CN102572431A - Overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method - Google Patents
Overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method Download PDFInfo
- Publication number
- CN102572431A CN102572431A CN2011104514049A CN201110451404A CN102572431A CN 102572431 A CN102572431 A CN 102572431A CN 2011104514049 A CN2011104514049 A CN 2011104514049A CN 201110451404 A CN201110451404 A CN 201110451404A CN 102572431 A CN102572431 A CN 102572431A
- Authority
- CN
- China
- Prior art keywords
- matrix
- homogeneous blocks
- blocks
- value
- est
- 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
Images
Landscapes
- Compression Of Band Width Or Redundancy In Fax (AREA)
Abstract
The invention discloses an overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method, which comprises a coding process and a decoding process. The coding process mainly comprises the following steps of: dividing a gray image matrix into same-class blocks which can be overlapped by utilizing an extended Gouraud shadowing method and four criterions of an overlapping rectangular NAM; and coding all the same-class blocks to acquire a color table P and a coordinate table Q of the same-class blocks. The decoding process mainly comprises the following steps of: decoding coordinate matrixes H, V and I according to the coordinate table Q, reconstructing a decoded image by utilizing the extended Gouraud shadowing method according to the H, the V, the I and the color table P, and calculating the peak signal to noise ratio (PSNR) of the reconstructed image. On the premise of ensuring the quality of the image, the method has a low bit rate, a small number of blocks and high processing speed, and can be applied to the conventional joint photographic experts group (JPEG) market as well as the emerging fields of wireless communication, network transmission, medical images and the like.
Description
Technical field
The present invention relates to Computer Image Processing, particularly a kind of based on asymmetric inversed placement model (NAM) graphical representation method that can overlapping rectangle subpattern.
Background technology
Graphical representation is a major issue in the fields such as computer graphics, computer vision, robot, image processing and pattern recognition.Effectively the graphical representation method can not only be saved memory space, and can also improve the speed of image processing.Graphical representation is one of present most active research field.With regard to the method for expressing of bianry image, the method for expressing of code word set method for expressing, string representation method and tree structure etc. is arranged, yet these method for expressing all are based on spatial data structure.The method for expressing that is different from spatial data structure; Mohamed has proposed a kind of bianry image method for expressing (Mohamed SA based on non-overlapped rectangle; Fahmy MM.Binary image compression using efficient partitioning into rectangular regions.IEEE Transactions on Communications; 1995; 43 (5): 1888-1892), and this method for expressing is applied in the compression of bianry image effectively, obtained effect preferably.Afterwards; Quddus has done further research to this method of Mohamed; Proposed a kind of based on bianry image method for expressing (Quddus A that can overlapping rectangle; Fahmy MM.Binary text image compression using overlapping rectangular partitioning.atter Recognition Letters; 1999,20 (2): 81-88), its experimental result shows that use can overlapping rectangle divides the sum of resulting rectangle and always be less than the sum that uses non-overlapped rectangle to divide resulting rectangle.Quddus in the compression to text image, has obtained this method for expressing successful Application than Mohamed better experiment results.For a width of cloth bianry image; Compare with non-overlapped rectangle division; Can overlapping rectangle division can obtain higher expression efficient (Quddus A; Fahmy MM.Binary text image compression using overlapping rectangular partitioning.Pattern Recognition Letters, 1999,20 (2): 81-88).
Yet, because the image in the reality is gray level image mostly, so the research of gray level image method for expressing is had purposes and practical meaning more widely.Based on B-tree bougainvillea shape coding (BTTC) method; People such as Distasi have proposed to represent algorithm (R. Distasi based on the gray level image of spatial data structure first; M.Nappi; S.Vitulano.Image compression by B-tree triangular coding.IEEE Transactions on Communications, 1997,45 (9): 1095-1100).Because Distasi comes up the design of spatial data structure first from the expression that the expression of bianry image has expanded to gray level image, therefore represent it is a pioneering job really based on the gray level image of the spatial data structure of BTTC.Afterwards, based on S data tree structure (W.D.Jonge, P.Scheuermann; A.Schijf.S+-Trees:An efficient structure for the representation of large pictures.Computer Vision and Image Understanding, 1994,59 (3): 265-280) with Gouraud shadowing method (J.D.Foley; A.V.Dam, S.K.Feiner, et al.Computer Graphics; Principle, and Practice, second ed.Reading; MA:Addision-Wesley; 1990), (K.Chung, J.Wu.Improved image compression using S-tree and shading approach.IEEE Transactions on Communications such as Chung; 2000,48 (5): the gray level image that 748-751) has proposed a kind of spatial data structure based on S tree is represented (STC) method.The experimental result of STC method shows: keeping picture quality and do not sacrificing under the situation of image compression rate, the STC method is lacked half at least than the time of implementation of BTTC method.Subsequently; Chung etc. have proposed a kind of mixing gray scale graphical representation method based on DCT territory and spatial domain; Abbreviate SDCT method for expressing (Chung KL, Liu YW, Yan WM.A hybrid gray image representation using spatial-and DCT-based approach with application to moment computation.Journal of Visual Communication and Image Representation as; 2006,17 (6): 1209-1226).Its experimental result shows: under the prerequisite that keeps picture quality, the SDCT method for expressing is on average improving 63.08% than STC method for expressing aspect the compression ratio raising rate, is a kind of effective gray level image method for expressing.Yet because the encoding and decoding time complexity of SDCT method for expressing is identical, and be higher than the STC method for expressing.Therefore, with respect to the STC method for expressing, SDCT representes that the higher compression ratio that obtains is is cost with the encoding and decoding time complexity of sacrificing algorithm.
