CN103077543A - Embedded zerotree coding method on basis of inverse Loop subdivision - Google Patents
Embedded zerotree coding method on basis of inverse Loop subdivision Download PDFInfo
- Publication number
- CN103077543A CN103077543A CN2012105923733A CN201210592373A CN103077543A CN 103077543 A CN103077543 A CN 103077543A CN 2012105923733 A CN2012105923733 A CN 2012105923733A CN 201210592373 A CN201210592373 A CN 201210592373A CN 103077543 A CN103077543 A CN 103077543A
- Authority
- CN
- China
- Prior art keywords
- play amount
- grid
- tree
- coefficient
- embedded
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention relates to an embedded zerotree coding method on the basis of inverse Loop subdivision, which comprises the following steps of: (1) constructing an offset wavelet tree, i.e. forming an offset d by obtaining difference of an actual vertex O and a forecast vertex O' and simultaneously, establishing a corresponding quad-tree relation of edges in each mesh Mj-1 and four edges in a thinner layer mesh Mj, wherein the offset d is generated on a corresponding edge of a delete vertex reconstruction mesh; (2) carrying out offset wavelet processing and selecting a threshold value; and (3) encoding and carrying out code-stream ordering, wherein important information is preferentially encoded by adopting an embedded encoding mode on the basis of a three-dimensional graph triangular mesh and a result obtained after compression is placed at an initial part of a code stream. The invention provides the embedded zerotree coding method on the basis of inverse Loop subdivision, through which compression efficiency is greatly improved.
Description
Technical field
The present invention relates to technical field of digital media, especially a kind of coding method of Computerized three-dimensional figure.
Background technology
The dimension compression of images has been widely used in application, and wherein Embedded Zerotree Wavelet is one of effective method.The image Embedded Zerotree Wavelet comprises three processes:
(1) zero tree prediction is with zero tree construction coding significance map picture, successive approximation to quantification.Image through wavelet transformation forms a tree structure from low to high by its frequency band, and tree root is the node of lowest frequency subband, and it has three children to lay respectively at the relevant position of three infra-low frequency subbands, sees Fig. 1 upper left corner.The node of all the other subbands (high-frequency sub-band except) have four children be positioned at the relevant position of high one-level subband (because high-frequency sub-band resolution increases, thus a low frequency sub-band node to four high-frequency sub-band nodes should be arranged, i.e. 2 * 2 adjacent matrixes).It is 4 tree that such three grades of wavelet decomposition have just formed the degree of depth.Piece image is correlated with at the conversion coefficient of the same position of different sub-band after wavelet decomposition, and this correlativity has formed zero tree construction.
(2) small echo is processed: all wavelet coefficients are divided into following three kinds of situations: zerotree root (ZTR); Isolated zero (IZ); Significant coefficient.For the needs of encoding are divided into positive significant coefficient (POS) to significant coefficient again and coefficient (NEG) is wanted in heavy burden, the judgement flow process of wavelet coefficient as shown in Figure 2.
(3) EZW coding: the EZW coding is finished by successive approximation to quantification, is exactly will be by use threshold series T one by one
0, T
1, T
2..., T
N-1Decide significant coefficient, wherein T
i=T
I-1/ 2, and show threshold value T
0Satisfy condition: all wavelet coefficients have a
i＜2T
0
Embedded encoded is to important image information priority encoding, and the result after the compression is placed on the initial part of code stream, then places successively other parts of code stream according to the significance level of information, like this, the code stream of low code check is embedded in the code stream of high code check.Embedded bitstream is supported progressive transmission, can stop in the arbitrfary point coding, can strictly satisfy target bit rate or the requirement of target distortion degree.Adopt embedded bitstream to control accurately code check, when coding distortion or encoder bit rate reach requirement, just can stop at any time cataloged procedure.Therefore be adapted to the progressive transmission of view data on the internet.
Along with the increase in demand of three-dimensional picture, people have higher requirement to precision and the details of three-dimensional picture, and this has also caused scale and the complexity sharp increase of 3 D graphic data.Huge data volume has all proposed huge challenge to existing server and intelligent terminal, and simultaneously, the restriction of the network bandwidth has seriously hindered the propagation of this three-dimensional picture, so need to propose storage and the Internet Transmission that a kind of method can be convenient to three-dimensional picture.
Summary of the invention
In order to overcome the lower deficiency of compression efficiency of existing 3 D graphic data compression method, the invention provides a kind of Embedded Zerotree Wavelet method based on contrary Loop segmentation that significantly improves compression efficiency.
The technical solution adopted for the present invention to solve the technical problems is:
A kind of Embedded Zerotree Wavelet method based on contrary Loop segmentation, described coding method may further comprise the steps:
(1) make up the side-play amount wavelet tree:
When triangle gridding is done contrary Loop segmentation operation, per four triangles are simplified to a triangle, utilize predictor predicts summit O ', and actual summit O has just formed side-play amount d with the difference of predicting summit O ', side-play amount d is that the corresponding sides of rebuilding grid on the deletion summit produce, simultaneously each grid M
J-1In limit and thinner level grid M
jIn four limits set up corresponding quaternary tree relation;
For Progressive Mesh M
0→ M
1→ ... → M
N-1→ M
n, base net lattice M
0In triangle edges consisted of the root of quaternary tree, the height of tree namely is the level of Progressive Mesh, (n-1) the side-play amount d of layer
N-1Be the leaf of lowermost layer, except (n-1) layer, every layer has 4 nodes;
(2) the side-play amount small echo is processed and is selected threshold value:
The side-play amount small echo d of base net lattice
0Be defined as zerotree root, the side-play amount small echo d of thin level
N-1Be defined as isolated zero, the importance of grid side-play amount coefficient from coarse to fine reduces successively, and the coefficient of coarse grid is significant coefficient, and simultaneously, significant coefficient is divided into positive significant coefficient and coefficient is wanted in heavy burden;
In three side-play amounts of the triangle gridding of same level during all less than threshold value, it is unessential then looking this deviation ratio, and supposition: if deviation ratio is inessential in coarse grid, then its child is also inessential in refined net, and this node is set together with all being made as of child is zero; Zero tree needn't be transmitted in network, otherwise side-play amount is transmitted together with its position and symbolic information;
(3) coding and code stream tissue:
Embedded encoded with important information priority coding based on the three-dimensional picture triangle gridding, and the result after will compressing is placed on the initial part of code stream, its coding step is as follows:
(3.1) initialization: selected threshold epsilon
0
(3.2) set up two tabulations: master meter is all wavelet coefficients, and subtabulation is empty;
(3.3) first pass: whether the master meter coefficient is important, in the important adding subtabulation, and then analog value zero clearing;
(3.4) second times scanning: unessential coefficient is judged whether it is leaf, if it is do not encode;
(3.5)ε
0=ε
0/2；
(3.6) master meter is not empty, returns (3.1);
(3.7) finish.
Further, in the described step (2), initial threshold ε
0Estimation:
Threshold epsilon
0Value can decide according to the requirement of compression efficiency, to different levels j (0≤j≤n-1), the threshold value ε that successively decreases
J+1=ε
j/ 2.
Further again, in the described step (2), the classification zerotree image is adopted in the compression of progressive network.
Technical conceive of the present invention is: according to base net network in the Progressive Mesh and skew the relationship between quantities, take every limit of base net lattice as root, set up the quad-tree structure with thinner level Grid Edge, because the limit in the grid is corresponding one by one with side-play amount, so also just set up the quaternary tree relation of side-play amount small echo and next level side-play amount small echo.The little wave number of the side-play amount of base net lattice is defined as zerotree root, and the side-play amount small echo of thin level is defined as isolated zero, and the importance of grid side-play amount coefficient from coarse to fine reduces successively.
The Embedded Zerotree Wavelet method of triangle gridding is proposed.Select threshold value, during all less than threshold value, it is unessential then looking this side-play amount, if side-play amount is inessential in coarse grid at three deviation ratios of the triangle gridding of same level, then its side-play amount subtree is also inessential in refined net, and this node all is made as zero tree together with subtree.Zero tree needn't be transmitted in network, otherwise side-play amount is transmitted together with its positional information.Through behind the zerotree image, owing to side-play amount small echo majority need not coding for null value or near null value, so just greatly reduced the data volume of side-play amount small echo.
Proposition is based on the code stream method for organizing of triangle gridding, and with important information priority coding, and the result after will compressing is placed on the initial part of code stream, and Progressive Mesh is stored by following form: M
0→ d
0→ d
1→ ... → d
N-2→ d
N-1, and transmit in grid according to order.
Usually can simplify base net lattice M of generation to it through contrary Loop segmentation by the triangle gridding that subdivide technology generates
0With a series of side-play amount d
0→ d
1→ ... → d
N-2→ d
N-1The Progressive Mesh M that forms
0→ M
1→ ... → M
N-1→ M
nM of every simplification
j→ M
J-1(n 〉=j＞1), the summit of grid and leg-of-mutton number respectively reduce 75%, but because the summit number that reduces is identical with the side-play amount number of increase, so the compression effectiveness of grid is unsatisfactory.Progressive Mesh after three times are simplified with simplify before mesh compression than about 45%, compression efficiency is lower.If can provide a mechanism of eliminating redundant geological information to progressive network, the compression efficiency of grid can improve significantly.
Beneficial effect of the present invention is mainly manifested in: the method realizes the priority encoding of three-dimensional picture important information, and the data behind the coding are placed on the initial part of code stream, other parts of then placing successively code stream according to the significance level of data.The figure code stream of low resolution is embedded in the high-resolution code stream.Embedded bitstream is supported progressive transmission, can stop in the arbitrfary point coding, can strictly satisfy target resolution or the requirement of target distortion degree.When coding distortion or code distinguishability reach requirement, just can stop at any time cataloged procedure like this.Therefore be adapted to the progressive transmission of 3 D graphic data on the internet and the use of different resolution user terminal.The key issue that this invention will solve is the tree construction that how to make up based on the three-dimensional picture triangle gridding.
Description of drawings
Fig. 1 is the synoptic diagram in the space-frequency tree structure of wavelet transformation.
Fig. 2 is the decision flow chart of wavelet coefficient.
Fig. 3 is contrary segmentation rough schematic view, wherein, and after (a) simplify on (b) prediction summit (c) before the simplification.
Fig. 4 is M
J-1With M
jThe tree construction synoptic diagram that grid forms.
Embodiment
The invention will be further described below in conjunction with accompanying drawing.
With reference to Fig. 1～Fig. 4, a kind of Embedded Zerotree Wavelet method based on contrary Loop segmentation, the Progressive Mesh that the Loop subdivision curved surface may decompose generate Triangular Mesh Simplification by inverse transformation (sees patent for details: a kind of Progressive Mesh generation method based on contrary Loop segmentation, the patent No.: ZL2006101241528), contain the summit in its base net lattice and triangle reduces in large quantities, but limit of every deletion just produces an offset information, so offset information has occupied very large storage space.Owing to do not have very large sudden change between abutment points in its triangle gridding of model that is simplified, has good correlativity between abutment points, so often numerical value is all less for the side-play amount that generates through the forecasting process simplified, majority has near the characteristics the null value of concentrating on, and the characteristic that successively decays.Can utilize like this side-play amount to carry out Embedded Zerotree Wavelet, its cataloged procedure is as follows:
(1) makes up the side-play amount wavelet tree
When triangle gridding was done contrary Loop segmentation operation, per four triangles were simplified to a triangle, grid M
jBe reduced to M from Fig. 3 (a)
J-1During Fig. 3 (c), pass through forecasting process, utilize the process of the normal Fig. 3 (b) of predictor predicts summit O ', and actual summit O has just formed side-play amount d with the difference of predicting summit O '.And side-play amount is to produce in the corresponding sides of deleting summit reconstruction grid, simultaneously each grid M
J-1In limit and thinner level grid M
jIn four limits may set up corresponding quaternary tree relation, such as Fig. 4, illustrate respectively in the coarse grid three limits with arrow, real round dot, short side point in the position than refined net correspondence four edges.
Because side-play amount is corresponding one by one with the limit of triangular network, has so also just set up the side-play amount quaternary tree.As seen, for Progressive Mesh M
0→ M
1→ ... → M
N-1→ M
n, base net lattice M
0In triangle edges consisted of the root of quaternary tree, the height of tree namely is the level of Progressive Mesh, (n-1) the side-play amount d of layer
N-1Be the leaf of lowermost layer, except (n-1), every layer has 4 nodes.Because have correlativity between two levels, the data of side-play amount are not too large, process so can be considered small echo.
(2) the side-play amount small echo is processed and is selected threshold value
The side-play amount small echo d of base net lattice
0Be defined as zerotree root, the side-play amount small echo d of thin level
N-1Be defined as isolated zero, the importance of grid side-play amount coefficient from coarse to fine reduces successively, and the coefficient of coarse grid is significant coefficient, and simultaneously, significant coefficient is divided into positive significant coefficient and coefficient is wanted in heavy burden.In addition, during all less than threshold value, it is unessential then looking this deviation ratio, and supposition: if deviation ratio is inessential in coarse grid in three side-play amounts of the triangle gridding of same level, then its child is also inessential in refined net, and this node is set together with all being made as of child is zero.Zero tree needn't be transmitted in network, otherwise side-play amount is transmitted together with its position and symbolic information.
Initial threshold ε
0Estimation:
(for all i).Threshold epsilon
0Value can decide according to the requirement of compression efficiency, to different levels j (0≤j≤n-1), the threshold value ε that successively decreases
J+1=ε
j/ 2.
The classification zerotree image is adopted in the compression of progressive network, and its compression algorithm is: each triangle in the triangle gridding, call with minor function.
Wherein: T is the base net lattice, the level of j for simplifying, ε
0Be initial threshold.
For side-play amount d
i=(a
I1, a
I2, a
I3), ε
0Value can come according to the requirement of compression ratio fixed, to different levels j (0≤j≤n-1), the threshold value ε that successively decreases
J+1=ε
j/ 2.Only have three components when vector during all less than or equal to threshold value, just ignore this side-play amount.When threshold value was got 0 value, triangle gridding was not done zerotree image.
(3) coding and code stream tissue
Embedded encoded with important information priority coding based on the three-dimensional picture triangle gridding, and the result after will compressing is placed on the initial part of code stream, its coding step is as follows:
(3.1) initialization: selected ε
0
(3.2) set up two tabulations: master meter is all wavelet coefficients, and subtabulation is empty;
(3.3) first pass: whether the master meter coefficient is important, in the important adding subtabulation, and then analog value zero clearing;
(3.4) second times scanning: unessential coefficient is judged whether it is leaf, be: do not encode;
(3.5)ε
0=ε
0/2；
(3.6) master meter is not empty, returns (3.1);
(3.7) finish.
Through behind the zerotree image, owing to side-play amount small echo majority need not coding for null value or near null value, so just greatly reduced the data volume of side-play amount small echo.
Progressive Mesh is stored by following form: M
0→ d
0→ d
1→ ... → d
N-2→ d
N-1, and transmit in grid according to order.
Claims (3)
1. Embedded Zerotree Wavelet method based on contrary Loop segmentation, it is characterized in that: described coding method may further comprise the steps:
(1) make up the side-play amount wavelet tree:
When triangle gridding is done contrary Loop segmentation operation, per four triangles are simplified to a triangle, utilize predictor predicts summit O ', and actual summit O has just formed side-play amount d with the difference of predicting summit O ', side-play amount d is that the corresponding sides of rebuilding grid on the deletion summit produce, simultaneously each grid M
J-1In limit and thinner level grid M
jIn four limits set up corresponding quaternary tree relation;
For Progressive Mesh M
0→ M
1→ ... → M
N-1→ M
n, base net lattice M
0In triangle edges consisted of the root of quaternary tree, the height of tree namely is the level of Progressive Mesh, (n-1) the side-play amount d of layer
N-1Be the leaf of lowermost layer, except (n-1) layer, every layer has 4 nodes;
(2) the side-play amount small echo is processed and is selected threshold value:
The side-play amount small echo d of base net lattice
0Be defined as zerotree root, the side-play amount small echo d of thin level
N-1Be defined as isolated zero, the importance of grid deviation ratio from coarse to fine reduces successively, and the coefficient of coarse grid is significant coefficient, and simultaneously, significant coefficient is divided into positive significant coefficient and coefficient is wanted in heavy burden;
In three side-play amounts of the triangle gridding of same level during all less than threshold value, it is unessential then looking this deviation ratio, and supposition: if deviation ratio is inessential in coarse grid, then its child is also inessential in refined net, and this node is set together with all being made as of child is zero; Zero tree needn't be transmitted in network, otherwise side-play amount is transmitted together with its position and symbolic information;
(3) coding and code stream tissue:
Embedded encoded with important information priority coding based on the three-dimensional picture triangle gridding, and the result after will compressing is placed on the initial part of code stream, its coding step is as follows:
(3.1) initialization: selected threshold epsilon
0,
(3.2) set up two tabulations: master meter is all wavelet coefficients, and subtabulation is empty;
(3.3) first pass: whether the master meter coefficient is important, in the important adding subtabulation, and then analog value zero clearing;
(3.4) second times scanning: unessential coefficient is judged whether it is leaf, if it is do not encode;
(3.5)ε
0=ε
0/2
(3.6) master meter is not empty, returns (3.1)
(3.7) finish.
2. as claimed in claim 1 a kind of based on the Embedded Zerotree Wavelet method against the Loop segmentation, it is characterized in that: in the described step (2), initial threshold ε
0Estimation:
Threshold epsilon
0Value can decide according to the requirement of compression efficiency, to different levels j (0≤j≤n-1), the threshold value ε that successively decreases
J+1=ε
j/ 2.
3. as claimed in claim 1 or 2 a kind of based on the Embedded Zerotree Wavelet method against the Loop segmentation, it is characterized in that: in the described step (2), the classification zerotree image is adopted in the compression of progressive network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012105923733A CN103077543A (en) | 2012-12-29 | 2012-12-29 | Embedded zerotree coding method on basis of inverse Loop subdivision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012105923733A CN103077543A (en) | 2012-12-29 | 2012-12-29 | Embedded zerotree coding method on basis of inverse Loop subdivision |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103077543A true CN103077543A (en) | 2013-05-01 |
Family
ID=48154062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012105923733A Pending CN103077543A (en) | 2012-12-29 | 2012-12-29 | Embedded zerotree coding method on basis of inverse Loop subdivision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103077543A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303620A (en) * | 2015-12-07 | 2016-02-03 | 杭州电子科技大学 | Triangular mesh subdivision surface access method based on vertex coding |
| CN106408620A (en) * | 2016-09-08 | 2017-02-15 | 成都希盟泰克科技发展有限公司 | Compressive sensing-based three-dimensional grid model data processing method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1975786A (en) * | 2006-12-11 | 2007-06-06 | 中山大学 | Progressive lattice generating method based on inverse loop subdivision |
| CN102186069A (en) * | 2011-01-14 | 2011-09-14 | 王慧 | Remote sensing image data compression method capable of maintaining measurement performance |
-
2012
- 2012-12-29 CN CN2012105923733A patent/CN103077543A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1975786A (en) * | 2006-12-11 | 2007-06-06 | 中山大学 | Progressive lattice generating method based on inverse loop subdivision |
| CN102186069A (en) * | 2011-01-14 | 2011-09-14 | 王慧 | Remote sensing image data compression method capable of maintaining measurement performance |
Non-Patent Citations (2)
| Title |
|---|
| 朱为朝: "基于零树小波图像压缩编码算法的研究", 《万方数据》 * |
| 马建平等: "面向移动终端的三角网格逆细分压缩算法", 《软件学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303620A (en) * | 2015-12-07 | 2016-02-03 | 杭州电子科技大学 | Triangular mesh subdivision surface access method based on vertex coding |
| CN106408620A (en) * | 2016-09-08 | 2017-02-15 | 成都希盟泰克科技发展有限公司 | Compressive sensing-based three-dimensional grid model data processing method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10694210B2 (en) | Scalable point cloud compression with transform, and corresponding decompression | |
| CN1181690C (en) | Encoding method for compression of video sequence | |
| Cosman et al. | Vector quantization of image subbands: A survey | |
| Fan et al. | Point cloud compression based on hierarchical point clustering | |
| CN105303535A (en) | Global subdivision pyramid model based on wavelet transformation | |
| CN110248190B (en) | Multilayer residual coefficient image coding method based on compressed sensing | |
| CN104040592A (en) | Predictive position decoding | |
| CN103347187A (en) | Remote-sensing image compression method for discrete wavelet transform based on self-adaptation direction prediction | |
| CN103546759A (en) | Image compression coding method based on combination of wavelet packets and vector quantization | |
| Milani | Fast point cloud compression via reversible cellular automata block transform | |
| US11983905B2 (en) | Methods for level partition of point cloud, and decoder | |
| JP2003533952A (en) | Encoding method for video sequence compression | |
| CN108810534A (en) | Method for compressing image based on direction Lifting Wavelet and improved SPIHIT under Internet of Things | |
| CN103077543A (en) | Embedded zerotree coding method on basis of inverse Loop subdivision | |
| KR100511719B1 (en) | 3-dimension normal mesh data compression apparatus by using rate-distortion optimization | |
| KR100511717B1 (en) | 3-dimension normal mesh data compression and transmission apparatus by using view-dependent processing | |
| CN101754021B (en) | Method for realizing mobile phone mobile portal technology based on improved wavelet-transform image compression method | |
| CN102316324B (en) | Image coding prediction method based on local minimum entropy | |
| Zhu et al. | An improved SPIHT algorithm based on wavelet coefficient blocks for image coding | |
| Payan et al. | Multiresolution 3D mesh compression | |
| Zhu et al. | Embedded Image Compression Using Differential Coding and Optimization Method | |
| CN110365990A (en) | A kind of quasi- lossless video encoding system in narrowband | |
| Köse et al. | 3D model compression using connectivity-guided adaptive wavelet transform built into 2D SPIHT | |
| CN103561275A (en) | Image compression method applying improved bandelet transformation | |
| Karthikeyan et al. | A survey on embedded zero tree wavelet |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130501 |