CN104065974B - Method for compressing image and system - Google Patents
Method for compressing image and system Download PDFInfo
- Publication number
- CN104065974B CN104065974B CN201410306690.3A CN201410306690A CN104065974B CN 104065974 B CN104065974 B CN 104065974B CN 201410306690 A CN201410306690 A CN 201410306690A CN 104065974 B CN104065974 B CN 104065974B
- Authority
- CN
- China
- Prior art keywords
- bit
- code
- plane encoding
- result
- image
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
Abstract
The present invention proposes a kind of method for compressing image, comprises the following steps:Rarefaction representation is carried out to image, the rarefaction representation result of each picture frame is obtained;Bit-Plane Encoding is carried out to rarefaction representation result, to obtain Bit-Plane Encoding result;Bitplanes coding result carries out entropy code, to obtain entropy code result, and entropy code result is carried out code stream organization packing obtain final compressed bit stream.The method of the present invention, encoder complexity is low, efficiency high, fault-tolerant ability are strong, compression performance is high.The present invention also proposes a kind of image compression system.
Description
Technical field
The present invention relates to Image Compression field, more particularly to a kind of method for compressing image and system.
Background technology
Full-colour image (spectral coverage is generally in the range of 450nm~900nm) is that atural object is pushed away by panchromatic TDICCD cameras on star
Sweep the 2-D data with spatial information that imaging is obtained.The data can provide abundant ground object detail, be widely used in
The fields such as resource exploration, military surveillance and environmental protection.Spatial resolution, radiation resolution with the panchromatic TDICCD cameras in space
The indexs such as rate, temporal resolution, big visual field, covering wide are improved constantly, and the TDICCD for causing panchromatic TDICCD cameras to use splices
Piece number and read-out speed are also on the increase and are improved, average camera time increase, so that the image data amount after digitlization is big
Width increases.Existing spaceborne memory span is limited, satellite channel Bandwidth-Constrained, it is impossible to adapt to the magnanimity number of full-colour image on star
According to.It then becomes necessary to be compressed to full-colour image on star.
Full-colour image data have two kinds of redundancies on star:Meet redundancy between space between redundancy and data.Therefore, full-colour image
The purpose of compression is exactly to eliminate both redundancies.At present, many methods using wavelet transformation is based on of full-colour image compression on star, such as
The image compression algorithm for being used on BilSAT-1 (SSTL-Turkey) satellite for 2003 is JPEG2000 algorithms.JPEG2000
Algorithm uses the space decorrelation method of DWT (Discrete Wavelet Transform), and algorithm realizes that platform is FPGA+
DSP(XCV300E+TMS320C6701)., the Image Data Compression work of consultative committee for space data system (CCSDS) in 2005
Make the Standard of image compression CCSDS122.0-B-1 that group (IDC) has formulated space application of new generation, the algorithm is also adopted by small echo change
Change.However, wavelet transformation is for full-colour image on the informative star of texture level such as edge and profile, be not it is a kind of most
Excellent sparse expression, can produce a large amount of significantly high frequency coefficients, be unfavorable for follow-up sub-band coding so that the pressure of compression algorithm
Contracting performance is relatively low.
The content of the invention
It is contemplated that at least solving one of technical problem in correlation technique to a certain extent.
Therefore, first purpose of the invention is to propose that a kind of low encoder complexity, efficiency high, fault-tolerant ability are strong, press
Contracting performance method for compressing image high.
Second object of the present invention is to propose a kind of image compression system.
To achieve these goals, the method for compressing image of first aspect present invention embodiment, comprises the following steps:To figure
As carrying out rarefaction representation, the rarefaction representation result of each picture frame is obtained;Bit plane volume is carried out to the rarefaction representation result
Code, to obtain Bit-Plane Encoding result;Entropy code is carried out to the Bit-Plane Encoding result, it is to obtain entropy code result and right
The entropy code result carries out code stream organization packing and obtains final compressed bit stream.
Method for compressing image according to embodiments of the present invention, using the carrying out converted after a kind of single base dictionary of low complex degree
Image sparse is represented, and carries out dynamic code rate distribution using each section in bit rate controller bitplanes coding, and encoder is multiple
Miscellaneous degree is low, and code efficiency is high, and compression performance is high.
In some instances, it is described rarefaction representation is carried out to image to specifically include:3 grades of two dimensions are carried out to each picture frame
9/7 wavelet transform, obtains the low frequency sub-band and high-frequency sub-band of respective image frame;Single base dictionary is carried out to the high-frequency sub-band
After convert, to obtain AC coefficients and side information.
In some instances, selection is converted after single base dictionary and uses different evaluation functions, its specifically chosen process includes:
The byte number of compression frame overhead is calculated according to compression ratio parameter;Code check is calculated according to the byte number, when the code check is code high
During rate, conversion afterwards uses l-1Norm Method, when the code check is low bit- rate, conversion afterwards uses l-0Norm Method.
In some instances, the Bit-Plane Encoding is specifically included:DC coefficients are obtained, and the DC coefficients are carried out initially
Change;Extract paragraph header information and depth coding is carried out to the AC coefficient bits;Extract AC coefficients and residue DC coefficients enter line position and put down
Face code;The side information is embedded into the code stream of Bit-Plane Encoding, Bit-Plane Encoding result is obtained.
In some instances, when the code stream of Bit-Plane Encoding reaches predetermined threshold value, then Bit-Plane Encoding is stopped.
In some instances, using the control based on the intermediate quantity and code check dynamically distributes converted after single base dictionary
The code stream of the Bit-Plane Encoding.
In some instances, coding is also predicted to the low frequency sub-band, and entropy code is carried out to prediction residual.
The image compression system of second aspect present invention embodiment, including:Image sparse representation module, for entering to image
Row rarefaction representation, obtains the rarefaction representation result of each picture frame;Bit-Plane Encoding module, for the rarefaction representation knot
Fruit carries out Bit-Plane Encoding, to obtain Bit-Plane Encoding result;Entropy code module, for being carried out to the Bit-Plane Encoding result
Entropy code, to obtain entropy code result, and the entropy code result is carried out code stream organization packing obtain final compressed bit stream.
Image compression system according to embodiments of the present invention, using the carrying out converted after a kind of single base dictionary of low complex degree
Image sparse is represented, and carries out dynamic code rate distribution using each section in bit rate controller bitplanes coding, and encoder is multiple
Miscellaneous degree is low, and code efficiency is high, and compression performance is high.
In some instances, described image rarefaction representation module specifically performs sparse table of the following steps realization to image
Show:3 grades of wavelet transforms of two dimension 9/7 are carried out to each picture frame, low frequency sub-band and high frequency of respective image frame is obtained
Band;Converted after single base dictionary is carried out to the high-frequency sub-band, to obtain AC coefficients and side information.
In some instances, selection is converted after single base dictionary and uses different evaluation functions, its specifically chosen process
Including:The byte number of compression frame overhead is calculated according to compression ratio parameter;Code check is calculated according to the byte number, when the code check is
During code check high, conversion afterwards uses l-1Norm Method, when the code check is low bit- rate, conversion afterwards uses l-0Norm Method.
In some instances, the specific following steps that perform realize Bit-Plane Encoding in the Bit-Plane Encoding module:Obtain
DC coefficients, and the DC coefficients are initialized;Extract paragraph header information and depth coding is carried out to the AC coefficient bits;Carry
Taking AC coefficients and residue DC coefficients carries out Bit-Plane Encoding;The side information is embedded into the code stream of Bit-Plane Encoding, is obtained
Bit-Plane Encoding result.
In some instances, the entropy code module is additionally operable to be predicted the low frequency sub-band coding, and to prediction
Residual error carries out entropy code.
In some instances, also include:Rate control module, for realizing converting the different evaluations of selection after single base dictionary
Function, the dynamic code stream of distribution Bit-Plane Encoding, control the predetermined threshold value of the code stream to stop Bit-Plane Encoding and to entropy
The code check of coding is controlled.
The additional aspect of the present invention and advantage will be set forth in part in the description, and will partly become from the following description
Obtain substantially, or recognized by practice of the invention.
Brief description of the drawings
Fig. 1 is the flow chart of method for compressing image according to an embodiment of the invention;
Fig. 2 is the process schematic of the method for compressing image of one embodiment of the invention;
Fig. 3 is the structural representation of image compression system in accordance with another embodiment of the present invention;With
Fig. 4 is the hardware architecture diagram of the image compression system of one embodiment of the invention.
Specific embodiment
Embodiments of the invention are described below in detail, the example of the embodiment is shown in the drawings, wherein from start to finish
Same or similar label represents same or similar element or the element with same or like function.Below with reference to attached
It is exemplary to scheme the embodiment of description, it is intended to for explaining the present invention, and be not considered as limiting the invention.
It is the technical problems to be solved by the invention for the compression method of full-colour image on star provides new technological means, is
This, proposes a kind of method for compressing image of low complex degree in the embodiment of the first aspect of the present invention, comprise the following steps:It is right
Image carries out rarefaction representation, obtains the rarefaction representation result of each picture frame;Bit-Plane Encoding is carried out to rarefaction representation result,
To obtain Bit-Plane Encoding result;Bitplanes coding result carries out entropy code, to obtain entropy code result, and to entropy code knot
Fruit carries out code stream organization packing and obtains final compressed bit stream.
Fig. 1 is the flow chart of method for compressing image according to an embodiment of the invention.Fig. 2 is one embodiment of the invention
Method for compressing image process schematic.The method for compressing image of the embodiment of the present invention is specifically described with reference to Fig. 1 and Fig. 2.
By panchromatic CCD with the image configuration of behavior unit into being unit image with frame, X is designated asi(i=1,2,3 ..., N).
Step S101:Rarefaction representation is carried out to image, the rarefaction representation result of each picture frame is obtained.
Specifically, carrying out rarefaction representation to image includes:
(1) 3 grades of wavelet transforms of two dimension 9/7 are carried out to each picture frame, the low frequency sub-band of respective image frame is obtained
And high-frequency sub-band.That is present frame XiUsing 3 grade of 9/7 lifting wavelet transform, 1 low frequency sub-band LL and 9 high-frequency sub-bands are obtained
HL1, HL2, HL3, LH1, LH2, LH3, HH1, HH2, HH3.Each high-frequency sub-band HL3, LH3, HH3 and HL1, LH1 are small in HH1
Wave system number is organized into some pieces with 4*4 sizes, is designated as Ai(i=1,2,3 ..., J).Each high-frequency sub-band HL2, LH2 is small in HH2
Wave system number is organized into some pieces with 2*2 sizes, is designated as Bi(i=1,2,3 ..., K).
(2) converted after single base dictionary is carried out to high-frequency sub-band, to obtain AC coefficients and side information.
Specifically, in an embodiment of the present invention, each block AiAnd BiPosition it is constant, and fw will be designated as by transform block after
(w=1,2,3 ..., W).Converted after the mono- base dictionaries of Hadamard are carried out to each block, obtain rear conversion coefficient and side information.
The base dictionary only one of which Hadamard bases used when converting afterwards, Hadamard Base computings only have addition and shift operation, calculate
Very simple, and it is easy to hardware to realize, and with compression performance higher.
First to f1Converted after carrying out, by coefficient block f1Converted after performing single base, rear transformation calculations formula is:
Wherein, G is rear transform block size,It is rear conversion base vector, a [g] conversion coefficients afterwards.
Secondly, to coefficient block f1It is optimal after conversion selection, selection use different evaluation functions, the tool of its evaluation function
Body selection course is:
(1) byte number of compression frame overhead is calculated according to compression ratio parameter.
The compression ratio parameter that total bitrate computing unit injects according to camera controller calculates the byte number of compression frame overhead, word
Joint number is:Wherein, M and N are respectively image frame sign, i.e. M for the effective pixel numbers of CCD, and N is
CCD line numbers, L is each bit depths of pixels, and R is compression ratio.
(2) code check is calculated according to condensed frame overhead byte number, calculating formula is:R=8 × BC/(M×N).When code check is code high
During rate (such as r >=1bpp), conversion afterwards uses l-1Norm Method;When code check is low bit- rate (such as r < 1bpp), conversion afterwards is used
l-0Norm Method.
Therefore, optimal rear conversion selection uses l-pThe method of norm, when for low bit- rate, p=0, i.e., using l-0Norm
Method, i.e., calculated using following formula:
Work as p=1, i.e., using l-1Norm method, is calculated using following formula:
Wherein, NbIt is the number of base in dictionary, due to using single base dictionary, therefore Nb=1.Work as NbWhen=0,I.e.
Converted after not carrying out now, it is optimal after be transformed to original sparse block.Used not according to different code checks during conversion selection after optimal
Same evaluation function, so substantially increases computational efficiency, reduces computation complexity.
Above-mentioned steps S101 is repeated, until present frame XiAll high-frequency sub-bands after the completion of convert.So, to small echo
Coefficient in domain is converted again, can further remove the redundancy between wavelet coefficient, makes up full-colour image on star
The deficiency of small echo rarefaction representation.
Step S102:Bit-Plane Encoding is carried out to rarefaction representation result, to obtain Bit-Plane Encoding result.
Specifically, in one embodiment of the invention, the rear conversion coefficient tissue that above-mentioned steps S101 is obtained is turned into
Some sections, BPE codings are carried out in units of section.Each section is made up of 16 blocks, every piece by 1 DC coefficient (i.e. low frequency sub-band) and
63 AC coefficients (i.e. the rear conversion coefficient of high-frequency sub-band) compositions.Each section is designated as St(i=1,2,3 ..., T).
Section S1 is encoded, i.e.,:DC coefficients are obtained, and DC coefficients are initialized;Extract paragraph header information and to AC
Coefficient bit carries out depth coding;Extracting AC coefficients and residue DC coefficients carries out Bit-Plane Encoding;The side that step S101 is obtained
Information is embedded into the code stream of Bit-Plane Encoding, obtains escape word, i.e. section S1 Bit-Plane Encodings result.So not only hardware is realized
Simply, and with compression performance higher, further, it enables coding has gradual feature, improve the robust of compression algorithm
Property.
Especially, when the code stream of Bit-Plane Encoding reaches predetermined threshold value, then Bit-Plane Encoding is stopped.Specific code check point
Process with controller bite rate control is as follows:
(1) segment information is estimated and weight calculation.Section StIn gross information content be
The intermediate result of rear transformation calculations in step S101 is dexterously make use of during the estimation of the information content of section, meter is improve
Calculate efficiency.
(2) code check shared by each section is calculated.
Specifically, the summation of all segment information amounts is calculated, and calculates the ratio shared by each segment information.If i-th section
Gross information content is Ii, and R section is contained per two field picture.Then the information scales shared by every section are
If compression of images code check is RS, then each section of code check of distribution is ri,s=ri×RS.Distribution in final coding section i
Target word joint number be:PS=M × N × RS/B.Wherein, M, N are the size of image.B is image locating depth.Using according to each section
Code check is dynamically distributed containing different information content, the strategy of dividing equally for overcoming code check section causes that compression performance is low to ask
Topic.Rate Control is carried out not only simply but also with compression performance higher using control bit plane in each section, and is realized
It is more convenient.
When the code stream of Bit-Plane Encoding reaches predetermined threshold value, then application bit plane depth encoder control stops bit plane
Coding.
Repeat the above steps S102, until completing all sections of coding.
At the same time, coding is predicted to the low frequency sub-band that step S101 is obtained, is compiled with obtaining low frequency sub-band prediction
The prediction residual of code.
Step S103:Bitplanes coding result carries out entropy code, to obtain entropy code result, and entropy code result is entered
The packing of row code stream organization obtains final compressed bit stream.
The Bit-Plane Encoding result and the prediction residual of low frequency sub-band predictive coding that step S102 is obtained carry out entropy volume
Code, obtain entropy code result, and entropy code result is carried out code stream organization packing obtain final compressed bit stream.
Above-mentioned steps S101~S103 is repeated, until all of picture frame end-of-encode.
Method for compressing image according to embodiments of the present invention, using the carrying out converted after a kind of single base dictionary of low complex degree
Image sparse is represented, and carries out dynamic code rate distribution using each section in bit rate controller bitplanes coding, and encoder is multiple
Miscellaneous degree is low, and code efficiency is high, and compression performance is high.
A kind of image compression system is proposed in the embodiment of second aspect present invention, as shown in figure 3, including:Image sparse
Representation module 100, Bit-Plane Encoding module 200 and entropy code module 400.
Wherein, image sparse representation module 100, for carrying out rarefaction representation to image, obtains the dilute of each picture frame
Dredge and represent result.Bit-Plane Encoding module 200, for carrying out Bit-Plane Encoding to rarefaction representation result, to obtain bit plane volume
Code result.Entropy code module 300, entropy code is carried out for bitplanes coding result, to obtain entropy code result, and entropy is compiled
Code result carries out code stream organization packing and obtains final compressed bit stream.
Further, the image compression system of embodiments of the invention, also includes:Rate control module 400, for realizing
The different evaluation function of selection is converted after single base dictionary, the dynamic code stream of Bit-Plane Encoding and the default threshold of control stream is distributed
Value is stopping Bit-Plane Encoding.
Specifically, in image sparse representation module 100, by panchromatic CCD with the buffered unit structure of the image of behavior unit
It is unit image to cause with frame, is designated as Xi(i=1,2,3 ..., N).In one embodiment of the invention, buffer unit is used
The SDRAM of ping-pong operation is realized.
Rarefaction representation is carried out to image in image sparse representation module 100, the rarefaction representation of each picture frame is obtained
As a result.
Specifically, carrying out rarefaction representation to image includes:
(1) 3 grades of wavelet transforms of two dimension 9/7 are carried out to each picture frame, the low frequency sub-band of respective image frame is obtained
And high-frequency sub-band.That is present frame XiUsing 3 grade of 9/7 lifting wavelet transform, 1 low frequency sub-band LL and 9 high-frequency sub-bands are obtained
HL1, HL2, HL3, LH1, LH2, LH3, HH1, HH2, HH3.Each high-frequency sub-band HL3, LH3, HH3 and HL1, LH1 are small in HH1
Wave system number is organized into some pieces with 4*4 sizes, is designated as Ai(i=1,2,3 ..., J).Each high-frequency sub-band HL2, LH2 is small in HH2
Wave system number is organized into some pieces with 2*2 sizes, is designated as Bi(i=1,2,3 ..., K).
(2) converted after single base dictionary is carried out to high-frequency sub-band, to obtain AC coefficients and side information.
Specifically, in an embodiment of the present invention, each block AiAnd BiPosition it is constant, and fw will be designated as by transform block after
(w=1,2,3 ..., W).Converted after the mono- base dictionaries of Hadamard are carried out to each block, obtain rear conversion coefficient and side information.
The base dictionary only one of which Hadamard bases used when converting afterwards, Hadamard Base computings only have addition and shift operation, calculate
Very simple, and it is easy to hardware to realize, and with compression performance higher.
First to f1Converted after carrying out, by coefficient block f1Converted after performing single base, rear transformation calculations formula is:
Wherein, G is rear transform block size,It is rear conversion base vector, a [g] conversion coefficients afterwards.
Secondly, to coefficient block f1It is optimal after conversion selection, by evaluation function arbitration unit selection use different evaluations
Function, the specifically chosen process of its evaluation function is:
(1) byte number of compression frame overhead is calculated according to compression ratio parameter.
The compression ratio parameter that total bitrate computing unit injects according to camera controller calculates the byte number of compression frame overhead, word
Joint number is:Wherein, M and N are respectively image frame sign, i.e. M for the effective pixel numbers of CCD, and N is CCD
Line number, L is each bit depths of pixels, and R is compression ratio.
(2) code check is calculated according to condensed frame overhead byte number, calculating formula is:R=8 × BC/(M×N).When code check is code high
During rate (such as r >=1bpp), conversion afterwards uses l-1Norm Method;When code check is low bit- rate (such as r < 1bpp), conversion afterwards is used
l-0Norm Method.
Therefore, optimal rear conversion selection uses l-pThe method of norm, when for low bit- rate, p=0, i.e., using l-0Norm
Method, i.e., calculated using following formula:
Work as p=1, i.e., using l-1Norm method, is calculated using following formula:
Wherein, NbIt is the number of base in dictionary, due to using single base dictionary, therefore Nb=1.Work as NbWhen=0,I.e.
Converted after not carrying out now, it is optimal after be transformed to original sparse block.Used not according to different code checks during conversion selection after optimal
Same evaluation function, so substantially increases computational efficiency, reduces computation complexity.
Above-mentioned steps are repeated, until present frame XiAll high-frequency sub-bands after the completion of convert.So, in wavelet field
Coefficient converted again, can further remove the redundancy between wavelet coefficient, make up full-colour image small echo on star
The deficiency of rarefaction representation.
In Bit-Plane Encoding module 200, Bit-Plane Encoding is carried out to rarefaction representation result, to obtain Bit-Plane Encoding knot
Really.
Specifically, in one embodiment of the invention, the rear conversion that will be obtained in above-mentioned image sparse representation module 100
Coefficient is equally organized into tree construction according to wavelet sub-band, then turns into some sections through CCSDS-IDC bit plane encoders tissue, with
Section carries out BPE codings for unit.Each section is made up of 16 blocks, and every piece by 1 DC coefficient (i.e. low frequency sub-band) and 63 AC systems
Number (i.e. the rear conversion coefficient of high-frequency sub-band) composition.Each section is designated as St(i=1,2,3 ..., T).
Section S1 is encoded, i.e.,:DC coefficients are obtained, and DC coefficients are initialized;Extract paragraph header information and to AC
Coefficient bit carries out depth coding;Extracting AC coefficients and residue DC coefficients carries out Bit-Plane Encoding;The side that step S101 is obtained
Information is embedded into the code stream of Bit-Plane Encoding, obtains escape word, i.e. section S1 Bit-Plane Encodings result.So not only hardware is realized
Simply, while having compression performance higher, further, it enables coding has gradual feature, improve the robust of compression algorithm
Property.
Especially, when the code stream of Bit-Plane Encoding reaches predetermined threshold value, then Bit-Plane Encoding is stopped.Specific code stream control
The process of system is as follows:
(1) segment information is estimated and weight calculation.Section StIn gross information content be
The middle knot of rear transformation calculations in image sparse representation module 100 is dexterously make use of during the estimation of the information content of section
Really, improve computational efficiency.
(2) code check shared by each section is calculated.
Specifically, the summation of all segment information amounts is calculated, and calculates the ratio shared by each segment information.If i-th section
Gross information content is Ii, and R section is contained per two field picture.Then the information scales shared by every section are
If compression of images code check is RS, then each section of code check of distribution is ri,s=ri×RS.Distribution in final coding section i
Target word joint number be:PS=M × N × RS/B.Wherein, M, N are the size of image.B is image locating depth.Using according to each section
Code check is dynamically distributed containing different information content, the strategy of dividing equally for overcoming code check section causes that compression performance is low to ask
Topic.Rate Control is carried out not only simply but also with compression performance higher using control bit plane in each section, and is realized
It is more convenient.
When the code stream of Bit-Plane Encoding reaches predetermined threshold value, then application bit plane depth encoder control stops bit plane
Coding.
Repeat the above steps, until completing all sections of coding.
At the same time, coding is predicted to the low frequency sub-band that image sparse representation module 100 is obtained, to obtain the low frequency
The prediction residual of sub-band predictive coding.
In entropy code module 300, bitplanes coding result carries out entropy code, to obtain entropy code result, and to entropy
Coding result carries out code stream organization packing and obtains final compressed bit stream.The bit plane that will be obtained in Bit-Plane Encoding module 200
Coding result and the prediction residual of low frequency sub-band predictive coding carry out entropy code, obtain entropy code result, and to entropy code knot
Fruit carries out code stream organization packing and obtains final compressed bit stream.
By the evaluation function arbitration unit and Bit-Plane Encoding module 200 that are used in above-mentioned image sparse representation module 100
The rate control module 400 that the bit rate controller and Bit-Plane Encoding depth controller for using are constituted, realizes single base dictionary
The different evaluation function of conversion selection, distributes the dynamic code stream of Bit-Plane Encoding and the predetermined threshold value of control stream to stop afterwards
Bit-Plane Encoding.
The hardware configuration of the image compression system of the one embodiment of the invention realized by hardware processors such as FPGA, DSP
As shown in Figure 4.The hardware composition of the image compression system mainly includes:Full-colour image data sink, FPGA processor, NAND
Flash array, SDRAM cachings, dsp processor, program storage and data storage.
Full-colour image receiver is used for receiving the ccd image of CCD imaging units output and being input in FPGA processor.
FPGA processor realizes whole compression work.Wherein, FPGA processor includes:Data receiver controller, storage management states machine,
SDRAM read-write controllers, inner buffer compression host state machine, flash controller, Clock management.Clock management is whole image pressure
Compression system provides working frequency.Data receiver controller control data receiving unit normal work, and the picture number that will be received
According to interaction to storage management states machine.Storage management states machine controls inner buffer real-time storage image, and dumps to SDRAM readings
In writing controller, SDRAM read-write controllers control SDRAM caching normal works.SDRAM cachings are used for constructing a two field picture, so
Compression host state machine is handed to by storage management states machine.Compression host state machine is input an image into DSP, and DSP is pressed
The code stream of contracting is stored in nand flash memory through flash controller.Flash controller controls flash memory normal work.Dsp processor master
Realize method for compressing image of the invention.Program storage is used for storage program.Data storage is that DSP extends DSP data
Area, storage calculates intermediate variable.
Compressibility according to embodiments of the present invention, image is carried out using what is converted after a kind of single base dictionary of low complex degree
Rarefaction representation, and carry out dynamic code rate distribution, encoder complexity using each section in bit rate controller bitplanes coding
Low, code efficiency is high, and compression performance is high.
In the description of this specification, reference term " one embodiment ", " some embodiments ", " example ", " specifically show
The description of example " or " some examples " etc. means to combine specific features, structure, material or spy that the embodiment or example are described
Point is contained at least one embodiment of the invention or example.In this manual, to the schematic representation of above-mentioned term not
Identical embodiment or example must be directed to.And, the specific features of description, structure, material or feature can be with office
Combined in an appropriate manner in one or more embodiments or example.Additionally, in the case of not conflicting, the skill of this area
Art personnel can be tied the feature of the different embodiments or example described in this specification and different embodiments or example
Close and combine.
Although embodiments of the invention have been shown and described above, it is to be understood that above-described embodiment is example
Property, it is impossible to limitation of the present invention is interpreted as, one of ordinary skill in the art within the scope of the invention can be to above-mentioned
Embodiment is changed, changes, replacing and modification.
Claims (9)
1. a kind of method for compressing image, it is characterised in that comprise the following steps:
Rarefaction representation is carried out to image, the rarefaction representation result of each picture frame is obtained;
Bit-Plane Encoding is carried out to the rarefaction representation result, to obtain Bit-Plane Encoding result;
Entropy code is carried out to the Bit-Plane Encoding result, to obtain entropy code result, and code is carried out to the entropy code result
Stream tissue packing obtains final compressed bit stream;
Wherein, it is described rarefaction representation is carried out to image to specifically include:
3 grades of wavelet transforms of two dimension 9/7 are carried out to each picture frame, low frequency sub-band and high frequency of respective image frame is obtained
Band;
Converted after single base dictionary is carried out to the high-frequency sub-band, to obtain AC coefficients and side information;
Selection is converted after single base dictionary and uses different evaluation functions, its specifically chosen process includes:
The byte number of compression frame overhead is calculated according to compression ratio parameter;
Code check is calculated according to the byte number, when the code check is code check high, conversion afterwards uses l-1Norm Method, when the code
When rate is low bit- rate, conversion afterwards uses l-0Norm Method.
2. method according to claim 1, it is characterised in that the Bit-Plane Encoding is specifically included:
DC coefficients are obtained, and the DC coefficients are initialized;
Extract paragraph header information and depth coding is carried out to the AC coefficient bits;
Extracting AC coefficients and residue DC coefficients carries out Bit-Plane Encoding;
The side information is embedded into the code stream of Bit-Plane Encoding, Bit-Plane Encoding result is obtained.
3. method according to claim 1, it is characterised in that when the code stream of Bit-Plane Encoding reaches predetermined threshold value, then
Stop Bit-Plane Encoding.
4. method according to claim 3, it is characterised in that use based on the intermediate quantity converted after single base dictionary with
And the code stream of the control Bit-Plane Encoding of code check dynamically distributes.
5. method according to claim 1, it is characterised in that coding also is predicted to the low frequency sub-band, and to pre-
Surveying residual error carries out entropy code.
6. a kind of image compression system, it is characterised in that including:
Image sparse representation module, for carrying out rarefaction representation to image, obtains the rarefaction representation result of each picture frame;
Bit-Plane Encoding module, for carrying out Bit-Plane Encoding to the rarefaction representation result, to obtain Bit-Plane Encoding result;
Entropy code module, for carrying out entropy code to the Bit-Plane Encoding result, to obtain entropy code result, and to the entropy
Coding result carries out code stream organization packing and obtains final compressed bit stream;
Wherein, described image rarefaction representation module specifically performs rarefaction representation of the following steps realization to image:
3 grades of wavelet transforms of two dimension 9/7 are carried out to each picture frame, low frequency sub-band and high frequency of respective image frame is obtained
Band;
Converted after single base dictionary is carried out to the high-frequency sub-band, to obtain AC coefficients and side information;
Selection is converted after single base dictionary and uses different evaluation functions, its specifically chosen process includes:
The byte number of compression frame overhead is calculated according to compression ratio parameter;
Code check is calculated according to the byte number, when the code check is code check high, conversion afterwards uses l-1Norm Method, when the code
When rate is low bit- rate, conversion afterwards uses l-0Norm Method.
7. system according to claim 6, it is characterised in that specific in the Bit-Plane Encoding module to perform following steps
Realize Bit-Plane Encoding:
DC coefficients are obtained, and the DC coefficients are initialized;
Extract paragraph header information and depth coding is carried out to the AC coefficient bits;
Extracting AC coefficients and residue DC coefficients carries out Bit-Plane Encoding;
The side information is embedded into the code stream of Bit-Plane Encoding, Bit-Plane Encoding result is obtained.
8. system according to claim 6, it is characterised in that the entropy code module is additionally operable to enter the low frequency sub-band
Row predictive coding, and entropy code is carried out to prediction residual.
9. system according to claim 6, it is characterised in that also include:
Rate control module, for realizing converting the different evaluation function of selection after single base dictionary, distributing the dynamic of Bit-Plane Encoding
State code stream, the predetermined threshold value of the code stream is controlled to stop Bit-Plane Encoding.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410306690.3A CN104065974B (en) | 2014-06-30 | 2014-06-30 | Method for compressing image and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410306690.3A CN104065974B (en) | 2014-06-30 | 2014-06-30 | Method for compressing image and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104065974A CN104065974A (en) | 2014-09-24 |
CN104065974B true CN104065974B (en) | 2017-06-23 |
Family
ID=51553470
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410306690.3A Active CN104065974B (en) | 2014-06-30 | 2014-06-30 | Method for compressing image and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104065974B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104361614A (en) * | 2014-11-21 | 2015-02-18 | 西安电子科技大学 | Polarization SAR image compression method based on multi-direction dictionary learning |
CN108156462A (en) * | 2017-12-28 | 2018-06-12 | 上海通途半导体科技有限公司 | A kind of compression of images, decompression method, system and its ME of application frameworks |
CN112449185B (en) * | 2019-08-28 | 2022-01-25 | 腾讯科技(深圳)有限公司 | Video decoding method, video encoding device, video encoding medium, and electronic apparatus |
WO2022120829A1 (en) * | 2020-12-11 | 2022-06-16 | 深圳市大疆创新科技有限公司 | Image encoding and decoding methods and apparatuses, and image processing apparatus and mobile platform |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102830409A (en) * | 2012-08-30 | 2012-12-19 | 西安电子科技大学 | Navigation signal acquiring method based on compressed sensing |
CN103179396A (en) * | 2013-03-04 | 2013-06-26 | 中国科学院长春光学精密机械与物理研究所 | System and method for controlling CCSDS (consultative committee for space data system) image compressing code in spatial TDICCD (time delayed integration charge coupled device) camera application |
CN103402043A (en) * | 2013-08-16 | 2013-11-20 | 中国科学院长春光学精密机械与物理研究所 | Image compression unit for large visual field TDICCD camera |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070071331A1 (en) * | 2005-09-24 | 2007-03-29 | Xiteng Liu | Image compression by economical quaternary reaching method |
-
2014
- 2014-06-30 CN CN201410306690.3A patent/CN104065974B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102830409A (en) * | 2012-08-30 | 2012-12-19 | 西安电子科技大学 | Navigation signal acquiring method based on compressed sensing |
CN103179396A (en) * | 2013-03-04 | 2013-06-26 | 中国科学院长春光学精密机械与物理研究所 | System and method for controlling CCSDS (consultative committee for space data system) image compressing code in spatial TDICCD (time delayed integration charge coupled device) camera application |
CN103402043A (en) * | 2013-08-16 | 2013-11-20 | 中国科学院长春光学精密机械与物理研究所 | Image compression unit for large visual field TDICCD camera |
Non-Patent Citations (1)
Title |
---|
李进等.适于空间TDICCD相机的图像压缩系统设计.《电子技术应用》.2013,17-19. * |
Also Published As
Publication number | Publication date |
---|---|
CN104065974A (en) | 2014-09-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104065974B (en) | Method for compressing image and system | |
CN100534186C (en) | JPEG2000 self-adapted rate control system and method based on pre-allocated code rate | |
US20050244071A1 (en) | Adaptive quantization of depth signal in 3D visual coding | |
Yang et al. | Discrete wavelet transform-based structural similarity for image quality assessment | |
CN102186076B (en) | Image compression method and device for real-time code rate pre-distribution | |
Yadav et al. | A review on image compression techniques | |
CN102833546B (en) | High-speed image compression method and device based on optimally quantized wavelet sub-band interlacing | |
KR20000059799A (en) | Device and method for motion compensation coding using wavelet coding | |
CN105144231A (en) | Method and device for selecting an image dynamic range conversion operator | |
CN103796018B (en) | A kind of remote sensing image Real Time Compression and progressive transmission system | |
CN108810534B (en) | Image compression method based on direction lifting wavelet and improved SPIHT under Internet of things | |
CN103118259B (en) | JPEG2000 image coding method | |
ZainEldin et al. | A modified listless strip based SPIHT for wireless multimedia sensor networks | |
Dubey et al. | 3D medical image compression using Huffman encoding technique | |
Joshua et al. | Comparison of DCT and DWT image compression | |
CN108718409A (en) | The remote sensing image compression method encoded based on Block direction Lifting Wavelet and adative quadtree | |
Khosravi et al. | Modified data aggregation for aerial ViSAR sensor networks in transform domain | |
CN103179396A (en) | System and method for controlling CCSDS (consultative committee for space data system) image compressing code in spatial TDICCD (time delayed integration charge coupled device) camera application | |
Singh et al. | Neuro-wavelet based efficient image compression using vector quantization | |
CN101977309B (en) | Code rate control method and device thereof | |
US8306341B2 (en) | Image data compression apparatus and decoding apparatus | |
CN104486631B (en) | A kind of remote sensing image compression method based on human eye vision Yu adaptive scanning | |
CN103402043A (en) | Image compression unit for large visual field TDICCD camera | |
WO2008042659A9 (en) | Predictive-transform source coding with subbands | |
Nagamani et al. | Image Compression Techniques for High Resolution Satellite Imageries using Classical Lifting Scheme |
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 | ||
GR01 | Patent grant |