Although above-mentioned spatial data structure representes to have many advantages, they too stress the symmetry cut apart, therefore are not optimum method for expressing.Thought by means of the Packing problem; Cutting apart maximized asymmetric dividing method with searching is target; The inventor had once proposed a kind of coloured image method for expressing (Zheng Yunping of the asymmetric inversed placement model (NAM) based on the rectangle subpattern; Chen Chuanbo. a kind of coloured image method for expressing based on asymmetric inversed placement model. the software journal; 2007,18 (11): 2932-2941), the basic thought of this method for expressing is: given layout pattern and predefined difform rectangle subpattern of having got well; From this given pattern, extract the rectangle subpattern of non-overlapping copies then out, represent given pattern with the combination of these rectangle subpatterns.Yet this article propose based on the coloured image method for expressing of rectangle NAM be a kind of harmless method for expressing of image, and be not suitable for the expression that diminishes of image.Afterwards; In order more effectively image to be represented and to be compressed, the inventor representes to have proposed the thought of algorithm a kind of new gray level image and represented algorithm by means of BTTC and STC gray level image; Abbreviate RNAMC as and represent algorithm (Zheng Yunping; Chen Chuanbo. a kind of new gray level image is represented algorithm research. Chinese journal of computers, 2010,33 (12): 2397-2406).Represent in the algorithm at RNAMC; Because contrary layout result is non-overlapped; Therefore only need the code word of given upper left, bottom right and isolated piece, can obtain the matrix R of a sign summit type, in matrix R; Symbol " 1 " and " 2 " are used for identifying the upper left corner and the lower right corner of rectangle respectively, and symbol " 1 " only is used for identifying the isolated point rectangle.RNAMC representes that the homogeneous blocks of the non-overlapping copies after the contrary layout of algorithm has following characteristics, if that is: with the mode scan matrix R of raster scan, the left upper apex of each rectangle and summit, bottom right are nearest on column direction.As long as original image is by the rectangle non-overlapping copies after cutting apart, matrix R can decode.Therefore representing in the algorithm at RNAMC, can only be the homogeneous blocks of non-overlapping copies with the contrary layout of original image, otherwise this algorithm can't be decoded.The complexity of this algorithm is the same with the complexity of BTTC and STC, that is: the time complexity of encoding and decoding part is respectively O (m log m) and O (m), and wherein m is the pixel count of gray level image.Experimental result shows: compare with the SDCT method for expressing with popular STC; Under the prerequisite that keeps picture quality; RNAMC representes that algorithm has higher compression ratio and piece number still less, thereby can more effectively reduce data space, is a kind of good method that gray level image is represented.But represent in the algorithm at RNAMC, subpattern be do not allow overlapping, and can be overlapping rectangle divide and can obtain to divide higher expression efficient than non-overlapped rectangle, therefore, need provide a kind of based on NAM graphical representation method that can overlapping rectangle subpattern.
Summary of the invention
The objective of the invention is to overcome the shortcoming and defect of above-mentioned prior art; Provide a kind of based on NAM graphical representation method that can overlapping rectangle subpattern; Homogeneous blocks when the compression in the time of can more effectively improving graphical representation when reduces graphical representation is total, further improves the expression and the operating efficiency of image model.
The object of the invention is realized through following technical scheme:
A kind ofly comprise cataloged procedure and decode procedure based on NAM graphical representation method that can overlapping rectangle subpattern,
Said cataloged procedure may further comprise the steps:
S1 with gray level image matrix G be divided into can be overlapping homogeneous blocks, specifically may further comprise the steps:
S1.1 definition homogeneous blocks is to satisfy the rectangle subpattern of following condition: the gray value g of all pixels in this rectangle subpattern (x y) all satisfies condition | and g (x, y)-g
Est(x, y) |≤ε; Wherein, ε is the error allowance of user's setting; g
Est(x y) representes that (it defines as follows coordinate in this rectangle subpattern: establish (x for x, the approximate gray value of y) locating
1, y
1), (x
2, y
2) be respectively the coordinate figure in this rectangle subpattern upper left corner and the lower right corner, x
1≤x≤x
2, y
1≤y≤y
2
If x
1＜x
2And y
1＜y
2, g then
Est(x, y)=g
5+ (g
6-g
5) * i
1,
If x
1≠ x
2And y
1=y
2, g then
Est(x, y)=g
1+ (g
4-g
1) * [(x-x
1)/(x
2-x
1)];
If x
1=x
2And y
1≠ y
2, g then
Est(x, y)=g
1+ (g
4-g
1) * [(y-y
1)/(y
2-y
1)];
If x
1=x
2And y
1=y
2, g then
Est(x, y)=g
1
G wherein
1, g
2, g
3, g
4Be respectively the upper left corner, the lower right corner, the lower left corner of this rectangle subpattern, the gray value in the upper right corner; g
5=g
1+ (g
2-g
1) * i
2, g
6=g
3+ (g
4-g
3) * i
2, i
1=(y-y
1)/(y
2-y
1), i
2=(x-x
1)/(x
2-x
1);
Define horizontal block matrix H, vertical blocks matrix V and single-point block matrix I that 3 sizes are M * N, wherein, all elements all is initialized as 0 among H, V and the I; With the counting variable n assignment of homogeneous blocks is 0; Wherein, M and N are natural number;
S1.2 scans homogeneous blocks:
2 sizes of S1.2.1 definition are the provisional matrix of M * N: TempH and TempV are respectively applied for the distribution condition that is identified at horizontal block and vertical blocks in the scanning process; All elements among TempH and the TempV all is initialized as 0;
S1.2.2 begins scanning from first of the gray level image matrix G upper left corner, confirms the starting point (x of a homogeneous blocks that is not identified according to the order of raster scan
1, y
1), in order abscissa and ordinate are scanned, determine the coordinate (x in the homogeneous blocks lower right corner according to the Gouraud shadowing method of expansion
2, y
2), obtain the maximum homogeneous blocks (number of pixels that promptly this homogeneous blocks comprised is maximum) of area, and this homogeneous blocks is made a check mark in gray level image matrix G;
In scanning process, follow following criterion:
If the value of homogeneous blocks corresponding position in TempH that the area of confirming during scanning is maximum is 0, the value of then that this homogeneous blocks among the TempH is corresponding position is changed to 1;
If the value of the homogeneous blocks correspondence position in TempH that runs into during scanning is 1, promptly this homogeneous blocks is horizontal block and also underlapped mistake, then searches for the coordinate (x in the upper left corner of this homogeneous blocks earlier
1, y
1) and the coordinate (x in the lower right corner
2, y
2) again it is become vertical blocks; The said method that becomes vertical blocks is: earlier this homogeneous blocks value of correspondence position in TempH is changed to 0, the value of correspondence position is changed to 1 in TempV, again with this homogeneous blocks corresponding points (x in matrix H
1, y
1) and point (x
2, y
2) value be changed to 0 respectively, corresponding points (x in the V matrix
1, y
1) value be changed to 1, corresponding points (x
2, y
2) value be 2; Scanning process is proceeded simultaneously, and homogeneous blocks value of correspondence position in TempH behind the inferior end of scan that the area that obtains is maximum is changed to 2;
If the value of the homogeneous blocks correspondence position in TempV that runs into during scanning is 1, then continue scanning, will work as the maximum homogeneous blocks value of correspondence position in TempH of area that obtains behind time end of scan and be changed to 2;
If homogeneous blocks correspondence position value in TempH that scanning runs into the time is 2, promptly this homogeneous blocks is horizontal block and also overlapping mistake, then this moment the end of scan, the homogeneous blocks that scanning this time obtains is the homogeneous blocks of area maximum;
S1.3 adds 1 with the value of n, and the parameter of the maximum homogeneous blocks of the area that obtains of recording step S1.2: the coordinate (x in the upper left corner
1, y
1), the coordinate (x in the lower right corner
2, y
2) and the gray value g at 4 angles
1, g
2, g
3, g
4
S1.4 stores the described parameter of step S1.3 among the color table P into, and to corresponding points (x among matrix H, the I
1, y
1), point (x
2, y
2) the position identify;
S1.5 judges whether the homogeneous blocks among the gray level image matrix G all is identified and finishes;
If carry out step S1.6;
If not, repeating step S1.2～S1.5;
S1.6 output color table P;
S2 encodes to the coordinate of all nonzero elements in 3 matrixes by the order of H, V, I according to the coordinate data compression algorithm successively, and coding result is stored among the coordinates table Q;
Said decoding is specially: according to coordinates table Q, decode matrix H, V and I, according to H, V, I and color table P, utilize the Gouraud method of expansion to rebuild decoded picture.
