CN102148993A - Method and device for encoding wavelet image - Google Patents

Method and device for encoding wavelet image Download PDF

Info

Publication number
CN102148993A
CN102148993A CN2010101163258A CN201010116325A CN102148993A CN 102148993 A CN102148993 A CN 102148993A CN 2010101163258 A CN2010101163258 A CN 2010101163258A CN 201010116325 A CN201010116325 A CN 201010116325A CN 102148993 A CN102148993 A CN 102148993A
Authority
CN
China
Prior art keywords
subband
wavelet
module
coefficient
tabulation
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
Application number
CN2010101163258A
Other languages
Chinese (zh)
Other versions
CN102148993B (en
Inventor
左雯
王宁
于培松
郭秀江
卓力
田卫
李晓光
周真理
张菁
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
ZTE Corp
Original Assignee
ZTE Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by ZTE Corp filed Critical ZTE Corp
Priority to CN 201010116325 priority Critical patent/CN102148993B/en
Publication of CN102148993A publication Critical patent/CN102148993A/en
Application granted granted Critical
Publication of CN102148993B publication Critical patent/CN102148993B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Abstract

The invention discloses a method for encoding a wavelet image. The method comprises the following steps of: classifying corresponding wavelet coefficients, which are obtained by wavelet decomposition, of each subband by using a pyramid structure; performing corresponding encoding operation on the classified wavelet coefficients to acquire a compressed code stream of each subband; according to a rate distortion characteristic of the compressed code stream of each subband, optimally allocating a target code rate between the compressed code streams of each subband; and according to the allocated code rate of each subband, cutting off the code stream of each subband to acquire the code stream for encoding the wavelet image. The invention also discloses a device for encoding the wavelet image. By the method and the device, the error-code-resisting capability of the code stream acquired by encoding can be enhanced and the code stream is decoded to obtain images of multiple spatial resolutions.

