CN109660809A - Based on the decoded colmv data lossless compression method of inter and system - Google Patents
Based on the decoded colmv data lossless compression method of inter and system Download PDFInfo
- Publication number
- CN109660809A CN109660809A CN201811092181.XA CN201811092181A CN109660809A CN 109660809 A CN109660809 A CN 109660809A CN 201811092181 A CN201811092181 A CN 201811092181A CN 109660809 A CN109660809 A CN 109660809A
- Authority
- CN
- China
- Prior art keywords
- compression
- data
- difference
- bit
- colmv
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
- H04N19/423—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation characterised by memory arrangements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
- H04N19/43—Hardware specially adapted for motion estimation or compensation
- H04N19/433—Hardware specially adapted for motion estimation or compensation characterised by techniques for memory access
Abstract
The present invention provides a kind of based on the decoded colmv data lossless compression method of inter, colmv data to be compressed are divided into the compression section of independent compression one by one, each compression section is divided into compression unit one by one according to M kind compact model respectively, keep the first data in compression section constant, then the difference of adjacent data is successively found out, then finds out the maximum value in each compression unit difference;Total bit number of the various compact models of each compression unit is found out with the digit+1 of maximum value;Compare the size for total bit number that all compact models finally obtain, the smallest corresponding compact model of total bit number is selected as the last compact model of corresponding compression section, if certain compression sections upon compression than not compressing also big when, this independent compression unit replicates initial value.Present invention employs lossless compression algorithm and difference sectional compression method, and the characteristics of combine colmv data, compression ratio can be further improved, within make that the code stream obtained after compression is mostly original 30%.
Description
Technical field
The present invention relates to inter (interframe) decodings to remove read-write colmv (reference information for assisting frame) data method, especially relates to
And a kind of lossless compression method when colmv reading and writing data decoded based on inter.
Background technique
It is not have when read-write colmv (reference information for assisting frame) data are gone in existing hardware structure inter (interframe) decoding
By compression processing, colmv data are directly stored in ddr.
Colmv committed memory bandwidth is very big it can be seen from below table, especially h264 (high compression number
Video Codec standard) accounting it is bigger, for 420, acquire accounting (read-write two-way) and reach 66.67%.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of based on the decoded lossless date-compress side colmv inter
Method and system use lossless compression algorithm, mainly difference sectional compression method, and the characteristics of combination colmv data, can be more
It is further to improve compression ratio.
The method of the present invention is achieved in that a kind of based on the decoded colmv lossless compression method of inter, comprising:
Step S11, in colmv data to be compressed, by every 2nCompression section of a data as an independent compression, often
A compression section is divided into compression unit one by one according to M kind compact model respectively, wherein n is natural number, and M is voluntarily set by user
It is fixed;
Step S12, it keeps the first data in compression section constant, then successively finds out the difference of adjacent data, then find out each
Maximum value in compression unit difference;
Step S13, use the digit+1 of maximum value as the subsegment code length of the compression unit, further according to this subsegment code
The long difference subsequent compression unit is set out one by one to be come;
Step S14, total bit number of the various compact models of each compression unit is found out respectively;
Step S15, it is the smallest corresponding to select total bit number for the size for comparing total bit number that all compact models finally obtain
Compact model as corresponding compression section last compact model, if certain compression sections upon compression than not compressing also big when,
Then individually this compression unit replicates initial value.
Further, when 2nValue be 64 and M=4 when, then 4 kinds of modes are respectively as follows:
Mode0: point 8 compression units, 8 data of each compression unit;
Mode1: point 4 compression units, 16 data of each compression unit;
Mode2: point 2 compression units, 32 data of each compression unit;
Mode3: point 1 compression unit, 64 data of this compression unit.
Further, in the step S13, the virtual value of the subsegment code length is 0 and integer 2~15;
Group segment encode it is a length of 1 when, compressed bit stream main body is the initial data of respective compression units;
When a length of integer 2~15 of group segment encode, compressed bit stream main body be then in order by each difference by fixed bit wide,
And pieced together with " sign bit+difference absolute value of difference " combination, the sign bit of difference is 0, indicates that the difference is positive number;Difference
Sign bit be 1, indicate the difference be negative.
Further, in the step S13, total bit number be it is described under a certain compact model " sign bit of difference+
The summation of the bit number of the bit wide of the absolute value of difference ".
Present system is achieved in that a kind of based on the decoded colmv lossless compression system of inter, including coding
Device, the first temporary storage, DDR, the second temporary storage and decoder;
The encoder is stored colmv data to be compressed by the compressed code stream of any one of Claims 1-4 method
Into the first temporary storage, it is then written in DDR after reaching certain condition;
The decoder is banished in second temporary storage from DDR sense code again when decompressing, then is decoded.
It further, is to be sequentially read out the data being stored in temporary storage, then join when the decoder decompresses
According to following table, the compact model mode_sel of each compression section is parsed, then parses the subsegment code length sub_ of compression unit
Bit_len parses first initial data of compression section, then parses compressed code according to subsegment code length sub_bit_len
Main body bit_stream is flowed, obtains the corresponding sign bit of each difference and absolute value, then calculate initial value, and group segment encode is long
When sub_bit_len=1, then it represents that current compression unit is not compressed, and in code stream is raw value;
The beneficial effect that the present invention obtains is: present invention employs the difference sectional compression methods of lossless compression algorithm, and
And the characteristics of combining colmv data, it can further improve compression ratio.When bigger (the i.e. consecutive number of the correlation of image data
Relatively according to numerical value), finding out the difference come will be smaller, and compression effectiveness will be better.Because colmv data the characteristics of be phase
Closing property is very big, so be well suited for using this scheme of the present invention, the code stream obtained after compression is mostly within original 30%,
Largely alleviate the pressure that memory reads bandwidth.
Detailed description of the invention
The present invention is further illustrated in conjunction with the embodiments with reference to the accompanying drawings.
Fig. 1 is the method for the present invention execution flow chart.
Fig. 2 is the structural block diagram of present system.
Fig. 3 is the internal module block diagram of the encoder of present system.
Fig. 4 is the execution flow diagram of the encoder of present system.
Fig. 5 is the internal module block diagram of the decoder of present system.
Fig. 6 is the execution flow diagram of the decoder of present system.
Fig. 7 is the compression accounting schematic diagram of the method for the present invention and system.
Specific embodiment
Refering to Figure 1, it is of the invention based on the decoded colmv lossless compression method of inter, it is lossless using variable length code
Compression algorithm, " variable length code " here refer to that every section of code length is unfixed, comprising:
Step S11, in colmv data to be compressed, by every 2n(data amount check of a section can be adjusted as needed
Whole, in application process, can be selected for needing the characteristic of data source that compress to trade off) a data are as an independence
The compression section of compression, each compression section are divided into compression unit one by one according to M kind compact model respectively, wherein n is nature
Number, M is by user's sets itself.
Step S12, it keeps the first data in compression section constant, then successively finds out the difference of adjacent data, then find out each
Maximum value in compression unit difference.
Step S13, use the digit+1 of maximum value as the subsegment code length of the compression unit, further according to this subsegment code
The long difference subsequent compression unit is set out one by one to be come;
The virtual value of the subsegment code length is 0 and integer 2~15;
Group segment encode it is a length of 1 when, compressed bit stream main body is the initial data of respective compression units;
When a length of integer 2~15 of group segment encode, compressed bit stream main body be then in order by each difference by fixed bit wide,
And pieced together with " sign bit+difference absolute value of difference " combination, the sign bit of difference is 0, indicates that the difference is positive number;Difference
Sign bit be 1, indicate the difference be negative.
Step S14, total bit number of the various compact models of each compression unit is found out respectively;Total bit number is a certain
The summation of the bit number of the bit wide of " sign bit+difference absolute value of difference " under compact model.
Step S15, it is the smallest corresponding to select total bit number for the size for comparing total bit number that all compact models finally obtain
Compact model as corresponding compression section last compact model, if certain compression sections upon compression than not compressing also big when,
Then individually this compression unit replicates initial value.
For example:
By taking 64 colmv data are compression section (segment) as an example, facilitate tile (tile: by image block, one
As be longitudinal piecemeal, tile must be rectangle) or slice (slice: using ctu as unit scribing, dividing the image into multiple)
Data are read when having division.
Then each compression section is divided into compression unit one by one, i.e. M=4 further according to 4 kinds of modes respectively, described 4 kinds
Mode is respectively as follows:
Mode0: point 8 compression units, 8 data of each compression unit;
Mode1: point 4 compression units, 16 data of each compression unit;
Mode2: point 2 compression units, 32 data of each compression unit;
Mode3: point 1 compression unit, 64 data of this compression unit.It is as shown in the table:
There are 64 data in one compression section, and the parsing of variable length code lossless compression algorithm is as follows:
Step1: the data source A of compression section is extracted:
OrgA0, OrgA1, OrgA2..., OrgA63;
Assuming that data flow (assuming that data source maximum bit wide is 16) are as follows:
1100,1090,1082,1081,1085,1079,1082,1084,1086,1089,1090,1085,1080,
1083,1088,1086 ...
Step2: difference diffA is sought:
OrgA0, diffA1, diffA2..., diffA63;
First data OrgA0It is constant, diffA1=OrgA1-OrgA0, and so on, it obtains:
1100, -10, -8,-Isosorbide-5-Nitrae, -6,3,2,2,3,1, -5, -5,3,5, -2 ...
Step3: the maximum value of difference diffA in each compression unit in various modes: abs_mN:M is found out respectively
It is 0~3, shown herein as various mode M, maximum value abs_max:mode0 in compression unit: with 8 data for one
Compression unit;
Abs_m0 [0]=max { abs { diffA1..., diffA7}};// the first data OrgA0Disregard, only 7
Number;
Abs_m0 [1]=max { abs { diffA8..., diffA15}};// 8 numbers;
Abs_m0 [2]=max { abs { diffA16..., diffA23}};
Abs_m0 [3]=max { abs { diffA24..., diffA31}};
Abs_m0 [4]=max { abs { diffA32..., diffA39}};
Abs_m0 [5]=max { abs { diffA40..., diffA47}};
Abs_m0 [6]=max { abs { diffA48..., diffA55}};
Abs_m0 [7]=max { abs { diffA56..., diffA63}};
It is obtained in example:
Abs_m0 [0]=max { abs { -10, -8,-Isosorbide-5-Nitrae, -6,3,2 } }=10;
Subsegment code length sub_bit_len=log2(10)+1=4+1=5bit;
Abs_m0 [1]=max { abs { 2,3,1, -5, -5,3,5, -2 } }=5;
Subsegment code length sub_bit_len=log2(5)+1=3+1=4bit;
...
Mode1: with 16 data for a compression unit
Abs_m1 [0]=max { abs { diffA1..., diffA15}};
Abs_m1 [1]=max { abs { diffA16.., diffA31}};
Abs_m1 [2]=max { abs { diffA32..., diffA47}};
Abs_m1 [3]=max { abs { diffA48..., diffA63}};
It is obtained in example:
Abs_m1 [0]=max { abs { -10, -8,-Isosorbide-5-Nitrae, -6,3,2,2,3,1, -5, -5,3,5, -2 } }
=10;
Subsegment code length sub_bit_len=log2(10)+1=4+1=5bit;
...
Mode2: with 32 data for a compression unit
Abs_m2 [0]=max { abs { diffA1..., diffA31}};
Abs_m2 [1]=max { abs { diffA32..., diffA63}};
Mode3: with 64 data for a compression unit
Abs_m3 [0]=max { abs { diffA1..., diffA63}};
Step4: total bit number of various modes is found out respectively.
The table shows that compressed code stream pieces together format, when calculating total bit number of various modes, and according to this
Code stream call format counts.
Mode_sel in table is used to indicate any in 4 kinds of modes;
Sub_bit_len is subsegment code length, and virtual value is 0 or [2:15];When sub_bit_len is 0: indicating the compression
All, the difference of adjacent data is 0 to the value of all data of unit, and such case is not in.When sub_bit_len is 1
The case where, because the maximum value digit of difference is greater than 0 when difference is non-zero, and also need along with a bit sign
Position, sub_bit_len is at least 2 in this case, and meaning is expressed as follows situation: when the maximum of the compression unit is exhausted
To value >=15, as soon as along with sign bit, then the compression unit compression does not have advantage, compressed bit stream main body at this time
Bit_stream is exactly the initial data of the compression unit.
Bit_stream be compressed bit stream main body, sub_bit_len be 1 when, be the initial data of respective compression units, and
It is then in order by each difference by fixed bit wide, and with " sign bit+difference of difference when sub_bit_len is other values
Absolute value " combination piece together.Wherein, the sign bit of difference is 0, indicates that the difference is positive number;The sign bit of difference is 1, is indicated
The difference is negative.
Mode0:len0=2+4+16+7* (log2(abs_m0[0])+1);
+4+8*(log2(abs_m0[1])+1);
+4+8*(log2(abs_m0[2])+1);
+4+8*(log2(abs_m0[3])+1);
+4+8*(log2(abs_m0[4])+1);
+4+8*(log2(abs_m0[5])+1);
+4+8*(log2(abs_m0[6])+1);
+4+8*(log2(abs_m0[7])+1);
In first formula: in 2+4+16, " 2 " indicate the 2bit of model_sel, and " 4 " indicate the 4bit of sub_bit_len,
" 16 " indicate the 16bit of orgA0, and 7* (log2 (abs_m0 [0])+1) parsing is as follows: " 7 " in formula are that the first compression unit is wanted
First initial data is deducted, mono- compression unit of mode0 is 8 data, deducts first initial data, is left 7 (below
Several formulas in corresponding position be " 8 "), " log2 (abs_m0 [0]) " in formula refers to that the first compression unit is maximum in mode0
The number of significant digit of absolute value, and " (log2 (abs_m0 [0])+1) ": in "+1 " be 1 potential difference value sign bit, each difference
Sign bit is also intended to be placed in compressed bit stream, and decompression Shi Caineng in this way restores initial value.
And so on, total bit number that all compact models finally obtain can be calculated:
Mode1:len1=2+4+16+15* (log2(abs_m1[0])+1)
+4+16*(log2(abs_m1[1])+1)
+4+16*(log2(abs_m1[2])+1)
+4+16*(log2(abs_m1[3])+1);
Mode2:len2=2+4+16+31* (log2(abs_m2[0])+1)
+4+32*(log2(abs_m2[1])+1);
Mode3:len3=2+4+16+63* (log2(abs_m3[0])+1);
Step5: comparing the size for total bit number (len0, len1, len2, len3) that all compact models finally obtain, choosing
It is the smallest in this 4 compact models out to be worth corresponding mode as the last compact model of compression section.
Assuming that above example chooses is compact model mode0, then compressed bit stream is as follows:
{2’d0,4’d5,16’d1100,{1’d1,4’d10},{1’d1,4’d8},{1’d1,4’d1}, {1’d0,4’
d4},{1’d1,4’d6},{1’d0,4’d3},{1’d0,4’d2},4’d4,{1’d0,3’d2}, {1’d0,3’d3},{1’d0,
3’d1},{1’d1,3’d5},{1’d1,3’d5},{1’d0,3’d3},{1’d0,3’d5}, {1’d1,3’d2},...}
If, can be also bigger than not compressing after certain compression unit compressions, then individually this compression unit just replicates initial value.Assuming that
Log2 (abs_m0 [2])+1 >=16, that illustrates the compression unit of this 8 pixel, and adjacent data difference is too big,
Such as: 0,65535,0,1,2,3,4,5
=> difference are as follows: 0,65535, -65535,1,1,1,1,1
=> absolute value: 0,65535,65535,1,1,1,1,1
Log2 (abs_m0 [2])+1=17bit, it is also bigger than not compressing instead, in this case, just by sub_bit_len
It is assigned a value of 1.
Wherein Step2 is the difference for seeking adjacent two data, if data source OrgANCorrelation it is bigger, i.e., it is adjacent
OrgANNumerical value relatively when, pass through formula diffAN=OrgAN-OrgAN-1, obtained diffANAbsolute value can also compare
Small, to show that corresponding sub_bit_len also can be smaller, total bit number that compression section comes out also can be fewer, compression
Effect is with regard to more preferable.
Based on the above method of the present invention, the present invention also provides one kind to be based on the decoded colmv lossless compression system of inter,
As shown in Fig. 2, including encoder, first temporary storage mem0, DDR, the second temporary storage mem1 and decoder;It is described
Encoder is by colmv data to be compressed by the compressed code stream storage of the above method of the present invention to the first temporary storage
In, it is then written in DDR after reaching certain condition.The decoder is banished from DDR sense code again when decompressing to be faced described second
When memory in, then be decoded.
As shown in figure 3, the internal module of the encoder includes asking difference block, temporary storage mem, total bit number system
Count module and displacement buf.Specific execution process is handled as shown in figure 4, data source (date_src) is first segmented (segment),
Assuming that each segment sections takes 64 data, data source is first asked to difference (adjacent two data is subtracted each other), and difference is saved
Into temporary storage, then the maximum value in each compression unit difference is found out, then finds out total bit of 4 kinds of modes respectively
It counts (len0, len1, len2, len3), after entire compression section, compares total bit number size of four kinds of modes, select total bit
Number recklings corresponding to mode (assuming that len0 be it is the smallest in four, then illustrate selection mode be mode0, be the mould of 8*8
Formula), and by the difference being stored in temporary storage reading be sequentially read out, then it is seamless spliced (i.e. by data without compartment of terrain by
One pieces together output) (stream_out) is exported at code stream.
As shown in figure 5, the internal module of the decoder includes 256bit displacement buf, state machine control module, the solution
When code device decompression, as shown in connection with fig. 6,256bit displacement buf reads in the data stream_ being stored in temporary storage in order
In, refers again to following table, state machine control module parse each compression section compact model mode_sel (4 kinds of modes,
" 0 ": mode 0;" 1 ": mode 1;" 2 ": mode 2;" 3 ": mode 3), then parse the subsegment code length sub_bit_ of compression unit
Len parses first initial data (bit wide of first initial data is fixed and invariable) of compression section, then basis
Subsegment code length sub_bit_len parses compressed bit stream main body bit_stream, obtain the corresponding sign bit of each difference and absolutely
To value, then initial value output (org_data_out) is calculated, and when the long sub_bit_len=1 of group segment encode, then it represents that current pressure
Contracting unit is not compressed, and in code stream is raw value;
Present invention employs the difference sectional compression methods of lossless compression algorithm, and the characteristics of combination colmv data, can be more
It is further to improve compression ratio.When the correlation of image data is bigger (i.e. adjacent data numerical value relatively), the difference come is found out
Value will be smaller, and compression effectiveness will be better.
Because colmv data the characteristics of be that correlation is very big, so be well suited for using this scheme of the present invention, after compression
To code stream be mostly largely to alleviate the pressure that memory reads bandwidth within original 30%.As shown in fig. 7,
Horizontal axis in Fig. 7 is the frame number run, and the longitudinal axis is to indicate colmv number before compressing with former data source bit percentage, the straight line of series 1
According to source, the broken line of series 2 indicates compressed code stream accounting.
Although specific embodiments of the present invention have been described above, those familiar with the art should be managed
Solution, we are merely exemplary described specific embodiment, rather than for the restriction to the scope of the present invention, it is familiar with this
The technical staff in field should be covered of the invention according to modification and variation equivalent made by spirit of the invention
In scope of the claimed protection.
Claims (6)
1. one kind is based on the decoded colmv data lossless compression method of inter, it is characterised in that: include:
Step S11, in colmv data to be compressed, by every 2nCompression section of a data as an independent compression, each pressure
Contracting section is divided into compression unit one by one according to M kind compact model respectively, wherein n is natural number, and M is by user's sets itself;
Step S12, it keeps the first data in compression section constant, then successively finds out the difference of adjacent data, then find out each compression
Maximum value in unit difference;
Step S13, use the digit+1 of maximum value as the subsegment code length of the compression unit, further according to this subsegment code length handle
The difference of subsequent compression unit is set out one by one to be come;
Step S14, total bit number of the various compact models of each compression unit is found out respectively;
Step S15, the size for comparing total bit number that all compact models finally obtain, selects the smallest corresponding pressure of total bit number
Compressed mode as corresponding compression section last compact model, if certain compression sections upon compression than not compressing also big when, singly
This only compression unit replicates initial value.
2. according to claim 1 be based on the decoded colmv data lossless compression method of inter, it is characterised in that: when 2n
Value be 64 and M=4 when, then 4 kinds of modes are respectively as follows:
Mode0: point 8 compression units, 8 data of each compression unit;
Mode1: point 4 compression units, 16 data of each compression unit;
Mode2: point 2 compression units, 32 data of each compression unit;
Mode3: point 1 compression unit, 64 data of this compression unit.
3. according to claim 2 be based on the decoded colmv data lossless compression method of inter, it is characterised in that: described
In step S13, the virtual value of the subsegment code length is 0 and integer 2~15;
Group segment encode it is a length of 1 when, compressed bit stream main body is the initial data of respective compression units;
When a length of integer 2~15 of group segment encode, compressed bit stream main body be then in order by each difference by fixed bit wide, and with
" sign bit+difference absolute value of difference " combination is pieced together, and the sign bit of difference is 0, indicates that the difference is positive number;The symbol of difference
Number position is 1, indicates that the difference is negative.
4. according to claim 3 be based on the decoded colmv data lossless compression method of inter, it is characterised in that: described
In step S13, total bit number is the bit wide of " sign bit+difference absolute value of difference " described under a certain compact model
The summation of bit number.
5. one kind is based on the decoded colmv lossless date-compress system of inter, it is characterised in that: temporarily including encoder, first
Memory, DDR, the second temporary storage and decoder;
The encoder is by colmv data to be compressed by the compressed code stream storage of any one of Claims 1-4 method to the
In one temporary storage, it is then written in DDR after reaching certain condition;
The decoder is banished in second temporary storage from DDR sense code again when decompressing, then is decoded.
6. according to claim 5 be based on the decoded colmv lossless date-compress system of inter, it is characterised in that: described
When decoder decompresses, it is to be sequentially read out the data being stored in temporary storage, refers again to following table, parse each
The compact model mode_sel of compression section, then the subsegment code length sub_bit_len of compression unit is parsed, parse compression section
Then first initial data parses compressed bit stream main body bit_stream according to subsegment code length sub_bit_len, obtain each
The corresponding sign bit of a difference and absolute value, then calculate initial value, and when the long sub_bit_len=1 of group segment encode, then it represents that when
Preceding compression unit is not compressed, and in code stream is raw value;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811092181.XA CN109660809A (en) | 2018-09-19 | 2018-09-19 | Based on the decoded colmv data lossless compression method of inter and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811092181.XA CN109660809A (en) | 2018-09-19 | 2018-09-19 | Based on the decoded colmv data lossless compression method of inter and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109660809A true CN109660809A (en) | 2019-04-19 |
Family
ID=66110275
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811092181.XA Pending CN109660809A (en) | 2018-09-19 | 2018-09-19 | Based on the decoded colmv data lossless compression method of inter and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109660809A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111159121A (en) * | 2019-12-23 | 2020-05-15 | 深圳星康医疗科技有限公司 | Wearable electrocardiogram data compression storage method, system and storage system |
CN112019220A (en) * | 2020-08-21 | 2020-12-01 | 广东省新一代通信与网络创新研究院 | Block floating point data compression method and device based on differential bias detection |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070065026A1 (en) * | 2005-09-16 | 2007-03-22 | Industry-Academia Cooperation Group Of Sejong University | Method of and apparatus for lossless video encoding and decoding |
US20080089595A1 (en) * | 2006-10-13 | 2008-04-17 | Samsung Electronics Co.,Ltd. | Method of and apparatus for encoding/decoding data |
CN101841705A (en) * | 2010-03-12 | 2010-09-22 | 西安电子科技大学 | Video lossless compression method based on adaptive template |
CN104486624A (en) * | 2014-12-18 | 2015-04-01 | 中国科学院自动化研究所 | CAVLC entropy decoder and entropy decoding method |
CN107197286A (en) * | 2017-05-15 | 2017-09-22 | 华中科技大学 | A kind of Lossless Image Compression Algorithm method and system based on reversible long code |
-
2018
- 2018-09-19 CN CN201811092181.XA patent/CN109660809A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070065026A1 (en) * | 2005-09-16 | 2007-03-22 | Industry-Academia Cooperation Group Of Sejong University | Method of and apparatus for lossless video encoding and decoding |
US20080089595A1 (en) * | 2006-10-13 | 2008-04-17 | Samsung Electronics Co.,Ltd. | Method of and apparatus for encoding/decoding data |
CN101841705A (en) * | 2010-03-12 | 2010-09-22 | 西安电子科技大学 | Video lossless compression method based on adaptive template |
CN104486624A (en) * | 2014-12-18 | 2015-04-01 | 中国科学院自动化研究所 | CAVLC entropy decoder and entropy decoding method |
CN107197286A (en) * | 2017-05-15 | 2017-09-22 | 华中科技大学 | A kind of Lossless Image Compression Algorithm method and system based on reversible long code |
Non-Patent Citations (1)
Title |
---|
陈秋华: "一种内存读写数据压缩算法", 《中国集成电路》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111159121A (en) * | 2019-12-23 | 2020-05-15 | 深圳星康医疗科技有限公司 | Wearable electrocardiogram data compression storage method, system and storage system |
CN112019220A (en) * | 2020-08-21 | 2020-12-01 | 广东省新一代通信与网络创新研究院 | Block floating point data compression method and device based on differential bias detection |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8942502B2 (en) | Parallelization of variable length decoding | |
CN101588502B (en) | DCT compression using Golomb-Rice coding | |
US20020106019A1 (en) | Method and apparatus for implementing motion detection in video compression | |
CN103067022B (en) | A kind of integer data lossless compression method, decompression method and device | |
US8902992B2 (en) | Decoder for selectively decoding predetermined data units from a coded bit stream | |
US7365659B1 (en) | Method of context adaptive binary arithmetic coding and coding apparatus using the same | |
JPH07283739A (en) | Method and device to compress and extend data of short block | |
US20190370667A1 (en) | Lossless compression of sparse activation maps of neural networks | |
CN101197578A (en) | Data-modifying run length encoder to avoid data expansion | |
US7298297B1 (en) | Hardware-implemented Huffman decoder | |
WO2021031877A1 (en) | Methods and apparatus for image coding and decoding, and chip | |
CN113868206A (en) | Data compression method, decompression method, device and storage medium | |
RU2313174C2 (en) | Adaptive method and system for transforming values of parameters into indexes of code words | |
CN109660809A (en) | Based on the decoded colmv data lossless compression method of inter and system | |
US7148821B2 (en) | System and method for partition and pattern-match decoding of variable length codes | |
CN103716634A (en) | Method and apparatus for data compression using error plane coding | |
US8305244B2 (en) | Coding data using different coding alphabets | |
KR101023536B1 (en) | Lossless data compression method | |
US7880647B1 (en) | Huffman decoding method | |
CN100399833C (en) | High-speed image compression apparatus using last non-zero detection circuit | |
US7683809B2 (en) | Advanced lossless bit coding | |
US6433707B1 (en) | Universal lossless compressor for digitized analog data | |
WO2024007843A9 (en) | Encoding method and apparatus, decoding method and apparatus, and computer device | |
KR100686354B1 (en) | Huffman decoding method and device for using variable length tree | |
KR101174598B1 (en) | Apparatus for generating compression file using lossless data compression algorism and method thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190419 |
|
WD01 | Invention patent application deemed withdrawn after publication |