Scanning process described in the step S1.2 is specially according to the strategy of row major scanning:
In scanning process, earlier make abscissa X constantly increase separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, make ordinate Y constantly increase again separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε.
Scanning process described in the step S1.2 is specially according to the strategy of row priority scan:
In scanning process, earlier make ordinate Y constantly increase separately, up to scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, make abscissa X constantly increase again separately, scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε.
Step S1.4 is said to be stored the described parameter of step S1.3 among the color table P into, and to corresponding points (x among matrix H, the I
1, y
1), point (x
2, y
2) the position identify, be specially:
If x
1＜x
2And y
1＜y
2, then with parameter g
1, g
2, g
3, g
4Store among the color table P, that is: P{n} ← { (g
1, g
2, g
3, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1=x
2And y
1≠ y
2, then with parameter g
1, g
4Store among the color table P, that is: P{n} ← { (g
1, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1≠ x
2And y
1=y
2, then with parameter g
1, g
4Store among the color table P, that is: P{n} ← { (g
1, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1=x
2And y
1=y
2, then with g
1Store among the color table P, that is: P{n} ← { (g
1), and with corresponding points (x among the matrix I
1, y
1) the position identify with ' 1 '.
Step S2 presses the order of H, V, I according to the coordinate data compression algorithm, and the coordinate of all nonzero elements in each matrix is encoded, and is specially:
The size of lining by line scan is the matrix of M * N, if the current line all elements is zero, uses bit ' 0 ' to represent that there is not nonzero element from the beginning to the end in this row, does not encode to this row; If there is nonzero element in this row, then before each nonzero element, add a prefix symbol ' 1 ', behind the prefix symbol, add code word then, and represent the position of each nonzero element place row with b bit in order to sign nonzero element 1 and 2; When last element of this row is zero, after then in the end a nonzero element coding was accomplished, it was zero to represent that with bit ' 0 ' the remaining element of one's own profession is.
Saidly represent the position of each nonzero element place row to be specially with b bit:
If not neutral element is first nonzero element of being expert at, then b=[log
2N]; B the bit of this moment is used for indicating the position of first nonzero element about being expert at head end;
If not neutral element is not first nonzero element of being expert at, then b=[log
2(N-c)], wherein c is the preceding once position of the row of the nonzero element of coding; B the bit of this moment is used for representing that this nonzero element is about the preceding once position of the right-hand member of the nonzero element of coding.
Said according to coordinates table Q, decode matrix H, V and I, according to H, V, I and color table P, utilize the Gouraud method of expansion to rebuild decoded picture, be specially:
The matrix g that S3.1 is M * N with a size
EstAll elements compose any initial value, and be 0 with the counting variable n assignment of homogeneous blocks;
S3.2 is according to coordinates table Q, by the order of H, V, I H, V and I matrix that to decode 3 sizes successively be M * N;
S3.3 provides the total s of homogeneous blocks according to color table P;
S3.4 scans H, V and I matrix simultaneously, if the value of scanning is 1 pixel, then according to P{n}, judges the type of this homogeneous blocks and obtains its gray value;
S3.5 utilizes all g of this homogeneous blocks of Gouraud shadowing method calculating of expansion according to the type of homogeneous blocks
Est(x, y), g wherein
Est(x y) representes coordinate in this homogeneous blocks (x, the approximate gray value of y) locating;
S3.6 presses the order of raster scan and gives matrix g with the decoded result assignment of this homogeneous blocks
Est
S3.7 makes n add 1, and judges whether satisfy n＜s this moment, if, repeating step S3.4～S3.7; If not, carry out step S3.8;
S3.8 is according to matrix g
EstThe output gray level image, and calculate Y-PSNR PSNR according to following formula:
The present invention by means of bianry image can be overlapping the thought of rectangular area coding, provide a kind of based on NAM graphical representation method that can overlapping rectangle subpattern, abbreviate the ORNAMC method for expressing as.The present invention includes the Code And Decode process, wherein cataloged procedure mainly is a Gouraud shadowing method of utilizing expansion and 4 criterions that can overlapping rectangle NAM, with the gray level image matrix be divided into can be overlapped homogeneous blocks; All homogeneous blocks are encoded again, obtain its color table P and coordinates table Q, decode procedure mainly is according to coordinates table Q; Decode coordinates matrix H, V and I; According to H, V and I and color table P, utilize the Gouraud shadowing method of expansion to rebuild decoded picture, and calculate its PSNR again.
Compared with prior art, the present invention has the following advantages and beneficial effect:
1, compare with present popular STC method, have piece number still less, under the prerequisite that guarantees picture quality, the number of its homogeneous blocks has on average descended 31.31%, thereby has image processing speed faster.
2, the complexity with the popular square computational algorithm of representing based on STC compares; Theory analysis shows that the complexity based on the square computational algorithm of ORNAMC method for expressing is the complexity that is lower than square computational algorithm that STC representes; Be merely O (n); N is the number of the homogeneous blocks of gray level image when representing with ORNAMC here, thereby ORNAMC representes to have computational speed faster.
3, with commercialization the JPEG method compare, have lower bit rate and encoding and decoding speed faster, its encoding and decoding complexity is merely O (m log m) and O (m), m is the pixel count of gray level image here.The ORNAMC method for expressing is on average improving 91.84% than STC method for expressing aspect the compression ratio raising rate under the prerequisite that keeps picture quality.
4, compare with the RNAMC method, keeping under the prerequisite of picture quality, the ORNAMC method for expressing has subpattern number and bit rate still less, thereby can more effective raising graphical representation and the efficient of operation.
5, the present invention not only can be applicable to traditional JPEG market, like printer, scanner, digital camera etc., but also can be applicable to emerging field, like wireless telecommunications, networking transmission, medical image etc.
Description of drawings
Fig. 1 is the sketch map of cataloged procedure.
Fig. 2 is the sketch map of homogeneous blocks.
Fig. 3 is the sketch map of gray level image matrix G.
Fig. 4 is the horizontal block matrix H after identifying.
Fig. 5 is the vertical blocks matrix V after identifying.
Fig. 6 is the single-point block matrix I after identifying.
Fig. 7 is the sketch map of decode procedure.
Embodiment
Below in conjunction with embodiment and accompanying drawing, the present invention is done to specify further, but execution mode of the present invention is not limited thereto.
Embodiment
The present invention is a kind of based on NAM graphical representation method that can overlapping rectangle subpattern, comprises cataloged procedure and decode procedure:
As shown in Figure 1, cataloged procedure may further comprise the steps:
S1 with gray level image matrix G be divided into can be overlapping homogeneous blocks, specifically may further comprise the steps:
S1.1 definition homogeneous blocks is to satisfy the rectangle subpattern of following condition: the gray value g of all pixels in this rectangle subpattern (x y) all satisfies condition | and g (x, y)-g
Est(x, y) |≤ε; Wherein, ε is the error allowance of user's setting; g
Est(x y) representes that (it defines as follows coordinate in this rectangle subpattern: establish (x for x, the approximate gray value of y) locating
1, y
1), (x
2, y
2) be respectively the coordinate figure in this rectangle subpattern upper left corner and the lower right corner, x
1≤x≤x
2, y
1≤y≤y
2
If x
1＜x
2And y
1＜y
2, g then
Est(x, y)=g
5+ (g
6-g
5) * i
1,
If x
1≠ x
2And y
1=y
2, g then
Est(x, y)=g
1+ (g
4-g
1) * [(x-x
1)/(x
2-x
1)];
If x
1=x
2And y
1≠ y
2, g then
Est(x, y)=g
1+ (g
4-g
1) * [(y-y
1)/(y
2-y
1)];
If x
1=x
2And y
1=y
2, g then
Est(x, y)=g
1
G wherein
1, g
2, g
3, g
4Be respectively the upper left corner, the upper right corner, the lower left corner of this rectangle subpattern, the gray value in the lower right corner; g
5=g
1+ (g
2-g
1) * i
2, g
6=g
3+ (g
4-g
3) * i
2, i
1=(y-y
1)/(y
2-y
1), i
2=(x-x
1)/(x
2-x
1); Fig. 2 is the sketch map of homogeneous blocks.
Define horizontal block matrix H, vertical blocks matrix V and single-point block matrix I that 3 sizes are M * N, wherein, all elements all is initialized as 0 among H, V and the I; With the counting variable n assignment of homogeneous blocks is 0; Wherein, M and N are natural number.
S1.2 scans homogeneous blocks:
2 sizes of S1.2.1 definition are the provisional matrix of M * N: TempH and TempV are respectively applied for the distribution condition that is identified at horizontal block and vertical blocks in the scanning process; All elements among TempH and the TempV all is initialized as 0;
S1.2.2 begins scanning from first of the gray level image matrix G upper left corner, confirms the starting point (x of a homogeneous blocks that is not identified according to the order of raster scan
1, y
1), in order abscissa and ordinate are scanned, determine the coordinate (x in the homogeneous blocks lower right corner according to the Gouraud shadowing method of expansion
2, y
2), obtain the maximum homogeneous blocks of area, and this homogeneous blocks is made a check mark in gray level image matrix G;
Can be during scanning according to row major strategy or row preference strategy, wherein the row major strategy is:
In scanning process, earlier make abscissa X constantly increase separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, the abscissa of establishing this point is x, and then the more preceding abscissa of this point is the abscissa x in the homogeneous blocks lower right corner
2Make ordinate Y constantly increase again separately, until scan gray value g (x, y) for satisfying | g (x, y)-g
Est(x, y) | till the point of＞ε, the ordinate of establishing this point is y, and then the more preceding abscissa of this point is the ordinate y in the homogeneous blocks lower right corner
2
Wherein the row preference strategy is:
In scanning process, earlier make ordinate Y constantly increase separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, the ordinate of establishing this point is y ', and then the more preceding abscissa of this point is the ordinate y in the homogeneous blocks lower right corner
2Make abscissa X constantly increase again separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, the abscissa of establishing this point is x ', and then the more preceding abscissa of this point is the abscissa x in the homogeneous blocks lower right corner
2
In scanning process, follow following criterion:
If the value of homogeneous blocks corresponding position in TempH that the area of confirming when scanning is maximum is changed to 1 for (not running into any piece before), the value of then that this homogeneous blocks among the TempH is corresponding position;
If the value of the homogeneous blocks correspondence position in TempH that runs into during scanning is 1, promptly this homogeneous blocks is horizontal block and also underlapped mistake, then searches for the coordinate (x in the upper left corner of this homogeneous blocks earlier
1, y
1) and the coordinate (x in the lower right corner
2, y
2) again it is become vertical blocks; The said method that becomes vertical blocks is: earlier this homogeneous blocks value of correspondence position in TempH is changed to 0, the value of correspondence position is changed to 1 in TempV, again with this homogeneous blocks corresponding points (x in matrix H
1, y
1) and point (x
2, y
2) value be changed to 0 respectively, corresponding points (x in the V matrix
1, y
1) value be changed to 1 and corresponding points (x
2, y
2) value be 2; Scanning process is proceeded simultaneously, and homogeneous blocks value of correspondence position in TempH behind the inferior end of scan that the area that obtains is maximum is changed to 2;
If the value of the homogeneous blocks correspondence position in TempV that runs into during scanning is 1 (running into previous vertical blocks); Then begin with this vertical blocks overlapping; And continue scanning, will work as the maximum homogeneous blocks value of correspondence position in TempH of area that obtains behind time end of scan and be changed to 2;
If the homogeneous blocks that runs into during scanning is horizontal block and also overlapping mistake, promptly this homogeneous blocks correspondence position value in TempH is 2, then can not be overlapping with this horizontal block once more, and the end of scan this time scanned the homogeneous blocks that obtains and was the maximum homogeneous blocks of area this moment.
S1.3 adds 1 with the value of n, and the parameter of the maximum homogeneous blocks that obtains of recording step S1.2: the coordinate (x in the upper left corner
1, y
1), the coordinate (x in the lower right corner
2, y
2) and the gray value g at 4 angles
1, g
2, g
3, g
4Wherein the K representation is used in the concrete stored record of gray value;
S1.4 stores the described parameter of step S1.3 among the color table P into, and to corresponding points (x among matrix H, the I
1, y
1), point (x
2, y
2) the position identify (to corresponding points (x in the matrix V
1, y
1), point (x
2, y
2) the sign of position in step S1.2, accomplish), be specially:
If x
1＜x
2And y
1＜y
2, then with parameter g
1, g
2, g
3, g
4Store among the color table P, that is: P{n} ← { (g
1, g
2, g
3, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1=x
2And y
1≠ y
2, then with parameter g
1, g
4Store among the color table P, that is: P{n} ← { (g
1, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1≠ x
2And y
1=y
2, then with parameter g
1, g
4Store among the color table P, that is: P{n} ← { (g
1, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1=x
2And y
1=y
2, then with g
1Store among the color table P, that is: P{n} ← { (g
1), and with corresponding points (x among the matrix I
1, y
1) the position identify with ' 1 '.
Being that the size that extracts 512 * 512 the F16 image is that 16 * 16 subimage matrix (as shown in Figure 3) is an example from a width of cloth size; When ε=20 can matrix be divided into can be overlapping 3 types of homogeneous blocks; That is: horizontal block, vertical blocks and single-point piece; And in horizontal block matrix H, vertical blocks matrix V and single-point block matrix I, identify different homogeneous blocks and summit type thereof with individual tags respectively, the horizontal block matrix H after the sign, vertical blocks matrix V and single-point block matrix I are respectively shown in Fig. 4, Fig. 5, Fig. 6 (no single-point piece in the present embodiment).
S1.5 judges whether the homogeneous blocks among the gray level image matrix G all is identified and finishes;
If carry out step S1.6;
If not, repeating step S1.2～S1.5;
S1.6 output color table P;
S2 encodes to the coordinate of all nonzero elements in 3 matrixes by the order of H, V, I according to the coordinate data compression algorithm successively, and coding result is stored among the coordinates table Q;
Said coordinate to all nonzero elements in each matrix is encoded, and is specially:
The size of lining by line scan is the matrix of M * N, if the current line all elements is zero, uses bit ' 0 ' to represent that there is not nonzero element from the beginning to the end in this row, does not encode to this row; If there is nonzero element in this row; Then before each nonzero element, add a prefix symbol ' 1 '; Behind the prefix symbol, add code word (the prefix code word set of using in the cataloged procedure is as shown in table 1 below) then, and represent the position of each nonzero element place row with b bit in order to sign nonzero element 1 and 2; When last element of this row is zero, after then in the end a nonzero element coding was accomplished, it was zero to represent that with bit ' 0 ' the remaining element of one's own profession is.
Saidly represent the position of each nonzero element place row to be specially with b bit:
If not neutral element is first nonzero element of being expert at, then b=[log
2N]; B the bit of this moment is used for indicating the position of first nonzero element about being expert at head end;
If not neutral element is not first nonzero element of being expert at, then b=[log
2(N-c)], wherein c is the preceding once position of the row of the nonzero element of coding; B the bit of this moment is used for representing that this nonzero element is about the preceding once position of the right-hand member of the nonzero element of coding.
The code word set of table 13 type summit symbol
As shown in Figure 7, decode procedure is: according to coordinates table Q, decode matrix H, V and I, according to H, V, I and color table P, utilize the Gouraud method of expansion to rebuild decoded picture, be specially:
The matrix g that S3.1 is M * N with a size
EstAll elements compose any initial value, and be 0 with the counting variable n assignment of homogeneous blocks;
S3.2 is according to coordinates table Q, by the order of H, V, I H, V and I matrix that to decode 3 sizes successively be M * N;
S3.3 provides the total s of homogeneous blocks according to color table P;
S3.4 scans H, V and I matrix simultaneously, if the value of scanning is 1 pixel, then according to P{n}, judges the type of this homogeneous blocks and obtains its gray value;
S3.5 utilizes all g of this homogeneous blocks of Gouraud shadowing method calculating of expansion according to the type of homogeneous blocks
Est(x, y), g wherein
Est(x y) representes coordinate in this homogeneous blocks (x, the approximate gray value of y) locating;
S3.6 presses the order of raster scan and gives matrix g with the decoded result assignment of this homogeneous blocks
Est
S3.7 makes n add 1, and judges whether satisfy n＜s this moment, if, repeating step S3.4～S3.7; If not, carry out step S3.8;
S3.8 is according to matrix g
EstThe output gray level image, and calculate Y-PSNR PSNR according to following formula:
Below in the image processing field with 256 * 256 sizes habitual ' Lena ', ' Goldhill ', ' Peppers ', ' Boat ', ' Robot1 ', ' Robot2 ', ' Robot3 ' and ' Robot4 ' gray level image as tested object; Through directly or indirectly comparing, the validity and the superiority of ORNAMC method for expressing is described with method for expressing such as RNAMC, STC and SDCT.Given 4 kinds of different error tolerance ε=10,20,30 and 40, table 2～4, table 2 have specifically provided the number of the homogeneous blocks of ORNAMC and STC method for expressing.Table 3 has specifically provided the PSNR of ORNAMC and STC method for expressing, and table 4 has specifically provided the compression ratio of ORNAMC and STC method for expressing.
