WO1997033254A1 - System and method for the fractal encoding of datastreams - Google Patents
System and method for the fractal encoding of datastreams Download PDFInfo
- Publication number
- WO1997033254A1 WO1997033254A1 PCT/US1997/003397 US9703397W WO9733254A1 WO 1997033254 A1 WO1997033254 A1 WO 1997033254A1 US 9703397 W US9703397 W US 9703397W WO 9733254 A1 WO9733254 A1 WO 9733254A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- interval
- datastream
- data
- compressor
- intervals
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/001—Model-based coding, e.g. wire frame
Definitions
- This invention relates to data compression methods, and more particularly, to lossless data compression methods.
- Compression generally refers to the representation of a set of data by using fewer elements than contained in the set to represent the set of data. If each data element within the set may be regenerated from the compressed representation then the method is commonly known as lossless. If the data representation regenerated from the compressed representation of the data set does not accurately represent each and every element within it, the method is known as lossy. Because the set of compressed data may represent all elements but contain fewer elements than the actual set represented, the transmission of the compressed data usually requires less time and fewer resources than if the original set of data was transmitted.
- Conversion of the transmitted compressed data set is called decompression and is usually performed by a receiver to obtain the original data set or a close approximation thereto
- the compressed data must not only be capable of accurately representing the data set, but it must also contain sufficient information for the receiver to decompress the data representation.
- One known method for compressing a data set uses a data set transformation function to represent a data set.
- the data set transformation function may be constructed out of projective, affine or complex mapping functions that, when applied to any data set in iterative fashion, converges to the same attractor.
- the coefficients of the data set transformation function may be used to represent a data set that is substantially the same as the attractor.
- This type of compression method is known as an iterated functton system or IFS. This type of compression method is lossy as the attractor usually approximates the data set and does not exactly reproduce the data set.
- mapping transformations computed from the representative datastream are then used to compress subsequent datastreams from the zero-order Markov source and the mapping transformations are utilized to construct a decompressor as well.
- the limitations of the systems and methods previously known have been overcome by a system and method performed in accordance with the principles of the present invention.
- the method of the present invention includes the steps of capturing a representative datastream from a data source, determining an optimal data set transformation function from the representative datastream, and incorporating the optimal data set transformation function in a compressor/decompressor combination to be used to compress and decompress datastreams from the data source.
- the optimal data set transformation is comprised of mapping functions which are preferably piecewise linear functions determined from the sample datastream by ordering intervals of an attractor in correspondence with the values of the data symbols comprising the datastream. These mapping functions more efficiently encode data by incorporating long-term data correlation present in the datastream generated by the data source.
- the data set transformation function constructed in accordance with the method of the present invention may also be used to construct a decompressor using the inverse of the optimal data set transformation function for efficiently decompressing the data from the compressor.
- a datastream is received from a data source which is considered a stochastic process since the generation of data symbols by the source are correlated in the long term.
- the cyclical permutations of the symbols comprising this datastream correspond to intervals of an attractor which forms the one-dimensional linear graph of the interval 0-1 which may be represented by the notation [0,1).
- intervals are then ordered and the graph of the mapping functions or the nodes of the mapping functions which comprise the data set transformation function are generated.
- mapping functions may then be incorporated in a compressor and used to compress other datastreams generated by the data source.
- the symbols of an incoming datastream are used to drive the compressor to determine an interval on the interval [0,1) which may be used to represent the datastream.
- the address of this interval is transmitted by the compressor to represent the datastream. After digits of the address have been shifted out, the interval is preferably renormalized so it conforms to sixteen bit binary representation.
- the decompressor of the present invention receives the address or a portion of the address and determines the portion of an interval identified by a dynamical system in the decompressor that corresponds to the address. This interval portion is used to generate and shift out a data symbol which was originally in the datastream before compression. A mapping function of the optimal data set transformation function corresponding to the data symbol is then applied to the identified interval and the resulting interval is renormalized. The process continues until all of the data symbols have been generated. The decompressor terminates the iterative process once the number of symbols generated equals the length of the original datastream. The length of the original datastream may be a parameter passed to the decompressor by the compressor.
- the system and method of the present invention may be used to generate compressor/decompressor combinations which more efficiently compress datastreams generated by stochastic processes.
- Such compressor/decompressor combinations are tailored to the output of a stochastic data source by incorporating the optimal data set transformation function based on the long term correlation of the symbols in the datastreams generated by the stochastic process.
- Fig. 1 is a block diagram of a system which implements the principles of the present invention
- Fig. 2 is graphical depiction of an example of the ordering of attractor intervals based on an exemplary datastream;
- Fig. 3 is a graphical depiction of the mapping functions which correspond to the intervals of Fig. 2;
- Fig. 4 is an example of ordering the intervals of an attractor based on the symbols in a datastream and the corresponding mapping functions
- Fig. 5 is a graphical depiction of a composition based on the mapping functions shown in Fig. 4;
- Fig. 6 is a flowchart of a compression process using mapping functions determined in accordance with the principles of the present invention.
- Fig. 7 is a flowchart of a decompression process using mapping functions determined in accordance with the principles of the present invention.
- Fig. 1 shows a system for generating the data set transformation function installed in a compressor and decompressor, respectively, for stochastic data sources.
- the system 10 includes a buffer 12, a data symbol arranger 14, a mapping function generator 18, a memory for mapping functions 16, a compressor 20, and a decompressor 24.
- Buffer 12 is comprised of memory storage elements which are used to store data symbols of a representative datastream captured from the output of a stochastic data source.
- the capacity of the storage buffer may be the same as a datastream normally output by the data source, the length of a standard message output by the data source, the length of the longest datastream output by the source, or the length of an average datastream.
- Data symbol arranger 14 is used to arrange the symbols of the datastream stored in buffer 12 into ordered strings or vectors based upon the values of the data symbols. These ordered strings are then provided to mapping function generator 18 for determination of the mapping functions which comprise the data set transformation function incorporated in compressor 20.
- the mapping functions are stored in memory 16 and transmitted to compressor 20 and decompressor 24 of a compressor/decompressor combination. Thereafter, the datastream is provided to compressor 20 for compression and the compressed representation is transmitted to decompressor 24.
- Decompressor 24 uses a dynamical system response to decompress or decode the compressed representation generated by compressor 20.
- Data symbol arranger 14, mapping function generator 18, memory 16, compressor 20, and decompressor 24 are preferably implemented by a computer program operating on a general purpose computer such as a personal computer or the like, although other computer systems may be used.
- the programs implementing each of these elements are preferably written in the C programming language and are executed on computer systems operating under a DOS or Windows environment, although other operating systems may be used.
- the computer platform has at least an Intel 80/286 processor, 8MB of RAM, 240MB of hard drive, or the equivalent thereof.
- Buffer 12 is preferably comprised of the RAM elements of the system, although a separate capture and storage board may be implemented for the buffer.
- mapping function generator 18 may be a graphical representation or a set of nodes defining the mapping transformation function which may be used in a compressor/decompressor combination.
- mapping functions output by mapping function generator 18 adapt a compressor/decompressor combination to more efficiently compress a datastream from a stochastic data source.
- the method and system of the present invention is used to identify an interval on a one- dimensional attractor on the range of [0,1). Preferably, this attractor is divided into k equal sized segments where k corresponds to the number of symbols in a datastream.
- the mapping transformation function for an iterated function system and its corresponding dynamical system is comprised of two mapping functions w 0 , w 1 which correspond to the data symbols 0,1, respectively.
- a dynamical system is a pair (f, X) where X is a space and f:X ⁇ X is a mapping.
- An iterated function system is a collection of mappings ⁇ w j ⁇ , w j :X ⁇ X. The inverse of a particular IFS is its corresponding dynamical system.
- X denotes the code space of all binary data strings. That is, the code space for semi-infinite binary data strings of zeros and ones.
- This (dynamical system, IFS) pair for the code space has a corresponding (dynamical system, IFS) pair on the unit interval.
- the function S(x) may be defined as:
- IFS dynamical system, IFS
- This may be used to express a datastream comprised of data symbols in the code space as an IFS operating on [0,1).
- IFS expression of the binary digit representation does not reduce the amount of data required to identify the datastream.
- IFS which reduces the amount of data required to identify a datastream for a class of datastreams.
- This class may be defined by its probability measure which is related to a probability distribution function, F, for the stochastical process identified by the output of a data source.
- F probability distribution function
- the stochastical process is assumed to be time invariant.
- the IFS compresses the datastream.
- the IFS expands the representation ofthe datastream.
- W is comprised of mapping functions w 0 ,w 1 and W and F are related as follows:
- a binary string of data symbols s 1 s 2 ...s n, with each symbol having a value of zero or one, may be represented by the interval w s1 ° w s2 ° ... ° w sn [0,1).
- This interval is determined by applying the mapping functions w 0 ,w 1 to the interval [0,1) in correspondence to the values of the data symbols comprising the string. That is, the address of the interval representing the string is found by determining w s2 ° w s1 ° ... ° w sn [0,1).
- the string 01101 is represented by the interval determined by w 0 w 1 w 1 w 0 w 1 [0,1).
- the method ofthe present invention enumerates all cyclic permutations of a datastream and orders them according to their binary magnitude.
- the order of the permutations correspond to an ordering of the intervals comprising an attractor on the interval [0,1).
- a datastream s 1 s 2 ... s n comprised of five binary digits 00111.
- Each binary digit is viewed as being the result of a mapping function applied to a prior interval which is represented by another one of the digits in the stream.
- the stream is also considered to be cyclical in that the first digit of the stream is deemed to be the result of the application of a mapping function to the last digit in the stream.
- the binary digits correspond to the intervals of an attractor on the interval [0-1) which are equal in length and are designated by T 1 , T 2 , T 3 , T 4 , and T 5 , respectively.
- interval T 1 Since the interval T 1 has a length 1/n then the representation of a string on n symbols is achieved with ⁇ log 2 n bits. Thus, identifying T 1 using the data values in buffer 12 compresses the data string contained in the buffer. This interval may used to optimally compress the representative datastream and the n circular shifts of that datastream. It also compresses datastreams of m bits which have at least log 2 n contiguous bits which are the same as any group of digits of the same length in the representative datastream used to generate W.
- This set of equations represents the relationships which are used to account for the correlation between successive binary digits. Specifically, the first binary digit 0 in the stream is for the interval T 1 and it is generated by applying the mapping function w 0 to the T 2 interval.
- the second binary digit 0 represents the application of the w 0 mapping function to the interval T 1 represented by the first binary digit 0 in the stream. Whether the interval is a result of applying the mapping function w 0 or w 1 is determined by the state of the binary digit itself. That is, if a binary digit is equal to 1 then it is the result of the application of the mapping function w 1 to the following interval. If a digit is equal to 0, it is the result of applying w 0 to the following interval.
- T 1 w 0 w 0 w 1 w 1 w 1 (T 1 ). This means that T 1 is the invariant interval that results from the application of W to T 1 according to the representative datastream in the example.
- the ordering of the T intervals defines the mapping relationships w 0, w 1 . If the data string used to determine the ordering of the intervals and consequently, the optimal IFS for the compressor, is compressed, the output of the encoder is the interval T 1 of length 1/n. This interval needs log 2 n or log 2 n + 1 bits to define it as opposed to the « bits required to define the original data string. Thus, the original data string has been losslessly compressed.
- mapping functions w 0 , w 1 are piecewise linear with nodes at the ends of the n intervals into which the interval [0,1] is divided.
- the functions w 0 ,w 1 have slopes of either 0 or 1 between the nodes. For any interval, the slope of w 0 and w 1 are different and w 0 (t) ⁇ t ⁇ w 1 (t). Using these relationships, the data symbol arranger 14 divides the datastream into two subsets where one subset is comprised of zero digit symbols and the other is comprised of one digit symbols.
- T 1 and T 2 must be equal to or less than T 3 , T 4 , and T 5 in the example above since the zero symbols represent intervals generated by the application of w 0 to a prior interval and the one symbols were generated by the application of the w 1 function.
- generator 18 determines that T 1 ⁇ T 3 since T 1 is the result of the double application to T 3 of w 0 and T 2 is the result of a single application of w 0 to T 3 . This relationship may be used to ascertain that T 1 ⁇ T 2 ⁇ T 3 . As a result, all five intervals may now be ordered to form the attractor for the range of 0 to 1 which corresponds to the input datastream. This ordering of the intervals represented by the binary digits of the datastream is shown graphically in Fig. 2.
- mapping functions w 0 and w 1 for the ordered intervals of Fig. 2 are also graphically shown in Fig. 3. Specifically, the intervals on which one of the functions w 0 or w 1 is increasing over the range means that the mapping function is being applied to the interval to generate a next interval. Thus, in the example shown in Fig. 3, w 0 is applied to interval T 1 while the mapping function w 1 is applied to the interval T 3 , and so on.
- the application of these mapping functions which comprise a mapping transformation W correspond to the correlation relationships noted above between the intervals and the mapping functions.
- the graphical representation of the mapping transformation or equivalent thereof may now be incorporated in a compressor to encode datastreams from the data source used to generate the mapping transformation W.
- the datastreams stored in buffer 12 may be generally described as a stream comprised of symbols s 1 , s 2 . . . s n .
- each symbol may be either a 1 or 0 digit.
- Data symbol arranger 14 begins by segmenting the datastream into vectors of contiguously valued symbols.
- the symbols z 1 , z 2 , . . . , z p represent contiguous zero digit vectors while n 1 , n 2 , . . . ,n p represent contiguous one digit vectors in the datastream.
- ,k(p) be the number of zeros in the corresponding z 1 , . . . ,z p vectors and let z ij be the jth zero in the z i vector.
- z 34 will denote the 4th zero in z 3 vector.
- l(1), . . . ,l(p) be the number of ones in the corresponding n 1 , . . . ,n p vectors and denote by n ij the/th one in the n i vector.
- the intervals T 1 ,T 2 , . . . ,T n-1 ,T n of the cyclic permutation may be identified with the corresponding starting data bit. That is:
- the elements of each z column are increasing from top to bottom so that any element on a row is smaller than any element on a lower row.
- the first group of inequalities follow from w 0 (t) ⁇ t. That is, all the elements of the top row of the z vectors corresponding to a one more application of w 0 than an element in a lower row.
- the elements of each n column decrease from top to bottom so that any element on a row is larger than any element on a lower row. That is, the top elements of the n vectors correspond to more than one application of w 1 , while the elements of the lower rows require less applications of w 1 .
- the ordering of the vectors begins by determining whether any of the n vectors are of non equal length. If they are then the z vectors are ordered first. Otherwise, the n vectors are ordered first.
- the criterion used to order the elements of the z vectors is an evaluation of length of the n vector follows a z vector. For example, to order z j and z k , the length of the n vector following z j is compared to the n vector following z k . The z vector followed by the shorter n vector is determined to be greater than the other z vector.
- n vector which follows z j is shorter, i.e., has fewer elements, than the n vector which follows z k , then z j > z k . If the n vectors which follow z j and z k are equal, then z j and z k are grouped together and the process proceeds by finding the position of the group ⁇ z j , z k ⁇ with respect to the other z vectors. The determination of the relationship between z j and z k is postponed for the following iteration.
- the process continues by performing the first iteration in the ordering of the it vectors. These vectors are ordered according to the order ofthe z vectors following each n vector.
- the second iteration for the ordering of the remaining z vectors uses the partial ordering of the n vectors. That is, a z vector followed by a n vector which is less than another n vector following a second z vector determines the ordering of the two z vectors.
- n vectors following z j and z k have been ordered so that the n vector following z j is less than the n vector following z k , then z j > z k , by the monotonicity of w 1 .
- Successive iterations of ordering the z and n vectors based on the ordering derived from the previous iteration eventually order all of the z and n vectors.
- an interval in a column is smaller than any interval in a column to its right.
- the intervals on a column are then ordered by examining the length of the string of ones that follow each one. In this case, ⁇ T 16 ,T 11 ⁇ T 5 since T 5 is followed by three ones while T 16 and T 11 are followed by two ones. This ordering provides ⁇ T 15 ,T 10 ⁇ T 4 and ⁇ T 14 ,T 6 ⁇ T 3 .
- T 11 and T 16 may now be ordered by applying w 0 to T 12 ⁇ T 1 to determine that T 11 ⁇ T 16 .
- all of the intervals in the image of w 0 have been determined and may be stated as: T 9 ⁇ T 14 ⁇ T 3 ⁇ T 10 ⁇ T 15 ⁇ T 4 ⁇ T 11 ⁇ T 16 ⁇ T 5 .
- T 13 and T 8 may now be ordered by applying w 1 to T 9 ⁇ T 14 to determine that T 8 ⁇ T 13 .
- all of the intervals in the image of w 1 have been determined and may be stated as: T 8 ⁇ T 11 ⁇ T 2 ⁇ T 7 ⁇ T 12 ⁇ T 1 ⁇ T 6 . Since any interval in the image of w 0 is smaller than any element in the image of w 1 then T 9 ⁇ T 14 ⁇ T 3 ⁇ T 10 ⁇ T 15 ⁇ T 4 ⁇ T 11 ⁇ T 16 ⁇ T 5 ⁇ T 8 ⁇ T 13 ⁇ T 2 ⁇ T 7 ⁇ T 12 ⁇ T 1 ⁇ T 6 .
- the intervals T correspond to the digits in the data string as follows:
- the zero digit vectors z i and the one digit vectors n i for the string are: or by letting the zeros be elements in the z vectors and the ones be elements in the n vectors, the string is:
- any interval on a row is smaller than any interval on a lower row because the interval on the lower row is a composition generated from fewer applications of the w 0 mapping function.
- the table above is:
- T 9 [0,1/16]
- T 14 (1/16,2/16]
- T 3 [2/16,3/16] and so forth.
- the graphs of the corresponding w 0 and w 1 are shown in Fig. 4.
- the graphical data of the mapping functions or equivalent thereof, such as identification of the nodes of the intervals, may be used by a compressor to encode data.
- the compressor is optimally "tuned" to encode a data stream like the one used to generate the mapping functions. Compression of a data string is achieved by using a composition operation with the piecewise linear maps, like the ones shown in Fig. 4, to identify an interval within the [0,1] range that represents the data stream.
- the nodes of the functions are identified.
- the nodes are the points in the function where the slope of the function changes between 0 and 1.
- the nodes for w 0 may be denoted as ⁇ (s i , x i ) ⁇ and the nodes for w 1 may be denoted as ⁇ (t i ,y i ) ⁇ .
- the graph of w 0 respective of w 1 is the line segment connecting the two node points.
- the composition w 0 o w 1 includes nodes from both w 0 and w 1 .
- w 0 o w 1 has the value w 0 (y i ) and for those s i in the image of w 1 , at the points w 1 -1 (s i ), w 0 ° w 1 has the value x i .
- the nodes of w 0 ° w 1 are ⁇ (t i , w 0 (y i )) ⁇ (w 1 -1 (s i ),x i ) ⁇ where some points may repeat each other.
- the composition operation may be used with the mapping functions of Fig. 4 for the binary substring 01 within the sixteen bit string discussed above in the following manner.
- the string 01 indicates a composition operation w 0 w 1 [0,1] which may be represented as w 0 o w 1 .
- the nodes (s i ,x i ) of w 0 are (0,0), (3/16,0), (9/16,6/16), (13/16,6/16), and (1,9/16) and the nodes (t i ,y i ) of w 1 are (0,9/16), (3/16,12/16), (9/16,12/16), (13/16,1) and (1,1).
- the nodes s i in the image of w 1 are (9/16, 6/16), (13/16,6/16), and (1,9/16). These values may be used in the equations given above to find the points ⁇ t i ,w 0 (y i ) ⁇ which define w 0 o w 1 . They are (0,6/16), (3/16,6/16), (9/16,6/16), (13/16,9/16), (1,9/16), (0,6/16), (10/16,6/16), (13/16,9/16), (14/16,6/16), (15/16,6/16), and (1,6/16). Because some of these nodes are redundant, w 0 ° w 1 may be determined by the nodes (0,6/16), (10/16,6/16), (13/16,9/16), and (1,9/16).
- the graph of w 0 o w 1 as shown in Fig. 5 indicates the string 01 is represented by the interval [6/16,9/16]. However, this interval does not fit within a dyadic interval.
- the compressor computes the composition corresponding to the application of the mapping function w for the next symbol s to w 0 w 1 .
- a string representing the symbol string may be transmitted. For example, if the composition example shown in Fig. 5 had defined an interval [1/8,3/8] then because the interval lies within the dyadic interval of [0,1/2] the bits defining that interval may be transmitted to a decompressor. Subsequent symbols in the string further define the interval that represents the string that lies within the interval [0,1/2].
- mapping functions generated by mapping function generator 18 are incorporated in a compressor, datastreams output by the data source which provided the representative datastream may be compressed.
- f denotes the interval defined by the composition operation.
- the system and method of the present invention preferably, renormalize the interval identified by the composition operation to the [0,1) interval. Renormalization is required because an increase in the number of symbols in the datastream to be compressed decreases the size of the interval which represents the datastream which, in turn, increases the number of bits required to define the interval address. The process continues until no more data symbols remain in the datastream to process.
- Renormalization includes two elements.
- the first element is the shifting of bits representing an interval endpoint following the storage of bits that define a dyadic interval. This element is performed by shifting in bits according to the function S(x) which is comprised of:
- a renormalization R G o S d1 ° S d2 ° ... ° S dm where G may be applied to an address one or more times and S d1 etc. represent the shifting operation which corresponds to the digit shifted out of the address.
- processing of the next symbol s k in a datastream may result in the application of a corresponding mapping function which generates an interval with left and right endpoints defined by the addresses:
- the variable count indicates the number of underflow bits counted which have not yet resulted in the interval being within a dyadic interval.
- the left and right endpoints both begin with a '1' so the interval lies within the dyadic interval (1/2,1) and a '1' may be transmitted or stored for the string being processed.
- This output bit is followed by three output '0' bits which correspond to the count variable value.
- the '0' value is used to move the left endpoint into the interval [0,1/2].
- the remaining bits in the addresses of the two endpoints, starting from the left, which are the same, indicate stability of the interval address and they may be shifted out of the address into the output string.
- the output string corresponding to all of these actions is 10001001.
- the first bits indicate the interval is located across the halfway point of the [0,1] interval.
- the G function is applied to the interval until this condition is reached.
- the G function is applied twice to achieve that function so count is set equal to 2.
- These bits are then removed from each endpoint address and the number of bits needed to form a sixteen bit value are shifted into the least significant bit positions. For the address of the left endpoint the values shifted in are '0' and for the right endpoint the values are '1'. This further ensures that the interval approximates the [0,1) interval.
- the interval resulting from the renormalization R G ° G ° S 1 o S 0 ° S 0 ° S 1 ° S 1 o f has the address values:
- Compression continues by receiving the next symbol in the string being compressed and following the steps until all of the symbols in the datastream have been processed.
- the decoder operation is the inverse of the encoder operation.
- the process is shown in Fig. 7.
- the process receives a number c from compressor 20 (Block 80) and determines the relationship between c and the interval identified by the decompressor (Block 84).
- the identified interval is f k-1 ° w 0 [0,1) which is the interval identified by the previous iteration of the composition operation performed by decompressor 24. If the number is less than the interval, a data symbol of value '0' is shifted out and if the number is less than the interval, a data symbol of value '1' is shifted out (Block 88).
- mapping function corresponding to the shifted out digit is selected. For the '0' digit, w 0 is selected and for '1', w 1 is selected.
- the renormalized c is processed at the next iteration, if there is one. The process continues until all of the data symbols for the compressed datastream have been generated (Block 94).
- the number of symbols to be generated is passed as a parameter with c and the number decremented after each iteration until it reaches zero.
- a user captures a sample datastream from a data source and determines the mapping functions for an optimal data set transformation function which correspond to the representative datastream.
- These mapping functions are incorporated in a compressor and a decompressor of a compressor/decompressor combination. Thereafter, datastreams are input to the compressor which use the mapping functions of the optimal data set transformation function to determine composition operations which identify an interval of an attractor, the address of which may be used to represent the stream or a portion of the stream.
- This representation is provided to a decompressor which uses the number or a portion of the number generated by the compressor to generate the data symbols of the compressed datastream. The symbols are generated by determining the relationship between the number generated by the compressor and an interval defined by the composition operation in the decompressor.
- the mapping function which corresponds to the generated data symbol is then applied in the composition operation to the interval and the resulting interval and number are renormalized to generate the identified interval for the next iteration. The process continues until all of the data symbols of the compressed datastream have been
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
Description
Claims
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
AU21958/97A AU2195897A (en) | 1996-03-04 | 1997-03-04 | System and method for the fractal encoding of datastreams |
EP97914861A EP0885429A1 (en) | 1996-03-04 | 1997-03-04 | System and method for the fractal encoding of datastreams |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US08/610,302 US5857036A (en) | 1996-03-04 | 1996-03-04 | System and method for the fractal encoding of datastreams |
US08/610,302 | 1996-03-04 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1997033254A1 true WO1997033254A1 (en) | 1997-09-12 |
Family
ID=24444491
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/US1997/003397 WO1997033254A1 (en) | 1996-03-04 | 1997-03-04 | System and method for the fractal encoding of datastreams |
Country Status (4)
Country | Link |
---|---|
US (1) | US5857036A (en) |
EP (1) | EP0885429A1 (en) |
AU (1) | AU2195897A (en) |
WO (1) | WO1997033254A1 (en) |
Families Citing this family (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8352400B2 (en) | 1991-12-23 | 2013-01-08 | Hoffberg Steven M | Adaptive pattern recognition based controller apparatus and method and human-factored interface therefore |
US6400996B1 (en) | 1999-02-01 | 2002-06-04 | Steven M. Hoffberg | Adaptive pattern recognition based control system and method |
US6850252B1 (en) * | 1999-10-05 | 2005-02-01 | Steven M. Hoffberg | Intelligent electronic appliance system and method |
US6418424B1 (en) | 1991-12-23 | 2002-07-09 | Steven M. Hoffberg | Ergonomic man-machine interface incorporating adaptive pattern recognition based control system |
US10361802B1 (en) | 1999-02-01 | 2019-07-23 | Blanding Hovenweep, Llc | Adaptive pattern recognition based control system and method |
US6421464B1 (en) * | 1998-12-16 | 2002-07-16 | Fastvdo Llc | Fast lapped image transforms using lifting steps |
US7966078B2 (en) | 1999-02-01 | 2011-06-21 | Steven Hoffberg | Network media appliance system and method |
US7366719B2 (en) | 2000-01-21 | 2008-04-29 | Health Discovery Corporation | Method for the manipulation, storage, modeling, visualization and quantification of datasets |
EP1252588B1 (en) * | 2000-01-21 | 2007-06-27 | Health Discovery Corporation | Method for the manipulation, storage, modeling, visualization and quantification of datasets |
US20050026199A1 (en) * | 2000-01-21 | 2005-02-03 | Shaw Sandy C. | Method for identifying biomarkers using Fractal Genomics Modeling |
US20050079524A1 (en) * | 2000-01-21 | 2005-04-14 | Shaw Sandy C. | Method for identifying biomarkers using Fractal Genomics Modeling |
US20050158736A1 (en) * | 2000-01-21 | 2005-07-21 | Shaw Sandy C. | Method for studying cellular chronomics and causal relationships of genes using fractal genomics modeling |
US6411228B1 (en) | 2000-09-21 | 2002-06-25 | International Business Machines Corporation | Apparatus and method for compressing pseudo-random data using distribution approximations |
US20030206181A1 (en) * | 2001-04-13 | 2003-11-06 | Abb Ab | System and method for organizing two and three dimensional image data |
US20080075377A1 (en) * | 2003-07-29 | 2008-03-27 | Topiwala Pankaj N | Fast lapped image transforms using lifting steps |
CN105680868B (en) * | 2014-11-17 | 2019-04-12 | 华为技术有限公司 | The method and apparatus of compressed stream data |
CA3042478A1 (en) * | 2016-08-31 | 2018-03-08 | Pointerra Technologies Pty Ltd | Method and system for storing and retrieving multi-dimensional data |
DE102021106488A1 (en) * | 2020-12-23 | 2022-06-23 | Arnold & Richter Cine Technik Gmbh & Co. Betriebs Kg | Background display device, background display system, recording system, camera system, digital camera and method for controlling a background display device |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1992017981A1 (en) * | 1991-03-28 | 1992-10-15 | Siemens Aktiengesellschaft | Process for compressing video images |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4941193A (en) * | 1987-10-02 | 1990-07-10 | Iterated Systems, Inc. | Methods and apparatus for image compression by iterated function system |
US5065447A (en) * | 1989-07-05 | 1991-11-12 | Iterated Systems, Inc. | Method and apparatus for processing digital data |
US5416856A (en) * | 1992-03-30 | 1995-05-16 | The United States Of America As Represented By The Secretary Of The Navy | Method of encoding a digital image using iterated image transformations to form an eventually contractive map |
-
1996
- 1996-03-04 US US08/610,302 patent/US5857036A/en not_active Expired - Lifetime
-
1997
- 1997-03-04 WO PCT/US1997/003397 patent/WO1997033254A1/en not_active Application Discontinuation
- 1997-03-04 AU AU21958/97A patent/AU2195897A/en not_active Abandoned
- 1997-03-04 EP EP97914861A patent/EP0885429A1/en not_active Withdrawn
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1992017981A1 (en) * | 1991-03-28 | 1992-10-15 | Siemens Aktiengesellschaft | Process for compressing video images |
Non-Patent Citations (2)
Title |
---|
EDALAT A: "Domain theory in stochastic processes", PROCEEDINGS TENTH ANNUAL IEEE SYMPOSIUM ON LOGIC IN COMPUTER SCIENCE (CAT. NO.95CB35768), PROCEEDINGS OF TENTH ANNUAL IEEE SYMPOSIUM ON LOGIC IN COMPUTER SCIENCE, SAN DEIGO, CA, USA, 26-29 JUNE 1995, ISBN 0-8186-7050-9, 1995, NEW YORK, NY, USA, IEEE, USA, pages 244 - 254, XP000676043 * |
HUIGUO LUO ET AL: "Markov iterated function system model of images", APPLICATIONS OF DIGITAL IMAGE PROCESSING XV, SAN DIEGO, CA, USA, 21-24 JULY 1992, vol. 1771, ISSN 0277-786X, PROCEEDINGS OF THE SPIE - THE INTERNATIONAL SOCIETY FOR OPTICAL ENGINEERING, 1993, USA, pages 598 - 604, XP000676044 * |
Also Published As
Publication number | Publication date |
---|---|
US5857036A (en) | 1999-01-05 |
EP0885429A1 (en) | 1998-12-23 |
AU2195897A (en) | 1997-09-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5857036A (en) | System and method for the fractal encoding of datastreams | |
EP0695040B1 (en) | Data compressing method and data decompressing method | |
US5818877A (en) | Method for reducing storage requirements for grouped data values | |
US5003307A (en) | Data compression apparatus with shift register search means | |
US5717393A (en) | Apparatus for data compression and data decompression | |
US5374916A (en) | Automatic electronic data type identification process | |
US5045852A (en) | Dynamic model selection during data compression | |
US5973630A (en) | Data compression for use with a communications channel | |
US5467087A (en) | High speed lossless data compression system | |
US7365658B2 (en) | Method and apparatus for lossless run-length data encoding | |
US6445314B1 (en) | System and method for the decoding of variable length codes | |
US5959560A (en) | Data compression via alphabet partitioning and group partitioning | |
US20040202374A1 (en) | Compression system for integrated sensor devices | |
CN101095284A (en) | Device and data method for selective compression and decompression and data format for compressed data | |
US5353024A (en) | Method for data compression having an improved encoding algorithm which utilizes a token stacking technique | |
US6919826B1 (en) | Systems and methods for efficient and compact encoding | |
US5594435A (en) | Permutation-based data compression | |
KR20120018360A (en) | Method for variable length coding and apparatus | |
WO2001063772A1 (en) | Method and apparatus for optimized lossless compression using a plurality of coders | |
US6055273A (en) | Data encoding and decoding method and device of a multiple-valued information source | |
US6292115B1 (en) | Data compression for use with a communications channel | |
Arnavut | Inversion coding | |
JP2536422B2 (en) | Data compression device and data decompression device | |
US6794999B1 (en) | Resilient parameterized prefix codes for adaptive coding | |
CN114337680B (en) | Compression processing method and device, storage medium and electronic equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AL AM AT AU AZ BA BB BG BR BY CA CH CN CU CZ DE DK EE ES FI GB GE GH HU IL IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK TJ TM TR TT UA UG UZ VN YU AM AZ BY KG KZ MD RU TJ TM |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GH KE LS MW SD SZ UG AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT SE BF BJ |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 1997914861 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: JP Ref document number: 97531903 Format of ref document f/p: F |
|
WWP | Wipo information: published in national office |
Ref document number: 1997914861 Country of ref document: EP |
|
REG | Reference to national code |
Ref country code: DE Ref legal event code: 8642 |
|
NENP | Non-entry into the national phase |
Ref country code: CA |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1997914861 Country of ref document: EP |