Description

A kind of coding method of Wavelet image and device
Technical field
The present invention relates to digital image coding and technical field of video coding, relate in particular to a kind of coding method and device of Wavelet image.
Background technology
Since the eighties in 20th century, multiresolution analysis ability that wavelet transformation conforms to human-eye visual characteristic because of it is distinctive and direction selective power, and be widely used in the image compression field, and obtained good effect.
Image can obtain the subband of a series of different resolutions after wavelet decomposition, the frequency of the subband correspondence of different resolution is also different, again the subband that obtains is encoded afterwards, just Wavelet image is encoded, thereby reach the purpose of compressed image, so that carry out the operation of successive image transmission.The key of Wavelet image coding is exactly how to organize better and represent wavelet coefficient, thereby the statistical property of utilizing wavelet coefficient better and being had realizes maximum data compression.Wherein, described wavelet coefficient is corresponding with image each subband of gained after wavelet decomposition, and each subband all has corresponding with it a plurality of wavelet coefficients.
At present, the Wavelet image Methods for Coding has a lot, comprise: vector quantization, scalar quantization, zerotree image and zero block encoding etc., in these coding methods, code efficiency is than higher mainly contain two kinds: based on the wavelet coding method of zero tree (Zero-tree) with based on the wavelet coding method of zero piece (Zero-block).
Wherein, in the described wavelet coding method based on zero tree, more representational have an embedded zero-tree wavelet (EZW, Embedded Zero-tree Wavelet) coding method and layering wavelet tree set partitioning (SPIHT, Set Partitioning In Hierarchical Trees) coding method.Here, described EZW coding method has made full use of the similar characteristic of wavelet coefficient between different scale, has rejected the coding to the high frequency wavelet coefficient effectively, has greatly improved the code efficiency of wavelet coefficient; Described SPIHT coding method, it is a kind of Wavelet image more efficiently coding method that on EZW coding method basis, proposes, set by the design space direction and more effectively to organize wavelet coefficient, the data structure of described direction in space tree has not only made full use of the correlation of wavelet coefficient between different scale, also taken into full account the correlation of wavelet coefficient under the same yardstick, therefore can more effectively organize wavelet coefficient, improve code efficiency.
Described wavelet coding method based on zero piece, the wavelet coefficient that adopts quaternary tree to decompose is organized mode, be about to image and be divided into four zones, the corresponding collection in each zone, if these all pixel values of concentrating are all less than the threshold value of current preset, thinking that then this collection is inessential, is " zero piece ", just can represent the state that this concentrates all wavelet coefficients with bit 0; If these all pixel values of concentrating are all greater than the threshold value of current preset, think that then this collection is important, this zone subdivision is cleaved into four sub regions obtains four subclass, then these four subclass being repeated importance detects, still adopt described quaternary tree decomposition method to proceed division to important subclass, up to determining all important wavelet coefficients under the current threshold value.Wavelet coding based on zero piece mainly contains set partitioning embedded block coding method (SPECK, Set Partitioned Embedded Block Coder) and optimize the embedded block coding method (EBCOT, Embedded Block Codingwith Optimized Truncation) etc. of point of cut-off.
Though wavelet coding methods such as existing EZW, SPIHT and SPECK have the characteristic of embedding, but the embedded bitstream that these coding methods obtain is very responsive to error code, because described existing coding method utilizes the correlation between the different sub-band wavelet coefficient, so, in a single day if occur error code in the transmission course of code stream, correlation between wavelet coefficient is destroyed, then the code stream after this error code will lose synchronously, cause the follow-up image that code stream decoding is gone out " changed beyond recognition ", as seen, the anti-error code capacity of existing wavelet coding method is relatively poor.In addition, same owing to have the correlation between wavelet coefficient in the cataloged procedure now, make existing wavelet coding method not support multiple spatial resolution decoding, that is to say, the compressed bit stream that existing wavelet coding method obtains can only be fixed the decoded picture of resolution after decoding, range of application is restricted.
Summary of the invention
In view of this, main purpose of the present invention is to provide a kind of coding method and device of Wavelet image, can improve the anti-error code capacity of coding gained code stream, and code stream is supported the decoding of multiple spatial resolution.
For achieving the above object, technical scheme of the present invention is achieved in that
The invention provides a kind of coding method of Wavelet image, this method comprises:
Utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, more classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband; According to the rate distortion characteristic of each subband compressed bit stream, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, the code check that is assigned with according to each subband blocks the code stream of each subband, obtains the code stream that Wavelet image is encoded.
Wherein, the described pyramid structure of utilizing is classified to the wavelet coefficient through each subband correspondence of wavelet decomposition gained, more classified wavelet coefficient is carried out the process of corresponding codes operation, is specially:
For current sub makes up corresponding pyramid structure;
For current sub is provided with first pixel tabulation and Section Point tabulation, and initialization be provided with the tabulation of first pixel and Section Point in tabulating content and the initial value size of bit-planes coefficient n;
All wavelet coefficients in the constructed pyramid structure are classified;
Wavelet coefficient in first pixel tabulation that classification is obtained is carried out Refinement operation;