Can know from the experimental result of SDCT and RNAMC: under the prerequisite that keeps picture quality, SDCT method for expressing and RNAMC represent that algorithm is on average improving 63.08% and 82.91% respectively than STC method for expressing aspect the compression ratio raising rate.In addition, aspect the encoding and decoding complexity of algorithm, the encoding and decoding time complexity of SDCT method for expressing is identical, and is higher than the STC method for expressing.Therefore, with respect to the STC method for expressing, the higher compression ratio that the SDCT method for expressing obtains is to be cost with the encoding and decoding time complexity of sacrificing algorithm.And the complexity of RNAMC and STC method for expressing is identical, that is: the time complexity of encoding and decoding part is respectively O (n log n) and O (n), and wherein n is the pixel count of gray level image.Simultaneously, the experimental result of RNAMC also shows: keeping under the prerequisite of picture quality, RNAMC represent algorithm aspect the number of homogeneous blocks than STC method for expressing decreased average 18.54%.ORNAMC representes that the encoding and decoding complexity of algorithm is identical with RNAMC.
Be not difficult to find out that from table 2～4 along with the increase of ε, the compression ratio of ORNAMC and STC method for expressing all is increase trend, and homogeneous blocks number and PSNR are all on a declining curve.For 8 given width of cloth images; Result of calculation further shows: under different error tolerances (ε=10,20,30 and 40); The homogeneous blocks of ORNAMC (STC) is counted average out to 2907 (4232); The PSNR average out to 31.1106 (33.9774) of ORNAMC (STC), ORNAMC is with respect to the compression ratio raising rate average out to 91.84% of STC.Therefore; Although ORNAMC representes algorithm and aspect PSNR, has on average descended 8.44% than STC method for expressing; But on average improving 91.84% than STC method for expressing aspect the compression ratio raising rate; Simultaneously ORNAMC represent algorithm aspect the number of homogeneous blocks also than STC method for expressing decreased average 31.31%, thereby more help improving the efficient of graphical representation and the speed of image processing.
In addition; Under identical error tolerance, such as: when ε=20, for 8 given width of cloth images; The homogeneous blocks of RNAMC (STC) is counted average out to 2978 (4443); The compression ratio average out to 7.0508 (3.6886) of RNAMC (STC), the PSNR average out to 32.0583 (34.8910) of RNAMC (STC), obviously; RNAMC is being superior to STC aspect number of homogeneous blocks (decreased average 32.97%) and the compression ratio raising rate (on average having improved 91.15%), although the PSNR of RNAMC has on average descended 8.12% than STC.Usually, if the PSNR of reconstructed image reaches about 30, human eye is subjective to be can not differentiate difference between original image and the reconstructed image.Obviously, when ε=20, the PSNR of the image after these 2 kinds of algorithms are rebuild has all reached more than 30.
The experimental result of RNAMC shows: keeping under the prerequisite of picture quality, RNAMC representes that algorithm is on average improving 82.91% than STC method for expressing aspect the compression ratio raising rate, aspect the number of homogeneous blocks than STC method for expressing decreased average 18.54%.
And ORNAMC representes that algorithm is keeping on average improving 91.84% than STC method for expressing aspect the compression ratio raising rate under the prerequisite of picture quality, aspect the number of homogeneous blocks than STC method for expressing decreased average 31.31%.Obviously, keeping under the prerequisite of picture quality, the homogeneous blocks the when compression when ORNAMC representes that algorithm can more effectively improve graphical representation with the SDCT method for expressing than RNAMC when reduces graphical representation is total.
Can know by above-mentioned analysis; Compare with RNAMC, SDCT and STC method for expressing; Under the prerequisite that keeps picture quality; ORNAMC representes that algorithm has higher compression ratio and piece number still less, thereby can more effectively reduce data space and the processing speed that improves image manipulation, thereby is a kind of better method for expressing of gray level image.
The number of the homogeneous blocks that table 2ORNAMC and STC represent
The PSNR that table 3ORNAMC and STC represent
The comparison of the compression performance that table 4ORNAMC and STC represent
The foregoing description is a preferred implementation of the present invention; But execution mode of the present invention is not limited by the examples; Other any do not deviate from change, the modification done under spirit of the present invention and the principle, substitutes, combination, simplify; All should be the substitute mode of equivalence, be included within protection scope of the present invention.
Claims (7)
1. one kind based on NAM graphical representation method that can overlapping rectangle subpattern, comprises cataloged procedure and decode procedure, it is characterized in that,
Said cataloged procedure may further comprise the steps:
S1 with gray level image matrix G be divided into can be overlapping homogeneous blocks, specifically may further comprise the steps:
S1.1 definition homogeneous blocks is to satisfy the rectangle subpattern of following condition: the gray value g of all pixels in this rectangle subpattern (x y) all satisfies condition | and g (x, y)-g
Est(x, y) |≤ε; Wherein, ε is the error allowance of user's setting; g
Est(x y) representes that (it defines as follows coordinate in this rectangle subpattern: establish (x for x, the approximate gray value of y) locating
1, y
1), (x
2, y
2) be respectively the coordinate figure in this rectangle subpattern upper left corner and the lower right corner, x
1≤x≤x
2, y
1≤y≤y
2
If x
1＜x
2And y
1＜y
2, g then
Est(x, y)=g
5+ (g
6-g
5) * i
1,
If x
1≠ x
2And y
1=y
2, g then
Est(x, y)=g
1+ (g
4-g
1) * [(x-x
1)/(x
2-x
1)];
If x
1=x
2And y
1≠ y
2, g then
Est(x, y)=g
1+ (g
4-g
1) * [(y-y
1)/(y
2-y
1)];
If x
1=x
2And y
1=y
2, g then
Est(x, y)=g
1
G wherein
1, g
2, g
3, g
4Be respectively the upper left corner, the lower right corner, the lower left corner of this rectangle subpattern, the gray value in the upper right corner; g
5=g
1+ (g
2-g
1) * i
2, g
6=g
3+ (g
4-g
3) * i
2, i
1=(y-y
1)/(y
2-y
1), i
2=(x-x
1)/(x
2-x
1);
Define horizontal block matrix H, vertical blocks matrix V and single-point block matrix I that 3 sizes are M * N, wherein, all elements all is initialized as 0 among H, V and the I; With the counting variable n assignment of homogeneous blocks is 0; Wherein, M and N are natural number;
S1.2 scans homogeneous blocks:
2 sizes of S1.2.1 definition are the provisional matrix of M * N: TempH and TempV are respectively applied for the distribution condition that is identified at horizontal block and vertical blocks in the scanning process; All elements among TempH and the TempV all is initialized as 0;
S1.2.2 begins scanning from first of the gray level image matrix G upper left corner, confirms the starting point (x of a homogeneous blocks that is not identified according to the order of raster scan
1, y
1), in order abscissa and ordinate are scanned, determine the coordinate (x in the homogeneous blocks lower right corner according to the Gouraud shadowing method of expansion
2, y
2), obtain the maximum homogeneous blocks of area, and this homogeneous blocks is made a check mark in gray level image matrix G;
In scanning process, follow following criterion:
If the value of homogeneous blocks corresponding position in TempH that the area of confirming during scanning is maximum is 0, the value of then that this homogeneous blocks among the TempH is corresponding position is changed to 1;
If the value of the homogeneous blocks correspondence position in TempH that runs into during scanning is 1, promptly this homogeneous blocks is horizontal block and also underlapped mistake, then searches for the coordinate (x in the upper left corner of this homogeneous blocks earlier
1, y
1) and the coordinate (x in the lower right corner
2, y
2) again it is become vertical blocks; The said method that becomes vertical blocks is: earlier this homogeneous blocks value of correspondence position in TempH is changed to 0, the value of correspondence position is changed to 1 in TempV, again with this homogeneous blocks corresponding points (x in matrix H
1, y
1) and point (x
2, y
2) value be changed to 0 respectively, corresponding points (x in the V matrix
1, y
1) value be changed to 1, corresponding points (x
2, y
2) value be 2; Scanning process is proceeded simultaneously, and homogeneous blocks value of correspondence position in TempH behind the inferior end of scan that the area that obtains is maximum is changed to 2;
If the value of the homogeneous blocks correspondence position in TempV that runs into during scanning is 1, then continue scanning, will work as the maximum homogeneous blocks value of correspondence position in TempH of area that obtains behind time end of scan and be changed to 2;
If homogeneous blocks correspondence position value in TempH that scanning runs into the time is 2, promptly this homogeneous blocks is horizontal block and also overlapping mistake, then this moment the end of scan, the homogeneous blocks that scanning this time obtains is the homogeneous blocks of area maximum;
S1.3 adds 1 with the value of n, and the parameter of the maximum homogeneous blocks of the area that obtains of recording step S1.2: the coordinate (x in the upper left corner
1, y
1), the coordinate (x in the lower right corner
2, y
2) and the gray value g at 4 angles
1, g
2, g
3, g
4
S1.4 stores the described parameter of step S1.3 among the color table P into, and to corresponding points (x among matrix H, the I
1, y
1), point (x
2,
Y2) the position identify;
S1.5 judges whether the homogeneous blocks among the gray level image matrix G all is identified and finishes;
If carry out step S1.6;
If not, repeating step S1.2～S1.5;
S1.6 output color table P;
S2 encodes to the coordinate of all nonzero elements in 3 matrixes by the order of H, V, I according to the coordinate data compression algorithm successively, and coding result is stored among the coordinates table Q;
Said decoding is specially: according to coordinates table Q, decode matrix H, V and I, according to H, V, I and color table P, utilize the Gouraud method of expansion to rebuild decoded picture.
2. according to claim 1 based on NAM graphical representation method that can overlapping rectangle subpattern, it is characterized in that scanning process described in the step S1.2 is specially according to the strategy of row major scanning:
In scanning process, earlier make abscissa X constantly increase separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, make ordinate Y constantly increase again separately, until scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε.
3. according to claim 1 based on NAM graphical representation method that can overlapping rectangle subpattern, it is characterized in that scanning process described in the step S1.2 is specially according to the strategy of row priority scan:
In scanning process, earlier make ordinate Y constantly increase separately, up to scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε, make abscissa X constantly increase again separately, scan gray value g (x, y) satisfy | g (x, y)-g
Est(x, y) | till the point of＞ε.
4. according to claim 1ly it is characterized in that based on NAM graphical representation method that can overlapping rectangle subpattern step S1.4 is said to be stored the described parameter of step S1.3 among the color table P into, and to corresponding points (x among matrix H, the I
1, y
1), point (x
2, y
2) the position identify, be specially:
If x
1＜x
2And y
1＜y
2, then with parameter g
1, g
2, g
3, g
4Store among the color table P, that is: P{n} ← { (g
1, g
2, g
3, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1=x
2And y
1≠ y
2, then with parameter g
1, g
4Store among the color table P, that is: P{n} ← { (g
1, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1≠ x
2And y
1=y
2, then with parameter g
1, g
4Store among the color table P, that is: P{n} ← { (g
1, g
4), and with corresponding points (x in the matrix H
1, y
1), point (x
2, y
2) the position use ' 1 ' and ' 2 ' to identify respectively;
If x
1=x
2And y
1=y
2, then with g
1Store among the color table P, that is: P{n} ← { (g
1), and with corresponding points (x among the matrix I
1, y
1) the position identify with ' 1 '.
5. according to claim 4 based on NAM graphical representation method that can overlapping rectangle subpattern; It is characterized in that step S2 presses the order of H, V, I according to the coordinate data compression algorithm; Coordinate to all nonzero elements in each matrix is encoded, and is specially:
The size of lining by line scan is the matrix of M * N, if the current line all elements is zero, uses bit ' 0 ' to represent that there is not nonzero element from the beginning to the end in this row, does not encode to this row; If there is nonzero element in this row, then before each nonzero element, add a prefix symbol ' 1 ', behind the prefix symbol, add code word then, and represent the position of each nonzero element place row with b bit in order to sign nonzero element 1 and 2; When last element of this row is zero, after then in the end a nonzero element coding was accomplished, it was zero to represent that with bit ' 0 ' the remaining element of one's own profession is.
6. according to claim 5ly it is characterized in that, saidly represent the position of each nonzero element place row to be specially with b bit based on NAM graphical representation method that can overlapping rectangle subpattern:
If not neutral element is first nonzero element of being expert at, then b=[log
2N]; B the bit of this moment is used for indicating the position of first nonzero element about being expert at head end;
If not neutral element is not first nonzero element of being expert at, then b=[log
2(N-c)], wherein c is the preceding once position of the row of the nonzero element of coding; B the bit of this moment is used for representing that this nonzero element is about the preceding once position of the right-hand member of the nonzero element of coding.
7. according to claim 6ly it is characterized in that based on NAM graphical representation method that can overlapping rectangle subpattern, said according to coordinates table Q; Decode matrix H, V and I; According to H, V, I and color table P, utilize the Gouraud method of expansion to rebuild decoded picture, be specially:
The matrix g that S3.1 is M * N with a size
EstAll elements compose any initial value, and be 0 with the counting variable n assignment of homogeneous blocks;
S3.2 is according to coordinates table Q, by the order of H, V, I H, V and I matrix that to decode 3 sizes successively be M * N;
S3.3 provides the total s of homogeneous blocks according to color table P;
S3.4 scans H, V and I matrix simultaneously, if the value of scanning is 1 pixel, then according to P{n}, judges the type of this homogeneous blocks and obtains its gray value;
S3.5 utilizes all g of this homogeneous blocks of Gouraud shadowing method calculating of expansion according to the type of homogeneous blocks
Est(x, y), g wherein
Est(x y) representes coordinate in this homogeneous blocks (x, the approximate gray value of y) locating;
S3.6 presses the order of raster scan and gives matrix g with the decoded result assignment of this homogeneous blocks
Est
S3.7 makes n add 1, and judges whether satisfy n＜s this moment, if, repeating step S3.4～S3.7; If not, carry out step S3.8;
S3.8 is according to matrix g
EstThe output gray level image, and calculate Y-PSNR PSNR according to following formula:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110451404.9A CN102572431B (en) | 2011-12-29 | 2011-12-29 | Overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110451404.9A CN102572431B (en) | 2011-12-29 | 2011-12-29 | Overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102572431A true CN102572431A (en) | 2012-07-11 |
| CN102572431B CN102572431B (en) | 2014-05-07 |
Family
ID=46416740
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110451404.9A Expired - Fee Related CN102572431B (en) | 2011-12-29 | 2011-12-29 | Overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102572431B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104331883A (en) * | 2014-10-28 | 2015-02-04 | 华南理工大学 | Image boundary extraction method based on non-symmetry and anti-packing model |
| CN113706639A (en) * | 2021-07-21 | 2021-11-26 | 国网江苏省电力有限公司电力科学研究院 | Image compression method and device based on rectangular NAM, storage medium and computing equipment |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101364306A (en) * | 2008-09-12 | 2009-02-11 | 华中科技大学 | Stationary image compression coding method based on asymmetric inversed placement model |
-
2011
- 2011-12-29 CN CN201110451404.9A patent/CN102572431B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101364306A (en) * | 2008-09-12 | 2009-02-11 | 华中科技大学 | Stationary image compression coding method based on asymmetric inversed placement model |
Non-Patent Citations (2)
| Title |
|---|
| 郑运平，陈传波: "一种基于非对称逆布局模型的彩色图像表示方法", 《JOURNAL OF SOFTWARE》 * |
| 郑运平，陈传波: "一种新的灰度图像表示算法研究", 《计算机学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104331883A (en) * | 2014-10-28 | 2015-02-04 | 华南理工大学 | Image boundary extraction method based on non-symmetry and anti-packing model |
| CN104331883B (en) * | 2014-10-28 | 2017-06-06 | 华南理工大学 | A kind of image boundary extraction method based on asymmetric inversed placement model |
| CN113706639A (en) * | 2021-07-21 | 2021-11-26 | 国网江苏省电力有限公司电力科学研究院 | Image compression method and device based on rectangular NAM, storage medium and computing equipment |
| CN113706639B (en) * | 2021-07-21 | 2024-02-23 | 国网江苏省电力有限公司电力科学研究院 | Image compression method and device based on rectangular NAM, storage medium and computing equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102572431B (en) | 2014-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11216985B2 (en) | Point cloud attribute compression method based on deleting 0 elements in quantisation matrix | |
| CN112383780B (en) | Encoding and decoding method and device for point matching reference set and index back and forth scanning string matching | |
| DE60213657T2 (en) | GENERATION OF GRAPHIC CODES BY HALFTONE GRASPING WITH INTEGRATED GRAPHIC CODING | |
| CN109743570B (en) | Method for compressing screen content video | |
| JP2018520428A (en) | Image processing method and apparatus | |
| CN110892725A (en) | Method and apparatus for encoding/decoding a point cloud representing a 3D object | |
| EP2645697B1 (en) | Image processing apparatus and method | |
| TWI646467B (en) | Method and system for generating two dimensional barcode with hidden data | |
| US20100289798A1 (en) | Image processing method and image processing apparatus | |
| CN104318560B (en) | A kind of image partition method based on asymmetric inversed placement model | |
| CN106162140A (en) | The compression method of a kind of panoramic video and device | |
| CN101969567A (en) | Image coding method based on total variation | |
| CN101364306B (en) | Stationary image compression coding method based on asymmetric inversed placement model | |
| CN104331883B (en) | A kind of image boundary extraction method based on asymmetric inversed placement model | |
| CN104270624A (en) | Region-partitioning 3D video mapping method | |
| CN102572431B (en) | Overlapping rectangular subpattern-based non-symmetry and anti-packing model (NAM) image representation method | |
| EP3175423A1 (en) | Digital image watermarking system and method | |
| CN113179407B (en) | Video watermark embedding and extracting method and system based on interframe DCT coefficient correlation | |
| CN105654017B (en) | Two-dimentional decoding transmission method and system | |
| CN107682699A (en) | A kind of nearly Lossless Image Compression method | |
| CN104244008B (en) | Depth video coding method | |
| Liu et al. | Edge-based inpainting and texture synthesis for image compression | |
| CN107483961A (en) | Predict and go here and there position encoded data compression method and device using arcuate scanning string | |
| CN103020887B (en) | Non-destructive hidden transmission method for on-board data | |
| CN113034390B (en) | Image restoration method and system based on wavelet prior attention |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140507 Termination date: 20201229 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |