CN104065974A - Image compression method and system - Google Patents
Image compression method and system Download PDFInfo
- Publication number
- CN104065974A CN104065974A CN201410306690.3A CN201410306690A CN104065974A CN 104065974 A CN104065974 A CN 104065974A CN 201410306690 A CN201410306690 A CN 201410306690A CN 104065974 A CN104065974 A CN 104065974A
- Authority
- CN
- China
- Prior art keywords
- bit
- carried out
- plane encoding
- image
- coefficient
- 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.)
- Granted
Links
Abstract
The invention provides an image compression method. The method comprises the following steps: performing sparse representation of an image and obtaining a sparse representation result of each image frame; carrying out bit-plane coding of the sparse representation result to obtain a bit-plane coding result; conducting entropy coding of the bit-plane coding result to obtain an entropy coding result; and performing code stream organization and packing of the entropy coding result to obtain a final compressed code stream. The method is low in coding complexity, high in efficiency, good in fault tolerance capability and high in compression performance. The invention also provides an image compression system.
Description
Technical field
The present invention relates to Image Compression field, relate in particular to a kind of method for compressing image and system.
Background technology
Full-colour image on star (spectral coverage scope is generally 450nm～900nm) is the 2-D data with spatial information atural object push-scanning image being obtained by panchromatic TDICCD camera.These data can provide abundant ground object detail, are widely used in the fields such as resource exploration, military surveillance and environmental protection.Along with the indexs such as the spatial resolution of the panchromatic TDICCD camera in space, radiometric resolution, temporal resolution, large visual field, wide covering improve constantly, cause TDICCD splicing sheet number and read-out speed that panchromatic TDICCD camera adopts are also on the increase and improve, average camera time increases, thereby the image data amount after digitlization is significantly increased.Existing spaceborne memory span is limited, and satellite channel Bandwidth-Constrained cannot adapt to the mass data of full-colour image on star.Therefore, must compress full-colour image on star.
On star, full-colour image data have two kinds of redundancies: between space, between redundancy and data, meet redundancy.Therefore, the object of full-colour image compression is eliminated this two kinds of redundancies exactly.At present, the full-colour image compression methods that adopt based on wavelet transformation on star, if the image compression algorithm using on BilSAT-1 (SSTL-Turkey) satellite for 2003 is JPEG2000 algorithm more.JPEG2000 algorithm uses the space decorrelation method of DWT (Discrete Wavelet Transform), and algorithm implementation platform is FPGA+DSP (XCV300E+TMS320C6701).2005, the Image Data Compression working group (IDC) of consultative committee for space data system (CCSDS) formulated the Standard of image compression CCSDS122.0-B-1 of space of new generation application, and this algorithm also adopts wavelet transformation.But, wavelet transformation, for full-colour image on the informative star of the texture level such as edge and profile, is not a kind of sparse expression of optimum, can produce in a large number significantly high frequency coefficient, be unfavorable for follow-up sub-band coding, make the compression performance of compression algorithm lower.
Summary of the invention
The present invention is intended to solve at least to a certain extent one of technical problem in correlation technique.
For this reason, first object of the present invention is to propose the method for compressing image that a kind of encoder complexity is low, efficiency is high, fault-tolerant ability is strong, compression performance is 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: image is carried out to rarefaction representation, obtain the rarefaction representation result of each picture frame; Described rarefaction representation result is carried out to Bit-Plane Encoding, to obtain Bit-Plane Encoding result; Described Bit-Plane Encoding result is carried out to entropy coding, to obtain entropy coding result, and described entropy coding result is carried out to code stream tissue packing obtain final compressed bit stream.
According to the method for compressing image of the embodiment of the present invention, adopt the image sparse that carries out converting after a kind of single base dictionary of low complex degree to represent, and utilize each section in bit rate controller bitplanes coding to carry out dynamic code rate distribution, encoder complexity is low, code efficiency is high, and compression performance is high.
In some instances, describedly image is carried out to rarefaction representation specifically comprise: each picture frame is carried out to 3 grades of two dimension 9/7 wavelet transforms, obtain low frequency sub-band and the high-frequency sub-band of respective image frame; Described high-frequency sub-band is carried out converting after single base dictionary, to obtain AC coefficient and side information.
In some instances, after single base dictionary, conversion is selected to adopt different evaluation functions, and its concrete selection course comprises: according to the byte number of compression ratio calculation of parameter condensed frame expense; Calculate code check according to described byte number, in the time that described code check is high code check, rear conversion adopts l
-1norm Method, in the time that described code check is low code check, rear conversion adopts l
-0norm Method.
In some instances, described Bit-Plane Encoding specifically comprises: obtain DC coefficient, and described DC coefficient is carried out to initialization; Extract paragraph header information and described AC coefficient bit is carried out to depth coding; Extract AC coefficient and remain DC coefficient and carry out Bit-Plane Encoding; Described side information is embedded in the code stream of Bit-Plane Encoding, obtains Bit-Plane Encoding result.
In some instances, in the time that the code stream of Bit-Plane Encoding reaches predetermined threshold value, position of rest plane coding.
In some instances, adopt the code stream of the described Bit-Plane Encoding of control of intermediate quantity based on converting after described single base dictionary and code check dynamic assignment.
In some instances, also described low frequency sub-band is carried out to predictive coding, and prediction residual is carried out to entropy coding.
The image compression system of second aspect present invention embodiment, comprising: image sparse representation module, for image is carried out to rarefaction representation, obtains the rarefaction representation result of each picture frame; Bit-Plane Encoding module, for carrying out Bit-Plane Encoding to described rarefaction representation result, to obtain Bit-Plane Encoding result; Entropy coding module, for described Bit-Plane Encoding result is carried out to entropy coding, to obtain entropy coding result, and carries out code stream tissue packing to described entropy coding result and obtains final compressed bit stream.
According to the image compression system of the embodiment of the present invention, adopt the image sparse that carries out converting after a kind of single base dictionary of low complex degree to represent, and utilize each section in bit rate controller bitplanes coding to carry out dynamic code rate distribution, encoder complexity is low, code efficiency is high, and compression performance is high.
In some instances, described image sparse representation module is specifically carried out following steps and is realized the rarefaction representation to image: each picture frame is carried out to 3 grades of two dimension 9/7 wavelet transforms, obtain low frequency sub-band and the high-frequency sub-band of respective image frame; Described high-frequency sub-band is carried out converting after single base dictionary, to obtain AC coefficient and side information.
In some instances, after described single base dictionary, conversion is selected to adopt different evaluation functions, and its concrete selection course comprises: according to the byte number of compression ratio calculation of parameter condensed frame expense; Calculate code check according to described byte number, in the time that described code check is high code check, rear conversion adopts l
-1norm Method, in the time that described code check is low code check, rear conversion adopts l
-0norm Method.
In some instances, in described Bit-Plane Encoding module, the concrete following steps of carrying out realize Bit-Plane Encoding: obtain DC coefficient, and described DC coefficient is carried out to initialization; Extract paragraph header information and described AC coefficient bit is carried out to depth coding; Extract AC coefficient and remain DC coefficient and carry out Bit-Plane Encoding; Described side information is embedded in the code stream of Bit-Plane Encoding, obtains Bit-Plane Encoding result.
In some instances, described entropy coding module is also for described low frequency sub-band is carried out to predictive coding, and prediction residual is carried out to entropy coding.
In some instances, also comprise: Rate Control module, select different evaluation functions for realizing conversion after single base dictionary, distribute the dynamic code stream of Bit-Plane Encoding, the predetermined threshold value of controlling described code stream to control with position of rest plane coding and to the code check of entropy coding.
The aspect that the present invention is additional and advantage in the following description part provide, and part will become obviously from the following description, or recognize by practice of the present 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 diagram 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 configuration schematic diagram of the image compression system of one embodiment of the invention.
Embodiment
Describe embodiments of the invention below in detail, the example of described embodiment is shown in the drawings, and wherein same or similar label represents same or similar element or has the element of identical or similar functions from start to finish.Be exemplary below by the embodiment being described with reference to the drawings, be intended to for explaining the present invention, and can not be interpreted as limitation of the present invention.
Technical problem to be solved by this invention for the compression method of full-colour image on star provides new technological means, for this reason, a kind of method for compressing image of low complex degree has been proposed in the embodiment of a first aspect of the present invention, comprise the following steps: image is carried out to rarefaction representation, obtain the rarefaction representation result of each picture frame; Rarefaction representation result is carried out to Bit-Plane Encoding, to obtain Bit-Plane Encoding result; Bitplanes coding result carries out entropy coding, to obtain entropy coding result, and entropy coding result is carried out to code stream tissue packing 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 the process schematic diagram of the method for compressing image of one embodiment of the invention.Specifically describe the method for compressing image of the embodiment of the present invention in conjunction with Fig. 1 and Fig. 2.
Panchromatic CCD is become taking frame as unit image with the image configuration of behavior unit, be designated as X
i(i=1,2,3 ..., N).
Step S101: image is carried out to rarefaction representation, obtain the rarefaction representation result of each picture frame.
Particularly, image being carried out to rarefaction representation comprises:
(1) each picture frame is carried out to 3 grades of two dimension 9/7 wavelet transforms, obtain low frequency sub-band and the high-frequency sub-band of respective image frame.Be present frame X
iadopt 3 grade of 9/7 lifting wavelet transform, obtain 1 low frequency sub-band LL and 9 high-frequency sub-band HL1, HL2, HL3, LH1, LH2, LH3, HH1, HH2, HH3.Each high-frequency sub-band HL3, LH3, HH3 and HL1, LH1, in HH1, wavelet coefficient is made into some with 4*4 size groups, is designated as A
i(i=1,2,3 ..., J).Each high-frequency sub-band HL2, LH2, in HH2, wavelet coefficient is made into some with 2*2 size groups, is designated as B
i(i=1,2,3 ..., K).
(2) high-frequency sub-band is carried out converting after single base dictionary, to obtain AC coefficient and side information.
Particularly, in an embodiment of the present invention, each A
iand B
iinvariant position, and will until after transform block be designated as fw (w=1,2,3 ..., W).Each is carried out converting after the mono-base dictionary of Hadamard, obtain rear conversion coefficient and side information.The base dictionary using in rear when conversion only has a Hadamard base, and Hadamard Base computing only has addition and shift operation, calculates very simply, and is convenient to hardware realization, and has higher compression performance.
First to f
1carry out rear conversion, by coefficient block f
1after fill order's base, convert, rear transformation calculations formula is:
Wherein, G is rear transform block size,
for rear transform-based vector, a[g] rear conversion coefficient.
Secondly, to coefficient block f
1the best after convert selection, select to adopt different evaluation function, the concrete selection course of its evaluation function is:
(1) according to the byte number of compression ratio calculation of parameter condensed frame expense.
The byte number of the compression ratio calculation of parameter condensed frame expense that total bitrate computing unit injects according to camera controller, byte number is:
wherein, M and N are respectively picture frame size, and M is the effective pixel number of CCD, and N is CCD line number, and L is each bit depths of pixels, and R is compression ratio.
(2) calculate code check according to condensed frame overhead byte number, calculating formula is: r=8 × B
c/ (M × N).In the time that code check is high code check (as r>=1bpp), rear conversion adopts l
-1norm Method; In the time that code check is low code check (as r < 1bpp), rear conversion adopts l
-0norm Method.
Therefore, best rear conversion selects to adopt l
-pthe method of norm, in the time being low code check, p=0, adopts l
-0norm method, adopts following formula to calculate:
Work as p=1, adopt l
-1norm method, adopts following formula to calculate:
Wherein, N
bfor the number of base in dictionary, owing to adopting single base dictionary, therefore N
b=1.Work as N
b=0 o'clock,
now do not carry out rear conversion, after the best, be transformed to original sparse piece.When conversion is selected after best, adopt different evaluation functions according to different code checks, so greatly improved computational efficiency, reduced computation complexity.
Repeat above-mentioned steps S101, until present frame X
iall high-frequency sub-band complete after conversion.Like this, the coefficient in wavelet field is converted again, can further remove the redundant information between wavelet coefficient, make up the deficiency of full-colour image small echo rarefaction representation on star.
Step S102: rarefaction representation result is carried out to Bit-Plane Encoding, to obtain Bit-Plane Encoding result.
Particularly, in one embodiment of the invention, the rear conversion coefficient tissue that above-mentioned steps S101 is obtained becomes some sections, carries out BPE coding taking section as unit.Each section is made up of 16 pieces, and every is made up of 1 DC coefficient (being low frequency sub-band) and 63 AC coefficients (being the rear conversion coefficient of high-frequency sub-band).Each section is designated as S
t(i=1,2,3 ..., T).
To section, S1 encodes, and obtains DC coefficient that is:, and DC coefficient is carried out to initialization; Extract paragraph header information and AC coefficient bit is carried out to depth coding; Extract AC coefficient and remain DC coefficient and carry out Bit-Plane Encoding; The side information that step S101 is obtained is embedded in the code stream of Bit-Plane Encoding, obtains escape word, i.e. section S1 Bit-Plane Encoding result.So not only hardware is realized simply, and has higher compression performance, in addition, makes coding have gradual feature, has improved the robustness of compression algorithm.
Especially, in the time that the code stream of Bit-Plane Encoding reaches predetermined threshold value, position of rest plane coding.The process of concrete bit-allocation control device code stream control is as follows:
(1) segment information is estimated and weight calculation.Section S
tin gross information content be
After having utilized dexterously in step S101 when the estimation of amount of information of section, the intermediate object program of transformation calculations, has improved computational efficiency.
(2) calculate each section of shared code check.
Particularly, calculate the summation of all segment information amounts, and calculate the shared ratio of each segment information.If the gross information content of i section is Ii, every two field picture contains R section.Every section of shared information scales is
If image compression code check is R
s, the code check of each section of distribution is r
i,s=r
i× R
s.The target word joint number of the distribution in final coding section i is: P
s=M × N × R
s/ B.Wherein, M, the size that N is image.B is that image bit is dark.Employing contains different amount of information according to each section and dynamically distributes code check, and the strategy of dividing equally that has overcome code check section causes the problem that compression performance is low.In each section, adopt control bit plane to carry out Rate Control not only simple but also there is higher compression performance, and realize more convenient.
In the time that the code stream of Bit-Plane Encoding reaches predetermined threshold value, apply bit plane depth encoder control position of rest plane coding.
Repeat above-mentioned steps S102, until complete the coding of all sections.
Meanwhile, the low frequency sub-band that step S101 is obtained carries out predictive coding, to obtain the prediction residual of this low frequency sub-band predictive coding.
Step S103: bitplanes coding result carries out entropy coding, to obtain entropy coding result, and carries out code stream tissue packing to entropy coding result and obtains final compressed bit stream.
The Bit-Plane Encoding result that step S102 is obtained and the prediction residual of low frequency sub-band predictive coding are carried out entropy coding, obtain entropy coding result, and entropy coding result is carried out to code stream tissue packing obtain final compressed bit stream.
Repeat above-mentioned steps S101～S103, until all picture frame end-of-encodes.
According to the method for compressing image of the embodiment of the present invention, adopt the image sparse that carries out converting after a kind of single base dictionary of low complex degree to represent, and utilize each section in bit rate controller bitplanes coding to carry out dynamic code rate distribution, encoder complexity is low, code efficiency is high, and compression performance is high.
In the embodiment of second aspect present invention, propose a kind of image compression system, as shown in Figure 3, comprising: image sparse representation module 100, Bit-Plane Encoding module 200 and entropy coding module 400.
Wherein, image sparse representation module 100, for image is carried out to rarefaction representation, obtains the rarefaction representation result of each picture frame.Bit-Plane Encoding module 200, for rarefaction representation result is carried out to Bit-Plane Encoding, to obtain Bit-Plane Encoding result.Entropy coding module 300, carries out entropy coding for bitplanes coding result, to obtain entropy coding result, and entropy coding result is carried out to code stream tissue packing obtains final compressed bit stream.
Further, the image compression system of embodiments of the invention, also comprise: Rate Control module 400, select different evaluation functions for conversion after realizing single base dictionary, distribute the dynamic code stream of Bit-Plane Encoding and the predetermined threshold value of control stream with position of rest plane coding.
Particularly, in image sparse representation module 100, panchromatic CCD is configured to, taking frame as unit image, be designated as X through buffer unit with the image of behavior unit
i(i=1,2,3 ..., N).In one embodiment of the invention, buffer unit adopts the SDRAM of ping-pong operation to realize.
In image sparse representation module 100, image is carried out to rarefaction representation, obtain the rarefaction representation result of each picture frame.
Particularly, image being carried out to rarefaction representation comprises:
(1) each picture frame is carried out to 3 grades of two dimension 9/7 wavelet transforms, obtain low frequency sub-band and the high-frequency sub-band of respective image frame.Be present frame X
iadopt 3 grade of 9/7 lifting wavelet transform, obtain 1 low frequency sub-band LL and 9 high-frequency sub-band HL1, HL2, HL3, LH1, LH2, LH3, HH1, HH2, HH3.Each high-frequency sub-band HL3, LH3, HH3 and HL1, LH1, in HH1, wavelet coefficient is made into some with 4*4 size groups, is designated as A
i(i=1,2,3 ..., J).Each high-frequency sub-band HL2, LH2, in HH2, wavelet coefficient is made into some with 2*2 size groups, is designated as B
i(i=1,2,3 ..., K).
(2) high-frequency sub-band is carried out converting after single base dictionary, to obtain AC coefficient and side information.
Particularly, in an embodiment of the present invention, each A
iand B
iinvariant position, and will until after transform block be designated as fw (w=1,2,3 ..., W).Each is carried out converting after the mono-base dictionary of Hadamard, obtain rear conversion coefficient and side information.The base dictionary using in rear when conversion only has a Hadamard base, and Hadamard Base computing only has addition and shift operation, calculates very simply, and is convenient to hardware realization, and has higher compression performance.
First to f
1carry out rear conversion, by coefficient block f
1after fill order's base, convert, rear transformation calculations formula is:
Wherein, G is rear transform block size,
for rear transform-based vector, a[g] rear conversion coefficient.
Secondly, to coefficient block f
1the best after convert selection, selected to adopt different evaluation function by evaluation function arbitration unit, the concrete selection course of its evaluation function is:
(1) according to the byte number of compression ratio calculation of parameter condensed frame expense.
The byte number of the compression ratio calculation of parameter condensed frame expense that total bitrate computing unit injects according to camera controller, byte number is:
wherein, M and N are respectively picture frame size, and M is the effective pixel number of CCD, and N is CCD line number, and L is each bit depths of pixels, and R is compression ratio.
(2) calculate code check according to condensed frame overhead byte number, calculating formula is: r=8 × B
c/ (M × N).In the time that code check is high code check (as r>=1bpp), rear conversion adopts l
-1norm Method; In the time that code check is low code check (as r < 1bpp), rear conversion adopts l
-0norm Method.
Therefore, best rear conversion selects to adopt l
-pthe method of norm, in the time being low code check, p=0, adopts l
-0norm method, adopts following formula to calculate:
Work as p=1, adopt l
-1norm method, adopts following formula to calculate:
Wherein, N
bfor the number of base in dictionary, owing to adopting single base dictionary, therefore N
b=1.Work as N
b=0 o'clock,
now do not carry out rear conversion, after the best, be transformed to original sparse piece.When conversion is selected after best, adopt different evaluation functions according to different code checks, so greatly improved computational efficiency, reduced computation complexity.
Repeat above-mentioned steps, until present frame X
iall high-frequency sub-band complete after conversion.Like this, the coefficient in wavelet field is converted again, can further remove the redundant information between wavelet coefficient, make up the deficiency of full-colour image small echo rarefaction representation on star.
In Bit-Plane Encoding module 200, rarefaction representation result is carried out to Bit-Plane Encoding, to obtain Bit-Plane Encoding result.
Particularly, in one embodiment of the invention, the rear conversion coefficient obtaining in above-mentioned image sparse representation module 100 is equally organized into tree structure according to wavelet sub-band, then becomes some sections through CCSDS-IDC bit plane encoder tissue, carry out BPE coding taking section as unit.Each section is made up of 16 pieces, and every is made up of 1 DC coefficient (being low frequency sub-band) and 63 AC coefficients (being the rear conversion coefficient of high-frequency sub-band).Each section is designated as S
t(i=1,2,3 ..., T).
To section, S1 encodes, and obtains DC coefficient that is:, and DC coefficient is carried out to initialization; Extract paragraph header information and AC coefficient bit is carried out to depth coding; Extract AC coefficient and remain DC coefficient and carry out Bit-Plane Encoding; The side information that step S101 is obtained is embedded in the code stream of Bit-Plane Encoding, obtains escape word, i.e. section S1 Bit-Plane Encoding result.So not only hardware is realized simply, has higher compression performance simultaneously, in addition, makes coding have gradual feature, has improved the robustness of compression algorithm.
Especially, in the time that the code stream of Bit-Plane Encoding reaches predetermined threshold value, position of rest plane coding.The process of concrete code stream control is as follows:
(1) segment information is estimated and weight calculation.Section S
tin gross information content be
After having utilized dexterously in image sparse representation module 100 when the estimation of amount of information of section, the intermediate object program of transformation calculations, has improved computational efficiency.
(2) calculate each section of shared code check.
Particularly, calculate the summation of all segment information amounts, and calculate the shared ratio of each segment information.If the gross information content of i section is Ii, every two field picture contains R section.Every section of shared information scales is
If image compression code check is R
s, the code check of each section of distribution is r
i,s=r
i× R
s.The target word joint number of the distribution in final coding section i is: P
s=M × N × R
s/ B.Wherein, M, the size that N is image.B is that image bit is dark.Employing contains different amount of information according to each section and dynamically distributes code check, and the strategy of dividing equally that has overcome code check section causes the problem that compression performance is low.In each section, adopt control bit plane to carry out Rate Control not only simple but also there is higher compression performance, and realize more convenient.
In the time that the code stream of Bit-Plane Encoding reaches predetermined threshold value, apply bit plane depth encoder control position of rest plane coding.
Repeat above-mentioned steps, until complete the coding of all sections.
Meanwhile, the low frequency sub-band that image sparse representation module 100 is obtained carries out predictive coding, to obtain the prediction residual of this low frequency sub-band predictive coding.
In entropy coding module 300, bitplanes coding result carries out entropy coding, to obtain entropy coding result, and entropy coding result is carried out to code stream tissue packing obtains final compressed bit stream.The prediction residual of the Bit-Plane Encoding result of obtaining in Bit-Plane Encoding module 200 and low frequency sub-band predictive coding is carried out to entropy coding, obtain entropy coding result, and entropy coding result is carried out to code stream tissue packing obtain final compressed bit stream.
The Rate Control module 400 that the bit rate controller being used by the evaluation function arbitration unit adopting in above-mentioned image sparse representation module 100 and Bit-Plane Encoding module 200 and Bit-Plane Encoding depth controller form, after having realized single base dictionary, different evaluation functions is selected in conversion, distributes the dynamic code stream of Bit-Plane Encoding and the predetermined threshold value of control stream with position of rest plane coding.
The hardware configuration of the image compression system of the one embodiment of the invention being realized by the hardware processor such as FPGA, DSP as shown in Figure 4.The hardware composition of this image compression system mainly comprises: full-colour image data sink, FPGA processor, nand flash memory array, SDRAM buffer memory, dsp processor, program storage and data storage.
Full-colour image receiver is used for receiving the ccd image of CCD image-generating unit output and being input in FPGA processor.FPGA processor is realized whole compression work.Wherein, FPGA processor comprises: data receiver controller, storage management states machine, SDRAM read-write controller, inner buffer compression host state machine, flash controller, Clock management.Clock management provides operating frequency for whole image compression system.Normally work in data receiver controller control data receiver unit, and the view data receiving is arrived to storage management states machine alternately.Storage management states machine control inner buffer real-time storage image, and dump in SDRAM read-write controller, SDRAM read-write controller control SDRAM buffer memory is normally worked.SDRAM buffer memory is used for constructing a two field picture, then hands to compression host state machine through storage management states machine.Compression host state machine is input to image in DSP, and the code stream that DSP is compressed is through flash controller, stores in nand flash memory.Flash controller control flash memory is normally worked.Dsp processor is mainly realized method for compressing image of the present invention.Program storage is used for storage program.Data storage is DSP expansion DSP data field, and intermediate variable is calculated in storage.
According to the compressibility of the embodiment of the present invention, adopt the image sparse that carries out converting after a kind of single base dictionary of low complex degree to represent, and utilize each section in bit rate controller bitplanes coding to carry out dynamic code rate distribution, encoder complexity is low, code efficiency is high, and compression performance is high.
In the description of this specification, the description of reference term " embodiment ", " some embodiment ", " example ", " concrete example " or " some examples " etc. means to be contained at least one embodiment of the present invention or example in conjunction with specific features, structure, material or the feature of this embodiment or example description.In this manual, to the schematic statement of above-mentioned term not must for be identical embodiment or example.And, specific features, structure, material or the feature of description can one or more embodiment in office or example in suitable mode combination.In addition,, not conflicting in the situation that, those skilled in the art can carry out combination and combination by the feature of the different embodiment that describe in this specification or example and different embodiment or example.
Although illustrated and described embodiments of the invention above, be understandable that, above-described embodiment is exemplary, can not be interpreted as limitation of the present invention, and those of ordinary skill in the art can change above-described embodiment within the scope of the invention, amendment, replacement and modification.
Claims (13)
1. a method for compressing image, is characterized in that, comprises the following steps:
Image is carried out to rarefaction representation, obtain the rarefaction representation result of each picture frame;
Described rarefaction representation result is carried out to Bit-Plane Encoding, to obtain Bit-Plane Encoding result;
Described Bit-Plane Encoding result is carried out to entropy coding, to obtain entropy coding result, and described entropy coding result is carried out to code stream tissue packing obtain final compressed bit stream.
2. method according to claim 1, is characterized in that, describedly image is carried out to rarefaction representation specifically comprises:
Each picture frame is carried out to 3 grades of two dimension 9/7 wavelet transforms, obtain low frequency sub-band and the high-frequency sub-band of respective image frame;
Described high-frequency sub-band is carried out converting after single base dictionary, to obtain AC coefficient and side information.
3. method according to claim 2, is characterized in that, after single base dictionary, conversion is selected to adopt different evaluation functions, and its concrete selection course comprises:
According to the byte number of compression ratio calculation of parameter condensed frame expense;
Calculate code check according to described byte number, in the time that described code check is high code check, rear conversion adopts l
-1norm Method, in the time that described code check is low code check, rear conversion adopts l
-0norm Method.
4. method according to claim 2, is characterized in that, described Bit-Plane Encoding specifically comprises:
Obtain DC coefficient, and described DC coefficient is carried out to initialization;
Extract paragraph header information and described AC coefficient bit is carried out to depth coding;
Extract AC coefficient and remain DC coefficient and carry out Bit-Plane Encoding;
Described side information is embedded in the code stream of Bit-Plane Encoding, obtains Bit-Plane Encoding result.
5. method according to claim 1, is characterized in that, in the time that the code stream of Bit-Plane Encoding reaches predetermined threshold value, and position of rest plane coding.
6. method according to claim 5, is characterized in that, adopts the code stream of the described Bit-Plane Encoding of control of intermediate quantity based on converting after described single base dictionary and code check dynamic assignment.
7. method according to claim 2, is characterized in that, also described low frequency sub-band is carried out to predictive coding, and prediction residual is carried out to entropy coding.
8. an image compression system, is characterized in that, comprising:
Image sparse representation module, for image is carried out to rarefaction representation, obtains the rarefaction representation result of each picture frame;
Bit-Plane Encoding module, for carrying out Bit-Plane Encoding to described rarefaction representation result, to obtain Bit-Plane Encoding result;
Entropy coding module, for described Bit-Plane Encoding result is carried out to entropy coding, to obtain entropy coding result, and carries out code stream tissue packing to described entropy coding result and obtains final compressed bit stream.
9. system according to claim 8, is characterized in that, described image sparse representation module is specifically carried out following steps and realized the rarefaction representation to image:
Each picture frame is carried out to 3 grades of two dimension 9/7 wavelet transforms, obtain low frequency sub-band and the high-frequency sub-band of respective image frame;
Described high-frequency sub-band is carried out converting after single base dictionary, to obtain AC coefficient and side information.
10. system according to claim 9, is characterized in that, after single base dictionary, conversion is selected to adopt different evaluation functions, and its concrete selection course comprises:
According to the byte number of compression ratio calculation of parameter condensed frame expense;
Calculate code check according to described byte number, in the time that described code check is high code check, rear conversion adopts l
-1norm Method, in the time that described code check is low code check, rear conversion adopts l
-0norm Method.
11. systems according to claim 8, is characterized in that, in described Bit-Plane Encoding module, the concrete following steps of carrying out realize Bit-Plane Encoding:
Obtain DC coefficient, and described DC coefficient is carried out to initialization;
Extract paragraph header information and described AC coefficient bit is carried out to depth coding;
Extract AC coefficient and remain DC coefficient and carry out Bit-Plane Encoding;
Described side information is embedded in the code stream of Bit-Plane Encoding, obtains Bit-Plane Encoding result.
12. systems according to claim 9, is characterized in that, described entropy coding module is also for described low frequency sub-band is carried out to predictive coding, and prediction residual is carried out to entropy coding.
13. systems according to claim 8, is characterized in that, also comprise:
Rate Control module, select different evaluation functions for realizing conversion after single base dictionary, distribute Bit-Plane Encoding dynamic code stream, control described code stream predetermined threshold value with position of rest plane coding.
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 true CN104065974A (en) | 2014-09-24 |
| CN104065974B 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) |
Cited By (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 |
| CN112449185A (en) * | 2019-08-28 | 2021-03-05 | 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 (4)
| 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 |
| 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 |
-
2014
- 2014-06-30 CN CN201410306690.3A patent/CN104065974B/en active Active
Patent Citations (4)
| 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 |
| 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》 * |
Cited By (5)
| 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 |
| CN112449185A (en) * | 2019-08-28 | 2021-03-05 | 腾讯科技（深圳）有限公司 | Video decoding method, video encoding device, video encoding medium, and electronic apparatus |
| 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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104065974B (en) | 2017-06-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104301724B (en) | Method for processing video frequency, encoding device and decoding device | |
| CN100534186C (en) | JPEG2000 self-adapted rate control system and method based on pre-allocated code rate | |
| CN104065974A (en) | Image compression method and system | |
| CN105144231A (en) | Method and device for selecting an image dynamic range conversion operator | |
| CN102523453B (en) | Super large compression method and transmission system for images | |
| 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 | |
| Demaret et al. | Adaptive thinning for terrain modelling and image compression | |
| CN102036073A (en) | Method for encoding and decoding JPEG2000 image based on vision potential attention target area | |
| CN102833546A (en) | High-speed image compression method and device based on optimally quantized wavelet sub-band interlacing | |
| Nair et al. | Deep-learning with context sensitive quantization and interpolation for underwater image compression and quality image restoration | |
| CN108718409B (en) | Remote sensing image compression method | |
| CN102695055B (en) | JPEG_LS (Joint Pho-tographic Experts Group-Lossless Standard) bit rate control method under high bit rate | |
| CN103985096A (en) | Hyperspectral image regression prediction compression method based on off-line training | |
| Kaddachi et al. | FPGA-based image compression for low-power Wireless Camera Sensor Networks | |
| RU2419246C1 (en) | Method to compress and recover fixed halftone video images | |
| CN108200440B (en) | A kind of distributed video compressed sensing reconstructing method based on temporal correlation | |
| 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 | |
| CN104869426A (en) | JPEG coding method lowering image diamond effect under low compression code rate | |
| Singh et al. | Neuro-wavelet based efficient image compression using vector quantization | |
| CN104079930A (en) | Achieving method of remote-sensing image compression system | |
| US8306341B2 (en) | Image data compression apparatus and decoding apparatus | |
| CN103780904A (en) | Hyperspectral image self-adaption lossless predictive encoding system and method based on edge information | |
| CN102685498B (en) | Lossless compression method for three-dimensional sonar image sequence based on JPEG-LS (joint photographic experts group-lossless standard) interframe expansion |
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 |