Judge the magnitude relationship of current bit-planes coefficient n and 1,, and begin to continue to carry out follow-up cataloged procedure from described execution sort operation step if n more than or equal to 1, then makes n=n-1; If n, then finishes the complete cataloged procedure of wavelet coefficient of current sub less than 1.
Wherein, the pyramid structure that described current sub makes up is made up of the l level, and the maximum of described l is max{log 2 W, log 2 H.
Wherein, described W and H are respectively the width and the height of current sub.
Wherein, described when first pixel tabulation and Section Point tabulation being set for current sub, one first pixel tabulation and l Section Point are set tabulate.
Wherein, describedly distribute target bit rate between the compressed bit stream of each subband, the process of blocking code stream according to the code check that is assigned with is:
Determine to block the optimum point of cut-off of each subband correspondence according to the rate distortion characteristic of each subband according to the compressed bit stream of described optimum point of cut-off to each subband.
The present invention also provides a kind of code device of Wavelet image, and this device comprises: coding module and truncation module; Wherein,
Described coding module, be used to utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, again classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband, and the compressed bit stream of each subband of gained is transferred to truncation module;
Described truncation module, the rate distortion characteristic of each the subband compressed bit stream that is used for sending out according to coding module, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, and the code check that is assigned with according to each subband blocks the code stream of each subband, obtains the code stream that Wavelet image is encoded.
Wherein, described coding module utilizes pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, and more classified wavelet coefficient is carried out the corresponding codes operation, is specially:
Coding module earlier makes up corresponding pyramid structure for current sub, and for current sub is provided with the tabulation of first pixel and Section Point is tabulated, and initialization is provided with content during first pixel is tabulated and Section Point is tabulated and the initial value size of bit-planes coefficient n; All wavelet coefficients in the constructed pyramid structure are classified, and the wavelet coefficient in first pixel that classification the is obtained tabulation is carried out Refinement operation;
Also be used to judge the magnitude relationship of current bit-planes coefficient n and 1, determine that n more than or equal to 1 o'clock, makes n=n-1, and begin to continue to carry out follow-up cataloged procedure from described execution sort operation step; Determine n less than 1 o'clock, finish the complete cataloged procedure of wavelet coefficient of current sub.
Wherein, described coding module further comprises: pyramid construction module, initialization module, sort module, refinement module and quantization modules; Wherein,
Described pyramid construction module is used to the subband of present encoding to make up corresponding pyramid structure, and will make up the result and send to initialization module;
Described initialization module, be used to current sub that tabulation of first pixel and Section Point tabulation are set, and initialization be provided with Section Point tabulation and first pixel in tabulating content and the initial value size of bit-planes coefficient n, and the result that will be provided with sends to sort module;
Described sort module, the initial value that is used for being sent out with initialization module serve as that foundation is classified to all wavelet coefficients of the pyramid structure of current sub correspondence, and classification results is sent to the refinement module; After also being used to receive the current bit-planes coefficient n that quantization modules sends out, serve as according to continuing to carry out sort operation with new n;
Described refinement module, the wavelet coefficient that is used for first pixel tabulation that classification obtains to sort module is carried out Refinement operation, and triggers quantization modules after executing current Refinement operation; Be used to also to receive after the notice of quantization modules that the refinement result with all wavelet coefficients of current sub sends to truncation module;
Described quantization modules is used for judging the magnitude relationship of sort module current bit-planes coefficient n and 1, if n more than or equal to 1, then n subtracts 1, the n after will successively decreasing sends to sort module, and to trigger sort module serve as according to continuing to carry out sort operation with new n;
Described quantization modules also was used for determining the current bit-planes coefficient of sort module n less than 1 o'clock, and notice refinement module sends to truncation module with the refinement result of all wavelet coefficients of current sub.
In the such scheme, described truncation module is distributed target bit rate between the compressed bit stream of each subband, block being operating as of code stream according to the code check that is assigned with: determine to block the optimum point of cut-off of each subband correspondence according to the compressed bit stream of described optimum point of cut-off to each subband according to the rate distortion characteristic of each subband.
The coding method of Wavelet image provided by the invention and device utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, and more classified wavelet coefficient are carried out the corresponding codes operation, obtain the compressed bit stream of each subband; According to the rate distortion characteristic of each subband compressed bit stream, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, the code check that is assigned with according to each subband blocks the code stream of each subband, and the gained code stream is the code stream that Wavelet image is encoded.Cataloged procedure of the present invention is respectively each subband independently to be encoded, do not relate to the correlation between the different sub-band wavelet coefficient, cataloged procedure is existing relatively simple, therefore, if the code stream generation error code of certain subband can not have influence on the code stream of other subband in the transmission, anti-error code capacity is improved.
In addition, rearrange because the code stream that finally obtains of coding is the subband code stream of different progression, therefore, can decode to the code stream of subseries selectively when decoding, thereby obtain the decoded picture of different resolution, user experience is improved.
Description of drawings
Fig. 1 is the coding method realization flow schematic diagram of Wavelet image of the present invention;
Fig. 2 is the structural representation of the output code flow of gained behind the wavelet coding of the present invention;
Fig. 3 carries out complete Methods for Coding realization flow schematic diagram for the wavelet coefficient to each subband of the present invention;
Fig. 4 is the structural representation that pyramid structure method for organizing of the present invention makes up gained;
Fig. 5 is a Wavelet image apparatus for encoding structural representation of the present invention.
Embodiment
Basic thought of the present invention is: utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, more classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband; Rate distortion characteristic according to each subband compressed bit stream is carried out optimum allocation with target bit rate between the compressed bit stream of each subband, the code check that is assigned with according to each subband blocks the code stream of each subband, and the gained code stream is the code stream that Wavelet image is encoded.
Below in conjunction with drawings and the specific embodiments the present invention is described in further detail.
Fig. 1 is the coding method realization flow schematic diagram of Wavelet image of the present invention, and as shown in Figure 1, this flow process performing step is as follows:
Step 101: utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, more classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband;
Be specially: image obtains corresponding Wavelet image after wavelet decomposition, can as required original image be carried out the decomposition of different progression, as: carry out three grades of wavelet decomposition, decomposable process is a prior art, no longer describes in detail herein; Respectively the wavelet coefficient through each subband correspondence of wavelet decomposition gained is carried out encoding operation afterwards, obtain the compressed bit stream of each subband respectively.Wherein, described coding fully is meant each subband independently encoded, and the cataloged procedure of each subband is not related to correlation between each subband wavelet coefficient.
Here, in the described process that the wavelet coefficient of each subband is encoded fully, at first the wavelet coefficient of each subband is carried out l level pyramid structure and make up, described l is an integer, and the maximum of l is max{log 2 W, log 2 H, wherein, described W and H are respectively the width and the height of current sub; And an important pixel tabulation (LSP is set for each subband, List of Significant Pixels) and l inessential node listing (LIN, List ofInsignificant Node), corresponding LIN of each grade pyramid, among the present invention, LSP is described as the tabulation of first pixel; LIN is described as the Section Point tabulation.When the wavelet coefficient of each subband is encoded, at first want the tabulation of first pixel and the tabulation of all Section Points of initialization current sub, want initialization bit floor coefficient n simultaneously, described n is an integer, and its value is
Figure GSA00000023029300061
(i j) is the wavelet coefficient values of current sub to described c; According to n the coefficient in each grade pyramid structure is classified then, assorting process finishes up to lowermost level from the highest beginning of pyramid, obtain the tabulation of first pixel, again the coefficient in the tabulation of first pixel is carried out refinement, obtain n the Most Significant Bit (MSB, Most SignificantBit) of each coefficient; With n-1, repeat above-mentioned classification and thinning process, till n=0, obtain n+1 bit-planes; At last, utilize binary arithmetic coding further to improve compression efficiency.
Step 102: according to the rate distortion characteristic of each subband compressed bit stream, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, the code check that is assigned with according to each subband blocks the code stream of each subband;
Be specially: from the compressed bit stream of the encoded gained of each subband, utilize existing computational methods to obtain the rate distortion characteristic of each subband; According to the rate distortion characteristic of each subband correspondence, target bit rate is distributed between the compressed bit stream of each subband.Here, how described target bit rate distributes can be the optimum point of cut-off of determining each subband correspondence according to the rate distortion characteristic of each subband, make each subband block the code check sum less than target bit rate the time, it is minimum that the coding distortion rate of entire image reaches, and also just reached described optimum allocation.Like this, the described process that target bit rate is distributed between the compressed bit stream of each subband is just blocked each subband at the optimum point of cut-off place of each subband, and the code check that blocks that blocks the back gained is the code check that each subband is assigned with.
Wherein, because each subband has a plurality of point of cut-offs, therefore need find the optimum point of cut-off of each subband; Described rate distortion characteristic is meant that subband is blocked the pairing distortion rate in back by a certain point of cut-off, and subband is cut apart all available existing computational methods of the pairing distortion rate in back by arbitrary point of cut-off and obtained; The described code check that blocks is meant code check after each subband is blocked by point of cut-off.The process that above-mentioned target bit rate distributes can be expressed as follows with formula (1):
min D = Σ i = 1 N D i ( Z i ) s.t. L = Σ i = 1 N L i ( Z i ) ≤ L max - - - ( 1 )
Wherein, described D represents the coding distortion rate of entire image, what described L represented each subband blocks the code check sum, the number of described N presentation video gained subband after wavelet decomposition, pass between the progression p of N and wavelet decomposition is: N=3*p+1, for example: image obtains 10 subbands, L after three grades of wavelet decomposition MaxBe target bit rate,
Figure GSA00000023029300073
The point of cut-off of representing i subband respectively is z i, i=1,2 ..., during N, the pairing code check of this subband promptly blocks code check and distortion rate.
Can find that formula (1) is an optimization problem under the constraints, utilize method of Lagrange multipliers then the optimization problem under the constraints can be converted into unconfined optimization problem, so, introduce Lagrange multiplier λ and obtain:
D + λ · L = Σ i = 1 N ( D i ( Z i ) + λ · L i ( Z i ) ) - - - ( 2 )
If: J = D i ( Z i ) + λ · L i ( Z i ) - - - ( 3 )
L ′ ( λ ) = Σ i = 1 N L i ( Z i * ) - - - ( 4 )
According to method of Lagrange multipliers, under the situation of given λ, make formula (3) reach minimum, also satisfy the point of cut-off z of formula (4) simultaneously i *, i=1,2 ..., N necessarily also is the optimal solution of formula (1), like this, searches for the optimum point of cut-off z of each subband i *, i=1,2 ..., the process of N, just be reduced to find the solution L ' (λ)=L Maxλ * value.Because z iCorresponding is some discrete sampled points, so adopts dichotomy to search for λ among the present invention *Value.Under the situation of given λ, find the solution the z that makes formula (4) minimum i *, i=1,2 ..., the process of N can independently be carried out in each subband.Therefore, to each wavelet sub-band, the present invention proposes following fast search algorithm:
Initialization: establish z Opt=0;
for?z i=1,2,…,z i
If ΔL = L i ( Z i ) - L i ( Z opt ) , ΔD = D i ( Z i ) - D i ( Z opt )
If Δ D/ Δ L>λ, z Opt=z i
Finish: z i * = z opt *
Above-mentioned algorithm is: the optimum value of establishing point of cut-off is z Opt, and make z OptInitial value be 0, afterwards all point of cut-offs of i subband are carried out successively the comparison operation of Δ D/ Δ L and λ, when determining Δ D/ Δ L>λ, resulting z OptValue then be the optimum value of point of cut-off, promptly obtain the optimum point of cut-off z that is asked i *, afterwards at optimum point of cut-off z i *The code stream of each subband is blocked in the position, and the code stream of each subband through blocking forms last output code flow.
Fig. 2 is the structural representation of the output code flow of gained behind the wavelet coding of the present invention, as shown in Figure 2, the subband of different progression is arranged in order output, wherein, described LH1, HL1 and HH1 are used to identify corresponding subband after the one-level wavelet decomposition, LH3, HL3 and HH3 are used to identify subband corresponding after three grades of wavelet decomposition or the like, B among Fig. 2 LLBlocked the corresponding compressed bit stream in back Deng each subband of expression, so,, then can be decoded to compressed bit stream selectively as will the compressed bit stream among Fig. 2 being decoded.The resolution of supposing original image is 512 * 512, if select the compressed bit stream of LH3, HL3 and three subbands of HH3 to decode, the resolution of gained image then is 128 * 128 so; If the compressed bit stream of LH3, HL3, HH3, LH2, HL2 and six subbands of HH2 is decoded, the resolution of gained image then is 256 * 256 so.Certainly, also can obtain resolution and be 512 * 512 image, and relate to correlation between each wavelet coefficient in the existing Wavelet image cataloged procedure, therefore, after prior art is decoded to compressed bit stream, the resolution image identical can only be obtained, and other multiple image different can not be obtained with original image resolution with original image resolution.
In the said process, the described process that the wavelet coefficient of each subband is encoded fully of step 101, as shown in Figure 3, the cataloged procedure of each subband includes the structure of pyramid structure, initialization, classification, refinement and quantizes four parts, and the specific implementation flow process is as follows:
Step 301: for current sub makes up corresponding pyramid structure;
Be specially: adopt pyramid structure that all wavelet coefficients of current sub correspondence are classified, Figure 4 shows that the construction method of a two-stage pyramid structure, all wavelet coefficients of the corresponding current sub of pyramidal all the 0th grade of coefficients; Pyramidal the 1st grade of coefficient is to obtain by 4 the 0th grade of coefficient maximizings to the pyramid relevant position, and its method by that analogy, obtains pyramid structure shown in Figure 4 as shown in Equation (5).
c l(i,j)=max{c l-1(2i,2j),c l-1(2i,2j+1),c l-1(2i+1,2j),c l-1(2i+1,2j+1)} (5)
Wherein, l is an integer, and maximum is max{log 2 W, log 2 H, W and H are respectively the width and the height of current sub; Described c l(i, j) expression pyramid l level (i, the coefficient value of j) locating, when l=0, c l(i, j) corresponding is the wavelet coefficient of current sub.
Step 302: for current sub is provided with first pixel tabulation and Section Point tabulation, and initialization be provided with Section Point tabulation and first pixel in tabulating content and the initial value size of n;
Be specially: for current sub is provided with the tabulation of one first pixel and l Section Point tabulation, LIN[l] represent that the Section Point of the l level pyramid coefficient correspondence of current sub tabulates.Make that all coefficients are inessential coefficient in the pyramid structure, the coefficient in each grade pyramid structure belongs to corresponding LIN k[l], this moment, the tabulation of first pixel was an empty set then, did not have coefficient.The described initialization setting of this step can be expressed as formula (6), (7) and (8):
LIN[l]={ all coefficients of l level gold tower correspondence }; (6)
LSP=φ; (7)
Figure GSA00000023029300101
Step 303: all wavelet coefficients in the constructed pyramid structure are classified;
The black round dot is attached most importance to and is wanted node as shown in Figure 4, just significant coefficient described in the present invention.Here, the importance of coefficient is defined as formula (9):
Figure GSA00000023029300102
S n(i j) is the function of n, S n(i j)=1 o'clock, shows S this moment n(i, j) Dui Ying wavelet coefficient is a significant coefficient; S n(i j)=0 o'clock, shows S this moment n(i, j) Dui Ying wavelet coefficient is inessential coefficient.
Concrete assorting process is:
1) from the pyramid top, promptly l level pyramid begins, and its wavelet coefficient is carried out importance judge;
2) if S n(i, value j) is 0, then with the c of correspondence l(i j) stays LIN[l] in;
3) if S n(i, value j) is 1,
If l>0 o'clock, LIN[l] in corresponding wavelet coefficient be c l(i, j)-2 n
If during l=0, with the wavelet coefficient c of correspondence 0(i j) joins in first pixel tabulation, and LIN[l] in corresponding wavelet coefficient be c l(i, j)-2 n
4) if the wavelet coefficient all classification in the pyramid of prime finishes, then make l=l-1, promptly enter l-1 level pyramid, repeat above-mentioned 2) step, until l=0.
5) finally obtain the first pixel tabulation of current sub when the bit-planes coefficient is n.
Step 304: the wavelet coefficient in first pixel tabulation that classification is obtained is carried out Refinement operation;
Be specially: n MSB of each wavelet coefficient in first pixel tabulation when exporting current bit-planes coefficient and being n, empty first pixel then and tabulate, make it to be empty set.The output procedure of n MSB of described each coefficient is a prior art, no longer describes in detail herein.
Among the present invention, from the bit-planes coefficient Classification and thinning process are carried out in beginning, finish till n=0.
Step 305: judge the magnitude relationship of current bit-planes coefficient n and 1,, and return step 303 if n more than or equal to 1, then makes n=n-1; If n is less than 1, then execution in step 306;
This step is carries out quantization operation to bit floor coefficient n.
Wherein, described n shows that more than or equal to 1 o'clock the cataloged procedure to current sub does not also finish; Described n shows that less than 1 o'clock the cataloged procedure to current sub finishes, is 0 to finish because the cataloged procedure of each subband is got from n that maximum begins to n.
Step 306: the complete cataloged procedure of the wavelet coefficient of current sub finishes.
Here, after the complete cataloged procedure of current sub finishes, also need the code stream of final output is carried out conventional binary arithmetic entropy coding, so that further improve compression efficiency.
For realizing said method, the present invention also provides a kind of Wavelet image apparatus for encoding, and as shown in Figure 5, this device comprises: coding module and truncation module; Wherein,
Described coding module, be used to utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, again classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband, and the compressed bit stream of each subband of gained is transferred to truncation module;
Described truncation module, the rate distortion characteristic of each the subband compressed bit stream that is used for sending out according to coding module, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, and the code check that is assigned with according to each subband blocks the code stream of each subband, obtains the code stream that Wavelet image is encoded.
Described coding module utilizes pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, and more classified wavelet coefficient is carried out the corresponding codes operation, is specially:
Coding module earlier makes up corresponding pyramid structure for current sub, and for current sub is provided with the tabulation of first pixel and Section Point is tabulated, and initialization is provided with content during first pixel is tabulated and Section Point is tabulated and the initial value size of bit-planes coefficient n; All wavelet coefficients in the constructed pyramid structure are classified, and the wavelet coefficient in first pixel that classification the is obtained tabulation is carried out Refinement operation;
Also be used to judge the magnitude relationship of current bit-planes coefficient n and 1, determine that n more than or equal to 1 o'clock, makes n=n-1, and begin to continue to carry out follow-up cataloged procedure from described execution sort operation step; Determine n less than 1 o'clock, finish the complete cataloged procedure of wavelet coefficient of current sub.
Described coding module further comprises: pyramid construction module, initialization module, sort module, refinement module and quantization modules; Wherein,
Described pyramid construction module is used to the subband of present encoding to make up corresponding pyramid structure, and will make up the result and send to initialization module;
Described initialization module, be used to current sub that tabulation of first pixel and Section Point tabulation are set, and initialization be provided with Section Point tabulation and first pixel in tabulating content and the initial value size of bit-planes coefficient n, and the result that will be provided with sends to sort module;
Described sort module, the initial value that is used for being sent out with initialization module serve as that foundation is classified to all wavelet coefficients of the pyramid structure of current sub correspondence, and classification results is sent to the refinement module; After also being used to receive the current bit-planes coefficient n that quantization modules sends out, serve as according to continuing to carry out sort operation with new n;
Described refinement module, the wavelet coefficient that is used for first pixel tabulation that classification obtains to sort module is carried out Refinement operation, and triggers quantization modules after executing current Refinement operation; Be used to also to receive after the notice of quantization modules that the refinement result with all wavelet coefficients of current sub sends to truncation module;
Described quantization modules is used for judging the magnitude relationship of sort module current bit-planes coefficient n and 1, if n more than or equal to 1, then n subtracts 1, the n after will successively decreasing sends to sort module, and to trigger sort module serve as according to continuing to carry out sort operation with new n;
Described quantization modules also was used for determining the current bit-planes coefficient of sort module n less than 1 o'clock, and notice refinement module sends to truncation module with the refinement result of all wavelet coefficients of current sub.
The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention, all any modifications of being done within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. the coding method of a Wavelet image, its characteristic are that this method comprises:
Utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, more classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband; According to the rate distortion characteristic of each subband compressed bit stream, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, the code check that is assigned with according to each subband blocks the code stream of each subband, obtains the code stream that Wavelet image is encoded.
2. the coding method of Wavelet image according to claim 1, its characteristic is, the described pyramid structure of utilizing is classified to the wavelet coefficient through each subband correspondence of wavelet decomposition gained, more classified wavelet coefficient is carried out the process of corresponding codes operation, is specially:
For current sub makes up corresponding pyramid structure;
For current sub is provided with first pixel tabulation and Section Point tabulation, and initialization be provided with the tabulation of first pixel and Section Point in tabulating content and the initial value size of bit-planes coefficient n;
All wavelet coefficients in the constructed pyramid structure are classified;
Wavelet coefficient in first pixel tabulation that classification is obtained is carried out Refinement operation;
Judge the magnitude relationship of current bit-planes coefficient n and 1,, and begin to continue to carry out follow-up cataloged procedure from described execution sort operation step if n more than or equal to 1, then makes n=n-1; If n, then finishes the complete cataloged procedure of wavelet coefficient of current sub less than 1.
3. the coding method of Wavelet image according to claim 2, its characteristic are that the pyramid structure that described current sub makes up is made up of the l level, and the maximum of described l is max{log1 2 W, log 2 H.
4. the coding method of Wavelet image according to claim 3, its characteristic are that described W and H are respectively the width and the height of current sub.
5. according to the coding method of claim 3 or 4 described Wavelet images, its characteristic is, and is described when for current sub tabulation of first pixel and Section Point tabulation being set, and one first pixel tabulation and l Section Point is set tabulates.
6. according to the coding method of each described Wavelet image of claim 1 to 4, its characteristic is, describedly distributes target bit rate between the compressed bit stream of each subband, and the process of blocking code stream according to the code check that is assigned with is:
Determine to block the optimum point of cut-off of each subband correspondence according to the rate distortion characteristic of each subband according to the compressed bit stream of described optimum point of cut-off to each subband.
7. the code device of a Wavelet image, its characteristic is that this device comprises: coding module and truncation module; Wherein,
Described coding module, be used to utilize pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, again classified wavelet coefficient is carried out the corresponding codes operation, obtain the compressed bit stream of each subband, and the compressed bit stream of each subband of gained is transferred to truncation module;
Described truncation module, the rate distortion characteristic of each the subband compressed bit stream that is used for sending out according to coding module, target bit rate is carried out optimum allocation between the compressed bit stream of each subband, and the code check that is assigned with according to each subband blocks the code stream of each subband, obtains the code stream that Wavelet image is encoded.
8. the code device of Wavelet image according to claim 7, its characteristic is, described coding module utilizes pyramid structure that the wavelet coefficient through each subband correspondence of wavelet decomposition gained is classified, and more classified wavelet coefficient is carried out the corresponding codes operation, is specially:
Coding module earlier makes up corresponding pyramid structure for current sub, and for current sub is provided with the tabulation of first pixel and Section Point is tabulated, and initialization is provided with content during first pixel is tabulated and Section Point is tabulated and the initial value size of bit-planes coefficient n; All wavelet coefficients in the constructed pyramid structure are classified, and the wavelet coefficient in first pixel that classification the is obtained tabulation is carried out Refinement operation;
Also be used to judge the magnitude relationship of current bit-planes coefficient n and 1, determine that n more than or equal to 1 o'clock, makes n=n-1, and begin to continue to carry out follow-up cataloged procedure from described execution sort operation step; Determine n less than 1 o'clock, finish the complete cataloged procedure of wavelet coefficient of current sub.
9. the code device of Wavelet image according to claim 8, its characteristic is that described coding module further comprises: pyramid construction module, initialization module, sort module, refinement module and quantization modules; Wherein,
Described pyramid construction module is used to the subband of present encoding to make up corresponding pyramid structure, and will make up the result and send to initialization module;
Described initialization module, be used to current sub that tabulation of first pixel and Section Point tabulation are set, and initialization be provided with Section Point tabulation and first pixel in tabulating content and the initial value size of bit-planes coefficient n, and the result that will be provided with sends to sort module;
Described sort module, the initial value that is used for being sent out with initialization module serve as that foundation is classified to all wavelet coefficients of the pyramid structure of current sub correspondence, and classification results is sent to the refinement module; After also being used to receive the current bit-planes coefficient n that quantization modules sends out, serve as according to continuing to carry out sort operation with new n;
Described refinement module, the wavelet coefficient that is used for first pixel tabulation that classification obtains to sort module is carried out Refinement operation, and triggers quantization modules after executing current Refinement operation; Be used to also to receive after the notice of quantization modules that the refinement result with all wavelet coefficients of current sub sends to truncation module;
Described quantization modules is used for judging the magnitude relationship of sort module current bit-planes coefficient n and 1, if n more than or equal to 1, then n subtracts 1, the n after will successively decreasing sends to sort module, and to trigger sort module serve as according to continuing to carry out sort operation with new n;
Described quantization modules also was used for determining the current bit-planes coefficient of sort module n less than 1 o'clock, and notice refinement module sends to truncation module with the refinement result of all wavelet coefficients of current sub.
10. according to the code device of claim 7 or 8 described Wavelet images, its characteristic is, described truncation module is distributed target bit rate between the compressed bit stream of each subband, block being operating as of code stream according to the code check that is assigned with: determine to block the optimum point of cut-off of each subband correspondence according to the compressed bit stream of described optimum point of cut-off to each subband according to the rate distortion characteristic of each subband.
CN 201010116325 2010-02-10 2010-02-10 Method and device for encoding wavelet image Active CN102148993B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010116325 CN102148993B (en) 2010-02-10 2010-02-10 Method and device for encoding wavelet image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010116325 CN102148993B (en) 2010-02-10 2010-02-10 Method and device for encoding wavelet image

Publications (2)

Publication Number Publication Date
CN102148993A true CN102148993A (en) 2011-08-10
CN102148993B CN102148993B (en) 2013-03-27

Family

ID=44422953

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010116325 Active CN102148993B (en) 2010-02-10 2010-02-10 Method and device for encoding wavelet image

Country Status (1)

Country Link
CN (1) CN102148993B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107257473A (en) * 2017-07-18 2017-10-17 西安电子科技大学 A kind of efficient image compression algorithm
CN112532980A (en) * 2020-11-03 2021-03-19 宁波高新区思搏科技有限公司 Medical ultrasonic image processing method

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6137915A (en) * 1998-08-20 2000-10-24 Sarnoff Corporation Apparatus and method for error concealment for hierarchical subband coding and decoding
US20020080408A1 (en) * 1999-12-17 2002-06-27 Budge Scott E. Method for image coding by rate-distortion adaptive zerotree-based residual vector quantization and system for effecting same

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6137915A (en) * 1998-08-20 2000-10-24 Sarnoff Corporation Apparatus and method for error concealment for hierarchical subband coding and decoding
US20020080408A1 (en) * 1999-12-17 2002-06-27 Budge Scott E. Method for image coding by rate-distortion adaptive zerotree-based residual vector quantization and system for effecting same

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
YA-QIN ZHANG,ETL: "Motion-Compensated Wavelet Transform Coding for Color Video Compression", 《IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY》, vol. 2, no. 3, 30 September 1992 (1992-09-30), XP000299259, DOI: doi:10.1109/76.157160 *
丁素英: "基于小波变换的Laplacian金字塔图像数据压缩", 《潍坊学院学报》, vol. 9, no. 4, 31 July 2009 (2009-07-31) *
党百振,等: "数字图象的金字塔型预测编码", 《桂林电子工业学院学报》, vol. 20, no. 1, 31 March 2000 (2000-03-31) *
吴宗泽,等: "JPEG2000实时截断码率控制新算法及其VLSI结构设计", 《电子学报》, no. 8, 31 August 2006 (2006-08-31) *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107257473A (en) * 2017-07-18 2017-10-17 西安电子科技大学 A kind of efficient image compression algorithm
CN107257473B (en) * 2017-07-18 2020-01-10 西安电子科技大学 Efficient image compression algorithm
CN112532980A (en) * 2020-11-03 2021-03-19 宁波高新区思搏科技有限公司 Medical ultrasonic image processing method
CN112532980B (en) * 2020-11-03 2023-03-10 宁波高新区思搏科技有限公司 Medical ultrasonic image processing method

Also Published As

Publication number Publication date
CN102148993B (en) 2013-03-27

Similar Documents

Publication Publication Date Title
US8428379B2 (en) Method for distributed source coding of wavelet coefficients in zerotrees
CN103581691B (en) A kind of towards sparse coefficient efficiently can parallel image coding method
KR19990067703A (en) Apparatus and Method for Encoding Zerotrees Generated by Wavelet-based Coding Techniques
CN101984666A (en) Image lossless compression and decompression method based on lifting wavelet transform
CN102036073B (en) Method for encoding and decoding JPEG2000 image based on vision potential attention target area
CN101132530A (en) Method for implementing built-in image compression based on run-length coding
Aulí-Llinàs et al. Lossy-to-lossless 3D image coding through prior coefficient lookup tables
CN102006483B (en) Video coding and decoding method and device
CN101783939B (en) Picture coding method based on human eye visual characteristic
CN101540908B (en) Video coding processing method and device
CN102148993B (en) Method and device for encoding wavelet image
Liu et al. Quadtree coding with adaptive scanning order for space-borne image compression
CN104270642B (en) A kind of lossless compression method for high spectrum image chosen based on RKLT and principal component
CN102790882A (en) Encoding method of remote sensing image
Al-Janabi Low memory set-partitioning in hierarchical trees image compression algorithm
Zhu et al. An improved SPIHT algorithm based on wavelet coefficient blocks for image coding
CN102316324B (en) Image coding prediction method based on local minimum entropy
Demaret et al. Scattered data coding in digital image compression
CN102769744A (en) Satellite image encoding method
Yang Multiple description wavelet-based image coding using iterated function system
CN103618910B (en) Image conversion method and device
Wang et al. Coefficient statistic based modified spiht image compression algorithm
Liu Research on image compression algorithm based on SPHIT
Liang et al. A rate control algorithm in JPEG2000 based on code content prediction
WO2010064569A1 (en) Moving image data compressing method

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant