the said flagged coefficients; and,
(g) means fora mechanism configured to repeating steps (d) to (f)repeat operation of the mechanism configured to define a group and the mechanism configured to transmit information one or more times, with each new group starting with thea most significant bit plane not previously dealt with, and means for transmitting, at each repeated pass, the bits within the a current group of those coefficients which were previously flagged on an earlier pass, and
(2) the decoder being arranged to maintain a running record, as transmission between the coder and the decoder proceeds, of the coefficients which are currently significant.
22. A method as claimed in claim 2 in which the repeating step is carried out across an entire image to be compressed.
US11/119,499 1997-02-19 2005-04-29 Progressive block-based coding for image compression Expired - Lifetime USRE43649E1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US11/119,499 USRE43649E1 (en) 1997-02-19 2005-04-29 Progressive block-based coding for image compression

Applications Claiming Priority (5)

Application Number Priority Date Filing Date Title
GBGB9703441.7A GB9703441D0 (en) 1997-02-19 1997-02-19 Progressive block-based coding for image compression
GB9703441 1997-02-19
PCT/GB1998/000360 WO1998037700A1 (en) 1997-02-19 1998-02-05 Progressive block-based coding for image compression
US09/375,459 US6556719B1 (en) 1997-02-19 1999-08-17 Progressive block-based coding for image compression
US11/119,499 USRE43649E1 (en) 1997-02-19 2005-04-29 Progressive block-based coding for image compression

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
US09/375,459 Reissue US6556719B1 (en) 1997-02-19 1999-08-17 Progressive block-based coding for image compression

Publications (1)

Publication Number Publication Date
USRE43649E1 true USRE43649E1 (en) 2012-09-11

Family

ID=10807940

Family Applications (2)

Application Number Title Priority Date Filing Date
US09/375,459 Ceased US6556719B1 (en) 1997-02-19 1999-08-17 Progressive block-based coding for image compression
US11/119,499 Expired - Lifetime USRE43649E1 (en) 1997-02-19 2005-04-29 Progressive block-based coding for image compression

Family Applications Before (1)

Application Number Title Priority Date Filing Date
US09/375,459 Ceased US6556719B1 (en) 1997-02-19 1999-08-17 Progressive block-based coding for image compression

Country Status (5)

Country Link
US (2) US6556719B1 (en)
EP (1) EP0962099B1 (en)
DE (1) DE69806729T2 (en)
GB (1) GB9703441D0 (en)
WO (1) WO1998037700A1 (en)

Families Citing this family (57)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB9703441D0 (en) 1997-02-19 1997-04-09 British Tech Group Progressive block-based coding for image compression
AU735741B2 (en) * 1998-12-18 2001-07-12 Canon Kabushiki Kaisha Method and apparatus for decimation filtering
PL206212B1 (en) * 1999-08-27 2010-07-30 Koninklijke Philips Electronics Nvkoninklijke Philips Electronics Nv Scalable coding
JP2003508981A (en) * 1999-08-27 2003-03-04 コーニンクレッカ フィリップス エレクトロニクス エヌ ヴィ Extensible coding
DE19957685A1 (en) * 1999-11-30 2001-06-28 Siemens Ag Method and arrangement for coding or decoding a sequence of numbers
US6826232B2 (en) 1999-12-20 2004-11-30 Koninklijke Philips Electronics N.V. Fine granular scalable video with embedded DCT coding of the enhancement layer
DE10007171A1 (en) * 2000-02-17 2001-08-30 Siemens Ag Method and arrangement for coding or for coding and decoding a sequence of numbers
US6501397B1 (en) 2000-05-25 2002-12-31 Koninklijke Philips Electronics N.V. Bit-plane dependent signal compression
GB0019121D0 (en) 2000-08-03 2000-09-27 Wave Limited M Signal compression and decompression
CN1636394A (en) * 2000-10-11 2005-07-06 皇家菲利浦电子有限公司 Spatial scalability for fine granular video encoding
US6934337B2 (en) 2001-09-27 2005-08-23 Intel Corporation Video capture device and method of sending high quality video over a low data rate link
US20030156637A1 (en) * 2002-02-15 2003-08-21 Koninklijke Philips Electronics N.V. Memory-bandwidth efficient FGS encoder
JP2005522106A (en) * 2002-04-01 2005-07-21 メイトリックスヴュー リミテッド Iterative coding compression of strongly correlated image data
GB2388502A (en) * 2002-05-10 2003-11-12 Chris Dunn Compression of frequency domain audio signals
DE10236694A1 (en) 2002-08-09 2004-02-26 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Equipment for scalable coding and decoding of spectral values of signal containing audio and/or video information by splitting signal binary spectral values into two partial scaling layers
GB0321954D0 (en) 2003-09-18 2003-10-22 Wave Ltd M Data compression
AU2004284829A1 (en) * 2003-12-15 2005-07-07 Matrixview Limited Compressing image data
KR101303375B1 (en) 2004-05-13 2013-09-03 코닌클리케 필립스 일렉트로닉스 엔.브이. Method and device for encoding blocks of values
US7848584B2 (en) * 2005-09-08 2010-12-07 Monro Donald M Reduced dimension wavelet matching pursuits coding and decoding
US7813573B2 (en) * 2005-09-08 2010-10-12 Monro Donald M Data coding and decoding with replicated matching pursuits
US8121848B2 (en) * 2005-09-08 2012-02-21 Pan Pacific Plasma Llc Bases dictionary for low complexity matching pursuits data coding and decoding
US20070065034A1 (en) * 2005-09-08 2007-03-22 Monro Donald M Wavelet matching pursuits coding and decoding
US20070271250A1 (en) * 2005-10-19 2007-11-22 Monro Donald M Basis selection for coding and decoding of data
US8674855B2 (en) * 2006-01-13 2014-03-18 Essex Pa, L.L.C. Identification of text
US7783079B2 (en) * 2006-04-07 2010-08-24 Monro Donald M Motion assisted data enhancement
US7586424B2 (en) * 2006-06-05 2009-09-08 Donald Martin Monro Data coding using an exponent and a residual
US20070290899A1 (en) * 2006-06-19 2007-12-20 Donald Martin Monro Data coding
US7845571B2 (en) * 2006-06-19 2010-12-07 Monro Donald M Data compression
US7770091B2 (en) * 2006-06-19 2010-08-03 Monro Donald M Data compression for use in communication systems
US7689049B2 (en) 2006-08-31 2010-03-30 Donald Martin Monro Matching pursuits coding of data
US7508325B2 (en) * 2006-09-06 2009-03-24 Intellectual Ventures Holding 35 Llc Matching pursuits subband coding of data
US7974488B2 (en) 2006-10-05 2011-07-05 Intellectual Ventures Holding 35 Llc Matching pursuits basis selection
US20080084924A1 (en) * 2006-10-05 2008-04-10 Donald Martin Monro Matching pursuits basis selection design
US7707214B2 (en) * 2007-02-21 2010-04-27 Donald Martin Monro Hierarchical update scheme for extremum location with indirect addressing
US7707213B2 (en) * 2007-02-21 2010-04-27 Donald Martin Monro Hierarchical update scheme for extremum location
US20080205505A1 (en) * 2007-02-22 2008-08-28 Donald Martin Monro Video coding with motion vectors determined by decoder
US10194175B2 (en) * 2007-02-23 2019-01-29 Xylon Llc Video coding with embedded motion
US8144037B2 (en) * 2007-07-12 2012-03-27 Intellectual Ventures Fund 44 Llc Blocking for combinatorial coding/decoding for electrical computers and digital data processing systems
US7737869B2 (en) * 2007-07-12 2010-06-15 Monro Donald M Symbol based data compression
US7671767B2 (en) * 2007-07-12 2010-03-02 Donald Martin Monro LIFO radix coder for electrical computers and digital data processing systems
US8055085B2 (en) * 2007-07-12 2011-11-08 Intellectual Ventures Fund 44 Llc Blocking for combinatorial coding/decoding for electrical computers and digital data processing systems
US7511639B2 (en) * 2007-07-12 2009-03-31 Monro Donald M Data compression for communication between two or more components in a system
US7602316B2 (en) * 2007-07-12 2009-10-13 Monro Donald M Data coding/decoding for electrical computers and digital data processing systems
US7545291B2 (en) * 2007-07-12 2009-06-09 Donald Martin Monro FIFO radix coder for electrical computers and digital data processing systems
US7990289B2 (en) * 2007-07-12 2011-08-02 Intellectual Ventures Fund 44 Llc Combinatorial coding/decoding for electrical computers and digital data processing systems
US7511638B2 (en) * 2007-07-12 2009-03-31 Monro Donald M Data compression for communication between two or more components in a system
US7548176B2 (en) * 2007-07-12 2009-06-16 Donald Martin Monro Data coding buffer for electrical computers and digital data processing systems
US7864086B2 (en) * 2008-10-06 2011-01-04 Donald Martin Monro Mode switched adaptive combinatorial coding/decoding for electrical computers and digital data processing systems
US7791513B2 (en) * 2008-10-06 2010-09-07 Donald Martin Monro Adaptive combinatorial coding/decoding with specified occurrences for electrical computers and digital data processing systems
US7786907B2 (en) * 2008-10-06 2010-08-31 Donald Martin Monro Combinatorial coding/decoding with specified occurrences for electrical computers and digital data processing systems
US7786903B2 (en) * 2008-10-06 2010-08-31 Donald Martin Monro Combinatorial coding/decoding with specified occurrences for electrical computers and digital data processing systems
US8810562B2 (en) * 2009-05-19 2014-08-19 Advanced Micro Devices, Inc. Hierarchical lossless compression
CN102461170B (en) 2009-06-16 2015-06-17 皇家飞利浦电子股份有限公司 Encoding a signal into a scalable bitstream and decoding such bitstream
WO2011031592A2 (en) * 2009-09-09 2011-03-17 Sony Corporation Bitstream syntax for graphics-mode compression in wireless hd 1.1
JP2013506379A (en) * 2009-10-14 2013-02-21 ソニー株式会社 Combined scalar embedded graphics coding for color images
US9286643B2 (en) 2011-03-01 2016-03-15 Applaud, Llc Personalized memory compilation for members of a group and collaborative method to build a memory compilation
WO2018213499A1 (en) * 2017-05-16 2018-11-22 Google Llc Stop code tolerant image compression neural networks

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0327931A2 (en) 1988-02-06 1989-08-16 Dainippon Screen Mfg. Co., Ltd. Method of storing and transmitting image data as image file suitable for image search
EP0551672A2 (en) 1992-01-13 1993-07-21 Koninklijke Philips Electronics N.V. Method of and system for transmitting and receiving digitally encoded video images
US5416854A (en) * 1990-07-31 1995-05-16 Fujitsu Limited Image data processing method and apparatus
US5504484A (en) * 1992-11-09 1996-04-02 Matsushita Electric Industrial Co., Ltd. Variable-length data alignment apparatus for digital video data
US5563960A (en) 1993-01-22 1996-10-08 David Sarnoff Research Center, Inc. Apparatus and method for emphasizing a selected region in the compressed representation of an image
US5768437A (en) 1992-02-28 1998-06-16 Bri Tish Technology Group Ltd. Fractal coding of data
WO1998037700A1 (en) 1997-02-19 1998-08-27 Btg International Limited Progressive block-based coding for image compression
US5818877A (en) 1996-03-14 1998-10-06 The Regents Of The University Of California Method for reducing storage requirements for grouped data values
US7142720B1 (en) * 1990-07-31 2006-11-28 Fujitsu Limited Image data processing method and apparatus

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0327931A2 (en) 1988-02-06 1989-08-16 Dainippon Screen Mfg. Co., Ltd. Method of storing and transmitting image data as image file suitable for image search
US5416854A (en) * 1990-07-31 1995-05-16 Fujitsu Limited Image data processing method and apparatus
US7142720B1 (en) * 1990-07-31 2006-11-28 Fujitsu Limited Image data processing method and apparatus
EP0551672A2 (en) 1992-01-13 1993-07-21 Koninklijke Philips Electronics N.V. Method of and system for transmitting and receiving digitally encoded video images
US5768437A (en) 1992-02-28 1998-06-16 Bri Tish Technology Group Ltd. Fractal coding of data
US5504484A (en) * 1992-11-09 1996-04-02 Matsushita Electric Industrial Co., Ltd. Variable-length data alignment apparatus for digital video data
US5563960A (en) 1993-01-22 1996-10-08 David Sarnoff Research Center, Inc. Apparatus and method for emphasizing a selected region in the compressed representation of an image
US5818877A (en) 1996-03-14 1998-10-06 The Regents Of The University Of California Method for reducing storage requirements for grouped data values
WO1998037700A1 (en) 1997-02-19 1998-08-27 Btg International Limited Progressive block-based coding for image compression
US6556719B1 (en) 1997-02-19 2003-04-29 University Of Bath Progressive block-based coding for image compression

Non-Patent Citations (17)

* Cited by examiner, † Cited by third party
Title
"A DCT-Based Image Coder," Xiong, Guleryuz and Orchard, IEEE Sig. Proc. Lett., vol. 3, No. 11, Nov. 1996, p. 289.
"Bit-Plane Encoding: A Technique For Source Encoding," Schwartz, J.W., Barker, R.C., IEEE Transactions on Aerospace and Electronic Systems, vol. 2, No. 4, 1966, pp. 385-392, XP002067595.
"Image Compression Using the Spatial-Orientation Tree," Said, A. and Pearlman, W.A., IEEE International Symposium on Circuits and Systems, 1993, (694), pp. 279-282.
"JPEG: Still Image Data Compression Standard," Pennebaker, W.B., Mitchell, J.L., 1993, Van Nostrand Reinhold, New York, NY USA XP00206756.
"Prioritized DCT For Compression and Progressive Transmission of Images," Yunning H., IEEE Transactions on Image Processing, vol. 1, No. 4, Oct. 1, 1992, pp. 477-487, XP000380776.
"Progressive Coding and Transmission of Digital Diagnostic Pictures," Elnahas, S.E. et al., IEEE Transactions on Medical Imaging, Jun. 1986, USA, vol. MI-5, No. 2 pp. 73-83, XP002067594.
"The JPEG Still Picture compression Standard," Wallace, G.K., Comm. ACM, vol. 34, No. 4, pp. 30-44, 1991.
"The JPEG Still Picture Compression Standard," Wallace, G.K., Communications of the Association For Computing Machinery, vol. 34, No. 4, Apr. 1, 1991, pp. 30-44 XP000228786.
Elnahas, S.E. et al., "Progressive Coding and Transmission of Digital Diagnostic Pictures," IEEE Transactions on Medical Imaging, Jun. 1986, MI-5 (2), 73-83.
International Patent Application No. PCT/GB1998/000360: International Search Report dated Jun. 25, 1998, 3 pages.
International Search Report for PCT/GB98/00360 mailed Jun. 25, 1998.
Pennebaker, W.B. et al., "JPEG: Still Image Data Compression Standard," Van Nostrand Reinhold, New York, 1993.
Said, A. et al., "Image Compression Using the Spatial-Orientation Tree," IEEE International Symposium on Circuits and Systems, 1993, 694, 279-282.
Schwartz, J.W. et al., "Bit-Plane Encoding: A Technique For Source Encoding," IEEE Transactions on Aerospace and Electronic Systems, 1966, 2(4), 385-392.
Wallace, G.K., "The JPEG Still Picture Compression Standard," Comm. ACM, 1991, 34(4), 30-44.
Xiong et al., "A DCT-Based Image Coder," IEEE Sig. Proc. Lett., Nov. 1996, 3(11), 289.
Yunming, H., "Prioritized DCT for Compression and Progressive Transmission of Images," IEEE Transactions on Image Processing, Oct. 1, 1992, 1(4), 477-487.

Also Published As

Publication number Publication date
EP0962099A1 (en) 1999-12-08
US6556719B1 (en) 2003-04-29
GB9703441D0 (en) 1997-04-09
EP0962099B1 (en) 2002-07-24
WO1998037700A1 (en) 1998-08-27
DE69806729T2 (en) 2003-04-03
DE69806729D1 (en) 2002-08-29

Similar Documents

Publication Publication Date Title
USRE43649E1 (en) Progressive block-based coding for image compression
US10045034B2 (en) System and method for using pattern vectors for video and image coding and decoding
JP4111644B2 (en) Embedded encoding / decoding method and system
US8422808B2 (en) Image encoding method, image decoding method, image encoding apparatus, image decoding apparatus, image encoding program, and image decoding program
CA2498384C (en) Context-based adaptive variable length coding for adaptive block transforms
Tsai et al. Stack-run image coding
US5339164A (en) Method and apparatus for encoding of data using both vector quantization and runlength encoding and using adaptive runlength encoding
US6272180B1 (en) Compression and decompression of reference frames in a video decoder
US6243496B1 (en) Data compression
US5757973A (en) Compression of image data seperated into frequency component data in a two dimensional spatial frequency domain
JPH05206867A (en) Apparatus and method for compression of data and data arrangement device used for them
AU7221498A (en) Signal coding and decoding
Andreopoulos et al. A hybrid image compression algorithm based on fractal coding and wavelet transform
In et al. On RD optimized progressive image coding using JPEG
Kim et al. Wavelet transform image compression using human visual characteristics and a tree structure with a height attribute
JP3114796B2 (en) Variable length coding apparatus and method
WO2005001771A1 (en) Method of encoding a signal into a bit stream
KR100295802B1 (en) Image data compression device using multiple variable encoding tables
KR100189523B1 (en) Image compression apparatus and method and decoding method by using image separation and arithmatic coding
PACKETS STACK-RUN ADAPTIVE WAVELET IMAGE COMPRESSION
KR100327364B1 (en) Image compression device using multiple huffman table
JPH03285460A (en) Picture data compressing and encoding system
JPH02308673A (en) Picture data coding system
Guo Analysis and Design of Lossless Bi-level Image Coding Systems

Legal Events

Date Code Title Description
AS Assignment

Owner name: AYSCOUGH VISUALS LLC, CALIFORNIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:MONRO, DONALD M.;REEL/FRAME:021973/0836

Effective date: 20040909

CC Certificate of correction
FEPP Fee payment procedure

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

FPAY Fee payment

Year of fee payment: 12

AS Assignment

Owner name: ZARBANA DIGITAL FUND LLC, DELAWARE

Free format text: MERGER;ASSIGNOR:AYSCOUGH VISUALS LLC;REEL/FRAME:037219/0345

Effective date: 20150811

AS Assignment

Owner name: HANGER SOLUTIONS, LLC, GEORGIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:INTELLECTUAL VENTURES ASSETS 161 LLC;REEL/FRAME:052159/0509

Effective date: 20191206