CN104349165B - The elongated decoding method of high-performance and device - Google Patents
The elongated decoding method of high-performance and device Download PDFInfo
- Publication number
- CN104349165B CN104349165B CN201410469348.5A CN201410469348A CN104349165B CN 104349165 B CN104349165 B CN 104349165B CN 201410469348 A CN201410469348 A CN 201410469348A CN 104349165 B CN104349165 B CN 104349165B
- Authority
- CN
- China
- Prior art keywords
- golomb
- value
- residual error
- encoded
- pixel
- 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.)
- Active
Links
Abstract
The invention discloses a kind of elongated decoding method of high-performance and device, it is encoded in coding side code device to the mapping value of pixel residual error, and specific coding method is:The Golomb coefficients Q of the mapping value of Huffman coding methods encoded pixels point residual error is used first, then the Golomb remainder Rs of the mapping value of Golomb coding methods encoded pixels point residual error are used, bit stream is decoded using decoding apparatus in decoding end, specific coding/decoding method is:Read the bit stream of multiple pixels first within a clock cycle, then the Q and R of each pixel of parallel decoding, go out the mapping value of the residual error of each pixel finally according to Q and R parallel computations, by the present invention, can effectively solve the problems, such as in existing variable length encoding method when low code efficiency and decoding apparatus that extension, area are big etc..
Description
Technical field
The present invention on a kind of decoding method and device, the property high of more particularly to a kind of video or image processing field
Can elongated decoding method and device.
Background technology
Image or video are a kind of effectively lively information representation modes, at present in fields such as TV, video conference and DVD
It is widely used in a variety of applications.In order to solve image or the shortcomings of cost of video storage is high and transmission belt is roomy, relevant group in the world
Knit and formulated video compression standard such as MPEG-x series and image compression standard JPEG series etc..In image or video compression standard
Main code method have Huffman encode, Golomb-Rice coding and arithmetic coding so that code efficiency is higher.
Huffman is encoded:Huffman codings are a kind of lossless entropy code modes, and its main coding thinking is:To going out
Existing probability information higher is distributed with less code word, and the less information distribution of probability of occurrence is with more code word.It has the disadvantage:
When information category relatively more (such as 256 kinds) and uniform probability distribution, compression efficiency will be very low, and decoding apparatus need to use many
The MUX of level, not only produces huge delay to cause that decoding apparatus area increases simultaneously.
Golomb-Rice is encoded:Golomb-Rice codings are a kind of lossless entropy code mode, its main code mode
It is:According to K values set in advance, coefficient Q, coding Q 0 are calculated first;Then bit flag position M is encoded, now M values must
It is 1;Finally use K coded remainder R.It has the disadvantage:When K values set in advance are inaccurate, compression efficiency will be very low, and
Decoding apparatus need to use multistage MUX, not only produce huge delay to cause that decoding apparatus area increases simultaneously.
The content of the invention
To overcome the shortcomings of that above-mentioned prior art is present, the purpose of the present invention is to provide a kind of elongated encoding and decoding of high-performance
Method and device, extension, area are big when it efficiently solves low code efficiency in existing variable length encoding method and decoding apparatus
Deng problem, realize the elongated encoding and decoding of high-performance.
It is that, up to above and other purpose, the present invention proposes a kind of high-performance variable length encoding method, comprises the following steps:
Step one, according to current line pixel value and encoded pixel value, calculates predicted value;
Step 2, calculates residual values and is mapped as nonnegative integer according to the predicted value and current pixel value;
Step 3, the information according to encoded pixel value calculates the required K values of Golomb-Rice codings;
Step 4, using the Golomb coefficient Q of the mapping value of Huffman coding methods encoded pixels point residual error, then adopts
With the Golomb remainder Rs of the mapping value of Golomb coding methods encoded pixels point residual error, all residual error mapping values to be encoded are entered
Row coding;
Step 5, the avriable length codes to being produced after coding carry out packing and produce final bit stream.
Further, step 4 further includes following steps:
Step 2.1, the Golomb of residual error mapping value to be encoded is calculated according to K values using residual error mapping value computing formula
Coefficient Q and Golomb remainder R;
Step 2.2, the Golomb coefficients Q of residual error mapping value to be encoded is encoded using Huffman coded systems;
Step 2.3, using Golomb coding methods and using the Golomb of binary coding residual error mapping value to be encoded
Remainder R;
Step 2.4, produces the Golomb coefficients Q and the Golomb remainder Rs polymerization change of residual error mapping value to be encoded
Long code word;
Step 2.5, circulation step 2.1~34 encodes all of residual error mapping value to be encoded.
Further, the residual error mapping value computing formula in step 2.1 is:
Dm=Q × 2k+R
Wherein, dm is residual error mapping value to be encoded, and Q is Golomb coefficients Q, and R is Golomb remainder Rs.
Further, in step 2.2, the code table pre-established according to the Huffman coded systems encodes the Golomb
Coefficient Q values.
To reach above-mentioned purpose, the present invention also provides a kind of high-performance variable length decoding method, comprises the following steps:
Step one, the bit stream most long needed for de-packaging operation produces Golomb-TT decoding apparatus is carried out to former bit stream;
Step 2, predicted value is gone out according to current line pixel value and encoded calculated for pixel values;
Step 3, the information according to encoded pixel value calculates the required K values of Golomb-Rice decodings;
Step 4, using Golomb the coefficients Q and Golomb of Golomb-TT decoding apparatus parallel decoding multiple pixel
Remainder R, and according to K values and Golomb coefficient Q and Golomb remainder Rs, each picture is calculated using residual error mapping value computing formula
The residual error mapping value of vegetarian refreshments;
Step 5, carries out reflection and penetrates and calculate current pixel value according to predicted value and current residue mapping value.
Further, in step 4, Golomb coefficient Q are decoded using Huffman decoding processes, is decoded from bit stream
Go out binary Golomb remainder Rs value, and according to K, Q and R value, each pixel is calculated using the residual error mapping value computing formula
Residual error mapping value.
Further, the residual error mapping value computing formula is:
Dm=Q × 2k+R
Wherein, dm is residual error mapping value to be decoded, and Q is Golomb coefficients Q, and R is Golomb remainder Rs.
To reach above-mentioned purpose, the present invention also provides a kind of high-performance elongated decoding apparatus, at least includes:
Module is unpacked, the bit stream most long needed for de-packaging operation produces Golomb-TT decoding apparatus is carried out to former bit stream;
Prediction module, predicted value is gone out according to current line pixel value and encoded calculated for pixel values;
K value computing modules, the information according to the encoded pixel value calculates the required K values of Golomb-Rice decodings;
Golomb-TT decoding apparatus, it is parallel to use in three bit streams of pixel are at least read in a clock cycle
Huffman decoding processes decode the Golomb coefficient Q of each pixel, and binary each pixel is decoded from bit stream
Golomb remainder Rs, and go out the mapping of the residual error of each pixel according to K values, Golomb coefficient Q and Golomb remainder R parallel computations
Value;
Anti- mapping block, carries out reflection and penetrates according to predicted value pix_pred with current residue mapping value, calculates current picture
Element value.
Further, the Golomb-TT decoding apparatus are using the Huffman in dichotomy decoding Huffman decoding processes
Code table, to reduce the series of the MUX in decoding Golomb coefficients Q.
Further, the Golomb-TT decoding apparatus decode multiple pixels using parallel mode within a clock cycle
Point component, and (n-1)th pixel component of parallel addition is consumed when the Golomb coefficient Q of nth pixel point component are decoded
Bit number.
Compared with prior art, the elongated decoding method of a kind of high-performance of the invention and device are by first using Huffman
The Golomb coefficients (Quotient, Q) of the mapping value of coding method encoded pixels point residual error, then using Golomb coding methods
The Golomb remainders (Remainder, R) of the mapping value of encoded pixels point residual error, and subtracted using dichotomy when decoding end is decoded
The series of MUX, while extending when being solved the problems, such as using parallel form, is effectively solved needed for few Huffman decodings
Code efficiency in existing variable length encoding method of having determined is low and extension, area big etc. problem during decoding apparatus.
Brief description of the drawings
The step of Fig. 1 is a kind of high-performance variable length encoding method of the invention flow chart;
Fig. 2 is an a kind of application scenarios schematic diagram of high-performance variable length encoding method of the invention;
Fig. 3 is the thin portion flow chart of step 104 in a kind of high-performance variable length encoding method of the invention;
The step of Fig. 4 is a kind of high-performance variable length decoding method of the invention flow chart;
Fig. 5 is a kind of system architecture diagram of the elongated decoding apparatus of high-performance of the invention;
Fig. 6 is each clock 3 waveform diagrams of the dm values of component of decoding in the present invention;
Fig. 7 is longest path schematic diagram when decoding the Q values in Golomb-TT decoding apparatus using conventional method;
Longest path schematic diagram Fig. 8 is to decode Q values in Golomb-TT decoding apparatus of the present invention when
Fig. 9 is the parallel decoding processes schematic diagram of Golomb-TT decoding apparatus proposed by the invention
Figure 10 is the schematic diagram that the present invention solves delay problem using parallel mode.
Specific embodiment
Below by way of specific instantiation and embodiments of the present invention are described with reference to the drawings, those skilled in the art can
Further advantage of the invention and effect are understood by content disclosed in the present specification easily.The present invention also can be different by other
Instantiation implemented or applied, the various details in this specification also can based on different viewpoints with application, without departing substantially from
Various modifications and change are carried out under spirit of the invention.
Before explaining the present invention, explanation first is defined to several nouns involved in the present invention:
(1) map:If A, B are two nonempty sets, if there is a rule F so as to each element a in A, press
Rule F, has the element b for uniquely determining to correspond to therewith in B, then F is called the mapping from A to B;
(2) residual values:The difference of predicted value and currency;
(3) MUX:In multichannel data transmit process, can be elected wherein any all the way as needed
Circuit;
(4) dichotomy:The method being divided into two;
The step of Fig. 1 is a kind of high-performance variable length encoding method of the invention flow chart, Fig. 2 is a kind of high-performance variable of the invention
One application scenarios schematic diagram of long codes method.As shown in Figures 1 and 2, a kind of high-performance variable length encoding method of the invention, bag
Include following steps:
Step 101, according to the encoded pixel value such as current line pixel value and lastrow (f (x-1, x-2 ... y-1, y-
2 ...) predicted value (pix_pred)) is calculated.
Step 102, calculates residual values and is mapped as nonnegative integer according to pix_pred and current pixel value (pix_cur)
(diffmap, dm), such as residual values be -255~255 dm values be 0~510.
Step 103, calculates according to the encoded pixel value f such as lastrow (x-1, x-2 ... y-1, y-2 ...) information
K values needed for Golomb-Rice codings.
Step 104, using the Golomb coefficients of the mapping value dm of Huffman coding methods encoded pixels point residual error
(Quotient, Q), then using Golomb coding methods encoded pixels point residual error mapping value dm Golomb remainders
All dm values to be encoded are encoded by (Remainder, R);For sake of convenience, step 104 is referred to as Golomb-TT codings
Method.
Step 105, the avriable length codes (code word, cw) to being produced after coding carry out packing and produce final bit stream
(bitstream, bs).
Table 1 is carried by existing block code method and tradition Golomb-Rice (K==2) coding methods with the present invention
Code word and the contrast table of required bit number that Golomb-TT (K==2) coding methods for going out are produced after being encoded to dm values, wherein
" x " represents that value can be 0 or 1, and { y } represents the code word size, and y can all nonnegative integers of value.Concrete meaning is in table 1:
First row:Four numbers are the index of one group of dm values;
Secondary series:Residual values by -255~255 when dm values be possible to value, wherein one group of 4 number of preceding 127 behavior,
Last column is one group of 128 behavior, 2 number;
3rd row:The code word and code word size of block code;
4th row:The code word and code word size of Golomb-Rice codings;
5th row:The code word and code word size of Golomb-TT codings;
The Golomb-TT coding methods proposed by the invention of table 1 are contrasted with existing coding method
Index value | Dm values | Block code | Golomb-Rice (K=2) | Golomb-TT (K=2) |
1 | 0~3 | 00000000xx{9} | 1xx{3} | 1xx{3} |
2 | 4~7 | 00000001xx{9} | 01xx{4} | 010xx{5} |
3 | 8~11 | 00000010xx{9} | 001xx{5} | 011xx{5} |
4 | 12~15 | 00000011xx{9} | 0001xx{6} | 0010xx{6} |
5 | 16~19 | 00000100xx{9} | 00001xx{7} | 0011xx{6} |
6 | 20~23 | 00000101xx{9} | 000001xx{8} | 00010xx{7} |
7 | 24~27 | 00000110xx{9} | 0000001xx{9} | 00011xx{7} |
8 | 28~31 | 00000111xx{9} | 00000001xx{10} | 000000xx{8} |
9 | 32~35 | 00001000xx{9} | 000000001xx{11} | 000001xx{8} |
10 | 36~39 | 00001001xx{9} | 0000000001dm values { 19 } | 000010xx{8} |
11 | 40~43 | 00001010xx{9} | 0000000001dm values { 19 } | 000011dm values { 15 } |
… | … | … | … | … |
127 | 504~507 | 1111110xx{9} | 0000000001dm values { 19 } | 000011dm values { 15 } |
128 | 508~510 | 1111111xx{9} | 0000000001dm values { 19 } | 000011dm values { 15 } |
As it can be seen from table 1 when index value is 2, code word size needed for Golomb-Rice codings is compiled for 4, Golomb-TT
Code word size needed for code is 5, and in the case of remaining, code word size needed for Golomb-TT codings is shorter than Golomb-Rice.
Table 2 is tradition Golomb-Rice (K==3) coding methods and Golomb-TT (K==3) proposed by the invention
Code word and the contrast of required bit number that coding method is produced after being encoded to dm values.
The Golomb-TT proposed by the invention of table 2 (K=3) coding methods and Golomb-Rice (K=3) coding method pair
Than
Index value | Dm values | Golomb-Rice (K=3) | Golomb-TT (K=3) |
1 | 0~7 | 1xxx{4} | 1xxx{4} |
2 | 8~15 | 01xxx{5} | 010xxx{6} |
3 | 16~23 | 001xxx{6} | 011xxx{6} |
4 | 24~31 | 0001xxx{7} | 0010xxx{7} |
5 | 32~39 | 00001xxx{8} | 0011xxx{7} |
6 | 40~47 | 000001xxx{9} | 00010xxx{8} |
7 | 48~55 | 0000001xxx{10} | 00011xxx{8} |
8 | 56~63 | 00000001xxx{11} | 000000xxx{9} |
9 | 64~71 | 000000001xxx{12} | 000001xxx{9} |
10 | 72~79 | 0000000001dm values { 19 } | 000010xxx{9} |
11 | 80~87 | 0000000001dm values { 19 } | 000011dm values { 15 } |
… | … | … | … |
63 | 496~503 | 0000000001dm values { 19 } | 000011dm values { 15 } |
64 | 504~510 | 0000000001dm values { 19 } | 000011dm values { 15 } |
From table 2 it can be seen that when index value is 2, code word size needed for Golomb-Rice codings is compiled for 5, Golomb-TT
Code word size needed for code is 6, and in the case of remaining, code word size needed for Golomb-TT codings is shorter than Golomb-Rice.
Draw to draw a conclusion from table 1,2 results and reasoning:
Conclusion 1:When K values are less than or equal to 3, Golomb-TT coding methods proposed by the invention have 98% probability excellent
In Golomb-Rice coding methods;
Conclusion 2:When the prediction of K values is accurate, Golomb-TT code efficiencies proposed by the invention are held with Golomb-Rice
It is flat;
Conclusion 3:When K value forecasting inaccuracies are true, Golomb-TT code efficiencies proposed by the invention will be significantly better than
Golomb-Rice;
Fig. 3 is the thin portion flow chart of step 104 in a kind of high-performance variable length encoding method of the invention.As shown in figure 3, step
104 Golomb-TT coding methods further include following steps:
Step 301, Q the and R values of dm values to be encoded, specific formula for calculation such as " eq1 " are calculated according to K values;
Dm=Q × 2k+R (eq 1)
Step 302, Q values are encoded using Huffman coded systems, i.e., (used in table 1,2 according to the code table for pre-establishing
Code table is referring to table 3) coding Q values, explanation:The code table can be adjusted according to the specific distribution probability of information.
Table 3 encodes a kind of Huffman code tables of Q values
Q | Code word |
0 | 1 |
1 | 010 |
2 | 011 |
3 | 0010 |
4 | 0011 |
5 | 00010 |
6 | 00011 |
7 | 000000 |
8 | 000001 |
9 | 000010 |
other | 000011 |
Step 303, using Golomb coding methods and using the R values of binary coding dm values to be encoded.
Step 304, produces Q and R polymerizations the avriable length codes cw of dm values to be encoded.
Step 305, circulation step 301~304 encodes all of dm values to be encoded.
The further comparative illustration Golomb-TT coding staffs proposed by the invention by taking the dm values that index value in table 2 is 3 as an example
Method.As known from Table 2, the span of dm is 16~23, can calculate its Q value according to formula " eq1 " and be " 2 ", and the value of R
Scope is 0~7;Show that the corresponding code word of Q is " 011 " then according to the code table shown in table 3;Then it is with binary representation R values
It is " xxx ", wherein the value of " x " is " 0 " or " 1 ";Code word " xxx " polymerization of the code word " 011 " and R of Q is finally produced code word
Cw " 011xxx ", explanation:The sequence that multi-form can be carried out according to real needs of Q and R, another common sequence in cw
Mode is " xxx011 ".Fig. 4 is using Golomb-TT coded systems of the present invention and Golomb-Rice coded systems in the example
The comparison diagram of coding schematic diagram is carried out to the dm values that index value in table 2 is 3.
The step of Fig. 5 is a kind of high-performance variable length decoding method of the invention flow chart.As shown in figure 4, the present invention is a kind of high
Performance variable length decoding method, comprises the following steps:
Step 501, is carried out most long needed for de-packaging operation produces Golomb-TT decoding apparatus possible to former bit stream (bs)
It is bs16 in bit stream, such as the present embodiment.
Step 502, according to the encoded pixel value such as current line pixel value and lastrow (f (x-1, x-2 ... y-1, y-
2 ...) predicted value (pix_pred)) is calculated.
Step 503, the information according to the encoded pixel value f such as lastrow (x-1, x-2 ... y-1, y-2 ...) is calculated
K values needed for Golomb-Rice decodings.
Step 504, decodes Q values, i.e., according to the code table decoding Q values for pre-establishing, from bit using Huffman decoding processes
Binary R values are decoded in stream (bs), and according to K, Q and R value, dm values is calculated using aforementioned formula " eq1 ".
Step 505, carries out reflection and penetrates and calculate current pixel value pix_ according to predicted value pix_pred with current dm values
cur。
Fig. 6 is a kind of system architecture diagram of the elongated decoding apparatus of high-performance of the invention, as shown in fig. 6, the present invention is a kind of high
Performance variable length decoding device, including:Unpack module 601, prediction module 602, K values computing module 603, Golomb-TT decoding dress
Put 604 and anti-mapping block 605.
Wherein, unpacking the former bit stream (bs) of module 601 pairs carries out the possible institute of de-packaging operation generation Golomb-TT decoding apparatus
The bit stream most long for needing, such as be bs16 in the present embodiment;Prediction module 602, according to current line pixel value and lastrow etc.
Encoded pixel values (f (x-1, x-2 ... y-1, y-2 ...)) calculate predicted value (pix_pred);K values computing module 603, according to
The information of the encoded pixel value f such as lastrow (x-1, x-2 ... y-1, y-2 ...) calculate Golomb-Rice decodings needed for K
Value;Golomb-TT decoding apparatus 604 decode Q using Huffman decoding processes decoding Q values according to the code table for pre-establishing
Value, decodes binary R values, and according to K, Q and R value, dm is calculated using aforementioned formula " eq1 " from bit stream (bs)
Value;Anti- mapping block 605 carries out reflection and penetrates and calculate current pixel value pix_ according to predicted value pix_pred with current dm values
cur。
In hardware realization, it is often necessary to weigh the factors such as area, efficiency and power consumption.During Video post-processing, very
It is first compression to be carried out to video source and the code stream after compression is sent into chip external memory in the case of many, from piece when needing
Readout code stream is decompressed in external memory, because Video post-processing technology has various such as noise reductions, FRC and enhancing, so
The decompressing device of multichannel is needed, and the work effect required by Video post-processing technology must is fulfilled for per decompressing device all the way
Rate, this just proposes the requirement of area and efficiency to the decoding apparatus of one of main modular in decompressing device.For example for
The video of YUV444 forms, method each clock (clock) of two kinds of improvement decoding apparatus proposed by the present invention can decode 3
Luminance component (compenont), that is, need Golomb-TT decoding apparatus each clock 3 dm values of decoding, as shown in Figure 7.
In Golomb-Rice and Golomb-TT decoding apparatus, time delay path most long is the decoded portion to Q values, is adopted
With traditional method, the former need to use 9 grades of MUXs, and the latter need to use 6 grades of MUXs, and Fig. 8 is that the latter uses tradition side
Method decode Q values latency path figure, by taking six bit stream bs [0]-bs [5] as an example, it is most long from starting point sp to terminal ep need to be by 6
Level MUX.
The present invention proposes the improvement to method as shown in Figure 8, i.e., using similar dichotomy to Golomb-TT decoding apparatus
In Q decoded, after improved method proposed by the present invention, Golomb-TT decoding apparatus 504 decoding Golomb-TT
In Q values, from starting point sp to terminal ep most long need to be by 4 grades of MUXs, as shown in Figure 9.Preferably implement of the invention
In example, Golomb-TT decoding apparatus decode Huffman code tables proposed by the invention using dichotomy, and it is concretely comprised the following steps:
step1:Judge whether low 4 be all 0 first;
step2:Then minimum sum-bit the 5th is judged parallel;
step3:Then first and second and five are judged parallel;
step4:Finally judge third and fourth and five.
In high performance variable length decoding device, it is desirable to multiple pixel components, such as 3 are decoded within a clock cycle
Individual component.When Golomb-TT decoding apparatus are designed with traditional method for designing, when elongatedness be:18 grades of MUX
With 3 the 6 of 2 ends adders.The present invention solves delay problem using parallel mode, specific as shown in Figure 10, Golomb-TT solutions
Q and R (dec_Q0, dec_Q1, dec_ of code device three components of parallel decoding bit stream bs_48 within a clock cycle
Q2, dec_R0, dec_R1, dec_R2), the bit that parallel addition one-component Q0 is consumed when second component Q1 is decoded
Number len_y0, the bit number len_y1 that second component of parallel addition is consumed when the 3rd component Q2 is decoded.Using this hair
Bright Golomb-TT decoding apparatus carry out Golomb-TT decodings, then elongatedness is when:12 grades of MUX and 12 end
6 adders.
In sum, the elongated decoding method of a kind of high-performance of the invention and device are by first using Huffman coding staffs
The Golomb coefficients (Quotient, Q) of the mapping value of method encoded pixels point residual error, then using Golomb coding methods coding picture
The Golomb remainders (Remainder, R) of the mapping value of vegetarian refreshments residual error, and reduced using dichotomy when decoding end is decoded
The series of MUX, while extending when being solved the problems, such as using parallel form, is efficiently solved needed for Huffman decodings
Code efficiency is low in existing variable length encoding method and extension, area big etc. problem during decoding apparatus.
The above-described embodiments merely illustrate the principles and effects of the present invention, not for the limitation present invention.Any
Art personnel can be modified above-described embodiment and are changed under without prejudice to spirit and scope of the invention.Therefore,
The scope of the present invention, should be as listed by claims.
Claims (10)
1. a kind of high-performance variable length encoding method, comprises the following steps:
Step one, according to current line pixel value and encoded pixel value, calculates predicted value;
Step 2, calculates residual values and is mapped as nonnegative integer according to the predicted value and current pixel value;
Step 3, the information according to encoded pixel value calculates the required K values of Golomb-Rice codings;
Step 4, using the Golomb coefficient Q of the mapping value of Huffman coding methods encoded pixels point residual error, then uses
All residual error mapping values to be encoded are carried out by the Golomb remainder Rs of the mapping value of Golomb coding methods encoded pixels point residual error
Coding;
Step 5, the avriable length codes to being produced after coding carry out packing and produce final bit stream.
2. a kind of high-performance variable length encoding method as claimed in claim 1, it is characterised in that step 4 further includes as follows
Step:
Step 2.1, the Golomb coefficients of residual error mapping value to be encoded are calculated according to K values using residual error mapping value computing formula
Q and Golomb remainder Rs;
Step 2.2, the Golomb coefficients Q of residual error mapping value to be encoded is encoded using Huffman coded systems;
Step 2.3, using Golomb coding methods and using the Golomb remainders of binary coding residual error mapping value to be encoded
R;
Step 2.4, produces the Golomb coefficients Q and the Golomb remainder Rs polymerization variable length code of residual error mapping value to be encoded
Word;
Step 2.5, circulation step 2.1~2.4 encodes all of residual error mapping value to be encoded.
3. a kind of high-performance variable length encoding method as claimed in claim 2, it is characterised in that the residual error mapping in step 2.1
Value computing formula is:
Dm=Q × 2k+R
Wherein, dm is residual error mapping value to be encoded, and Q is Golomb coefficients Q, and R is Golomb remainder Rs.
4. a kind of high-performance variable length encoding method as claimed in claim 2, it is characterised in that:In step 2.2, the Huffman
The code table pre-established according to coded system encodes the Golomb coefficient Q values.
5. a kind of high-performance variable length decoding method, comprises the following steps:
Step one, the bit stream most long needed for de-packaging operation produces Golomb-TT decoding apparatus is carried out to former bit stream;
Step 2, predicted value is gone out according to current line pixel value and encoded calculated for pixel values;
Step 3, the information according to encoded pixel value calculates the required K values of Golomb-Rice decodings;
Step 4, using Golomb coefficient Q and the Golomb remainders of Golomb-TT decoding apparatus parallel decoding multiple pixel
R, and according to K values and Golomb coefficient Q and Golomb remainder Rs, each pixel is calculated using residual error mapping value computing formula
Residual error mapping value;
Step 5, carries out reflection and penetrates and calculate current pixel value according to predicted value and current residue mapping value.
6. a kind of high-performance variable length decoding method as claimed in claim 5, it is characterised in that:In step 4, use
Huffman decoding processes decode Golomb coefficient Q, decode binary Golomb remainder Rs value from bit stream, and according to K,
Q and R values, the residual error mapping value of each pixel is calculated using the residual error mapping value computing formula.
7. a kind of high-performance variable length decoding method as claimed in claim 6, it is characterised in that:The residual error mapping value computing formula
For:
Dm=Q × 2k+R
Wherein, dm is residual error mapping value to be decoded, and Q is Golomb coefficients Q, and R is Golomb remainder Rs.
8. a kind of elongated decoding apparatus of high-performance, at least include:
Module is unpacked, the bit stream most long needed for de-packaging operation produces Golomb-TT decoding apparatus is carried out to former bit stream;
Prediction module, predicted value is gone out according to current line pixel value and encoded calculated for pixel values;
K value computing modules, the information according to the encoded pixel value calculates the required K values of Golomb-Rice decodings;
Golomb-TT decoding apparatus, it is parallel to use in three bit streams of pixel are at least read in a clock cycle
Huffman decoding processes decode the Golomb coefficient Q of each pixel, and binary each pixel is decoded from bit stream
Golomb remainder Rs, and go out the mapping of the residual error of each pixel according to K values, Golomb coefficient Q and Golomb remainder R parallel computations
Value;
Anti- mapping block, carries out reflection and penetrates according to predicted value pix_pred with current residue mapping value, calculates current pixel value.
9. a kind of elongated decoding apparatus of high-performance as claimed in claim 8, it is characterised in that:The Golomb-TT decoding apparatus
Using the Huffman code tables in dichotomy decoding Huffman decoding processes, selected with reducing the multichannel in decoding Golomb coefficients Q
Select the series of device.
10. a kind of elongated decoding apparatus of high-performance as claimed in claim 9, it is characterised in that:The Golomb-TT decoding apparatus
Multiple pixel components are decoded within a clock cycle using parallel mode, and in decoding nth pixel point component
The bit number that (n-1)th pixel component of parallel addition is consumed during Golomb coefficient Q.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410469348.5A CN104349165B (en) | 2014-09-16 | 2014-09-16 | The elongated decoding method of high-performance and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410469348.5A CN104349165B (en) | 2014-09-16 | 2014-09-16 | The elongated decoding method of high-performance and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104349165A CN104349165A (en) | 2015-02-11 |
CN104349165B true CN104349165B (en) | 2017-06-16 |
Family
ID=52503848
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410469348.5A Active CN104349165B (en) | 2014-09-16 | 2014-09-16 | The elongated decoding method of high-performance and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104349165B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105915227B (en) * | 2016-04-08 | 2019-03-15 | 苏州大学 | Adaptive mixed lossless date-compress system |
CN105915228B (en) * | 2016-04-08 | 2018-12-14 | 苏州大学 | Adaptive mixed data lossless compression method |
CN108155969B (en) * | 2016-12-02 | 2021-02-19 | 赛灵思公司 | Decoding circuit for variable length coded data |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102215385A (en) * | 2010-04-09 | 2011-10-12 | 中国科学院沈阳自动化研究所 | Real-time lossless compression method for image |
CN102368385A (en) * | 2011-09-07 | 2012-03-07 | 中科开元信息技术(北京)有限公司 | Backward block adaptive Golomb-Rice coding and decoding method and apparatus thereof |
CN102422540A (en) * | 2009-05-20 | 2012-04-18 | 日本电信电话株式会社 | Coding method, coding device, decoding method, decoding device, program, and recording medium |
CN102595496A (en) * | 2012-03-08 | 2012-07-18 | 西北大学 | Context-adaptive quotient and remainder encoding method used for sensing data of wireless sensing nodes |
CN103702129A (en) * | 2012-09-27 | 2014-04-02 | 联咏科技股份有限公司 | Video coding method and video coding device |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7586424B2 (en) * | 2006-06-05 | 2009-09-08 | Donald Martin Monro | Data coding using an exponent and a residual |
CN101925925B (en) * | 2008-01-21 | 2013-05-01 | 艾利森电话股份有限公司 | Prediction-based image processing |
-
2014
- 2014-09-16 CN CN201410469348.5A patent/CN104349165B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102422540A (en) * | 2009-05-20 | 2012-04-18 | 日本电信电话株式会社 | Coding method, coding device, decoding method, decoding device, program, and recording medium |
CN102215385A (en) * | 2010-04-09 | 2011-10-12 | 中国科学院沈阳自动化研究所 | Real-time lossless compression method for image |
CN102368385A (en) * | 2011-09-07 | 2012-03-07 | 中科开元信息技术(北京)有限公司 | Backward block adaptive Golomb-Rice coding and decoding method and apparatus thereof |
CN102595496A (en) * | 2012-03-08 | 2012-07-18 | 西北大学 | Context-adaptive quotient and remainder encoding method used for sensing data of wireless sensing nodes |
CN103702129A (en) * | 2012-09-27 | 2014-04-02 | 联咏科技股份有限公司 | Video coding method and video coding device |
Also Published As
Publication number | Publication date |
---|---|
CN104349165A (en) | 2015-02-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
RU2630750C1 (en) | Device and method for encoding and decoding initial data | |
KR20130018413A (en) | An image compression method with random access capability | |
US20130019029A1 (en) | Lossless compression of a predictive data stream having mixed data types | |
CN112399181B (en) | Image coding and decoding method, device and storage medium | |
CN104349165B (en) | The elongated decoding method of high-performance and device | |
CN101841707B (en) | High-speed real-time processing arithmetic coding method based on JPEG 2000 standard | |
EP2962399A1 (en) | Data encoder, data decoder and method | |
CN103167289A (en) | Method and device for coding and decoding image | |
CN101878651A (en) | Be used for the variable length code of the coefficient family of image and video compression | |
CN101502122A (en) | Encoding device and encoding method | |
CN105409129B (en) | Encoder device, decoder apparatus and method | |
CN104104954A (en) | Tile-based compression and decompression for graphic applications | |
CN101878646B (en) | Coding method and device for joint amplitude and position of video compression coefficients | |
CN103428502B (en) | Decoding method and decoding system | |
CN112449191A (en) | Method for compressing a plurality of images, method and apparatus for decompressing an image | |
CN102388538A (en) | Method and device for encoding an input bit sequence and corresponding decoding method and device | |
CN109660809A (en) | Based on the decoded colmv data lossless compression method of inter and system | |
CN101998122A (en) | Method and device for decoding normal form Hoffman hardware in JPEG (Joint Photographic Expert Group) image | |
WO2013011355A1 (en) | Method and apparatus for encoding an image | |
US20130170556A1 (en) | Variable length coding and decoding using counters | |
CN110545437B (en) | Coefficient encoding method, coefficient decoding method, electronic device, and medium | |
CN105472395B (en) | A kind of Lossless Image Compression Algorithm method based on discrete Krawtchouk orthogonal polynomial | |
TWI549437B (en) | Codec system and method | |
US20130188885A1 (en) | Apparatus and method for coding image, and non-transitory computer readable medium thereof | |
US20160323603A1 (en) | Method and apparatus for performing an arithmetic coding for data symbols |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant |