CN109743583A - A method of it is compressed based on adjacent error image - Google Patents

Abstract

The invention belongs to technical field of image processing, disclose a kind of based on adjacent error image compression method；Image is imported, resequences to the pixel in image, all pixels of two dimensional image or 3-D image is arranged as one-dimensional sequence array；The pixel value adjacent to this one-dimension array carries out two array values that difference is calculated sequence of differences and individually retains image sequence table head and table tail, and sequence of differences is carried out Huffman coding, obtains coding schedule；Sequence of differences is encoded according to coding schedule to obtain compressed data.It decodes compressed data progress Huffman to obtain sequence of differences based on coding schedule, two array values of sequence of differences combination image sequence gauge outfit and table tail is obtained into the one-dimensional sequence array of original image；Finally one-dimension array is reconfigured, restores original image.The present invention can be than directly using the compression of images effect of Huffman compression method outstanding by the compression of images effect of adjacent difference arithmetic.

Description

A method of it is compressed based on adjacent error image

Technical field

The invention belongs to technical field of image processing more particularly to a kind of methods based on the compression of adjacent error image.

Background technique

Currently, the prior art commonly used in the trade is such that

With information technology continuous development it is present, people start enjoy data age bring benefit, to image Requirement it is also higher and higher, various high definition cameras and high definition television have come into huge numbers of families；As remote sensing technology, space flight are distant The fast development of sense technology, various novel sensors results in more and more high quality, the appearance of high-resolution image data, Its bring technical difficulty is also following.The data such as various images, video, software installation packet need it is to be stored, transmission, under It carries, use.As the high-definition movie of one section of 2 hour, its size just has 1.5G or so, according to the speed of download of 10M per second It needs to wait 150 seconds to calculate.But be based on currently, China's 4G network speed is there are time difference and area differentiation, there are many country The speed of download in place is per second no more than 3M, and download time is also with regard to 450 seconds to 1500 seconds.The transmission of data needs to consider user Experience, this requires data to complete the process of transmission in the case where speed of download is relatively slow with the shorter time, that is, needs Data are compressed, to shorten the time of downloading.

In conclusion problem of the existing technology is: in the prior art, due to the appearance of various novel sensors, figure Image quality has measured huge promotion, but brings data volume simultaneously and increased dramatically, and image transmitting takes long time, and storage Need the great number of issues such as more equipment.It is growing with satellite image data scale such as in remote sensing field, it is limited Satellite channel capacity and a large amount of remotely-sensed datas of transmission between contradiction become increasingly conspicuous, Image Compression, which becomes, to be solved this and asks The effective way of topic, necessity and economic and social benefits are more and more obvious.

Solve the difficulty and meaning of above-mentioned technical problem:

The continuous maturation of information science technology, corresponding hardware device are also rapidly developing, update.It cannot achieve in the past Theoretical, method can slowly realize that the speed of service of present computer and the related software function of handling data are all very powerful now. Compression of images also necessarily will appear new thinking, new method, just can solve the people of present Technology Times in this way to efficient Demand.

Summary of the invention

In view of the problems of the existing technology, the present invention provides a kind of methods based on the compression of adjacent error image.

The invention is realized in this way a method of it is compressed, image is carried out first adjacent based on adjacent error image Difference processing, then carries out Huffman coding to difference processing result again.

Specifically includes the following steps:

Step 1 imports image；

Step 2 carries out the compression of adjacent difference approach, customized letter with custom function using yasuo function to image Number yasuo first resequences to the pixel in image according to certain sequential manner, by two dimensional image or 3-D image All pixels be arranged as one-dimensional sequence array；The pixel value adjacent to this one-dimension array carries out difference and difference is calculated Sequence of differences is carried out Huffman coding, is compiled by sequence and two array values for individually retaining image sequence table head and table tail Code table；Sequence of differences is encoded according to coding schedule to obtain compressed data；

Step 3 carries out adjacent difference approach decompression, custom function to image with custom function jieya realization Jieya is based on coding schedule and decodes compressed data progress Huffman to obtain sequence of differences；By sequence of differences combination image sequence table Two array values of head and table tail obtain the one-dimensional sequence array of original image, finally reconfigure to one-dimension array, extensive Multiple original image.

Further, it in step 2 and step 3, is realized with custom function yasuo, jieya and adjacent difference is carried out to image Method compression, decompression, specifically:

Import original image；

[h, e, o, p, w, M, N, Z]=yasuo (original image)；

%h is huffman table, and e is coding result；

%o is the one-dimensional sequence picture element matrix post-processed by adjacent difference；

%p is the minimum value of o matrix；

It is gauge outfit value that %w, which is original matrix,；

%M, N, Z are the ranks numbers of pages of original matrix.

[I]=jieya (h, e, p, w, M, N, Z)；% images relations decompression function；

subplot(1,2,1)；Imshow (original image)；Title (' original image name')；% shows original image；

subplot(1,2,2)；imshow(I)；Title (' decompressed image ')；% shows decompressed image；

In conclusion advantages of the present invention and good effect are as follows: import image, arranged again the pixel in image All pixels of two dimensional image or 3-D image are arranged as one-dimensional sequence array, the picture adjacent to this one-dimension array by sequence First value carries out two array values that difference is calculated sequence of differences and individually retains image sequence table head and table tail, by difference sequence Column carry out Huffman coding, obtain coding schedule, are encoded to obtain compressed data to sequence of differences according to coding schedule.Solution presses through Journey is the inverse process of compression process, decodes compressed data progress Huffman to obtain sequence of differences based on coding schedule, by difference sequence Column combine two array values of image sequence gauge outfit and table tail to obtain the one-dimensional sequence array of original image；Finally to one-dimension array It is reconfigured, restores original image.

Compression of images and decompression algorithm provided by the invention based on adjacent difference, image strong correlation and adjacent difference are calculated When method compatible degree is high, compression multiplying power can greatly increase；It can be than straight by the compression of images effect of adjacent difference arithmetic It connects and wants outstanding using the compression of images effect of Huffman compression method.

Image compression algorithm provided by the invention based on adjacent difference, image strong correlation and adjacent difference arithmetic agree with When spending high, compression multiplying power can greatly increase；The emulation experiment of the embodiment of the present invention shows by adjacent difference arithmetic Compression of images effect can be than directly using compression of images effect promoting 1.5 times or more of Huffman compression method, and this The image compression algorithm based on adjacent difference that invention provides also belongs to lossless compression algorithm.

Detailed description of the invention

Fig. 1 is the flow chart provided in an embodiment of the present invention for carrying out adjacent difference compression and decompression.

Fig. 2 is Lena image schematic diagram provided in an embodiment of the present invention.

Fig. 3 is pout image schematic diagram provided in an embodiment of the present invention.

Fig. 4 is circuit image schematic diagram provided in an embodiment of the present invention.

Fig. 5 is peppers image schematic diagram provided in an embodiment of the present invention.

Fig. 6 is provided in an embodiment of the present invention such as pout compression of images experimental comparison figure.

Fig. 7 is circuit compression of images experimental comparison figure provided in an embodiment of the present invention.

Fig. 8 is lean compression of images experimental comparison figure provided in an embodiment of the present invention.

Fig. 9 is peppers compression of images experimental comparison figure provided in an embodiment of the present invention.

Specific embodiment

In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to embodiments, to the present invention It is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to Limit the present invention.

Application principle of the invention is described in detail with reference to the accompanying drawing.

Method provided in an embodiment of the present invention based on the compression of adjacent error image, specifically includes the following steps:

S101: all pixels of image are arranged as an one-dimensional sequence first；

S102: retaining first element (gauge outfit) and the last one element (table tail) of one-dimensional sequence obtained in the previous step, Then the adjacent picture elements of sequence carry out difference and obtain sequence of differences；

S103: Huffman coding is carried out to sequence of differences, obtains Huffman coding schedule；

S104: sequence of differences is encoded based on coding schedule, obtains compressed data；

S105: compressed data is carried out by Huffman decoding based on coding schedule, obtains sequence of differences；

S106: in conjunction with the headers and footers in S102 step, sequence of differences is reverted into original one-dimension array；

S107: reconfiguring one-dimension array, restores original image.

The present invention is the use custom function yasuo that embodiment provides, and jieya, which is realized, carries out adjacent difference approach to image Compression, decompression, specifically:

Import original image

[h, e, o, p, w, M, N, Z]=yasuo (original image)；

%h is huffman table, and e is coding result；

%o is the one-dimensional sequence picture element matrix post-processed by adjacent difference；

%p is the minimum value of o matrix；

It is gauge outfit value that %w, which is original matrix,；

%M, N, Z are the ranks numbers of pages of original matrix.

[I]=jieya (h, e, p, w, M, N, Z)；% images relations decompression function

subplot(1,2,1)；Imshow (original image)；Title (' original image name')；% shows original image

subplot(1,2,2)；imshow(I)；Title (' decompressed image ')；% shows decompressed image

Application principle of the invention is further elaborated below with reference to specific experiment；

1 experiment porch MATLAB is introduced

MATLAB is a senior editor's software based on mathematical computations, provides various powerful array operation functions For handling various data acquisition systems.Matrix and array are the cores of MATLAB, because data all in MATLAB are all It is to be indicated and stored with array.

Although MATLAB is the author language towards matrix, it also has a kind of similar with other computer programming languages Programming characteristic.While carrying out data processing, MATLAB additionally provides various graphical user interface tools, convenient for user into The various application developments of row.

2 experimental image data source brief introductions

(1) Lena image: being one of the image data being frequently used in the algorithm research compressed, and reason is have Smooth block, clear careful lines, the shadow, the profound level of color that change gradually etc., drill it in verifying image processing When unraveling silk rule, quite effectively, as shown in Figure 2.

(2) pout image: pout image is a width gray level image, and child personage therein is than more visible and background window column line Rationality is stronger.It is also commonly used width standard picture in image procossing, is commonly used as the experiment test of image enhancement processing Figure, as shown in Figure 3.

(3) circuit image: circuit image is a width greyscale image data, has stronger grain, and warp The piece image data of testing standard are often taken as, as shown in Figure 4.

(4) peppers image: static color image, vegetables classification is bright-colored and easy differentiation, color effect are excellent It is elegant.It is classical one of the test chart of comparison, as shown in Figure 5.

As shown in Fig. 2, Lena image schematic diagram.

As shown in figure 3, pout image schematic diagram.

As shown in figure 4, circuit image schematic diagram.

As shown in figure 5, peppers image schematic diagram.

The different zones data experiment of 3 same images compares

It takes the array of 3 8*8 formats at random in the Lena image of Fig. 2, then does three groups of correlation datas respectively: right Image carries out Huffman compressed encoding, compressed encoding is carried out according to this patent method, the image data of Zig-Zag arrangement carries out Huffman compressed encoding is being carried out after adjacent difference processing.

The matrix that 3 groups of 8*8 are randomly selected in the lena image of Fig. 2 is respectively

A=[lena0x2Ebmp (9:16,1:8)]

B=[lena0x2Ebmp (101:108,1:8)]

C=[lena0x2Ebmp (101:108,101:108)]

1) specific experiment Process Design

Before experiment starts, customized several function performances are needed, to be efficiently completed compression experiment, Hyasuo letter Number, function are that Huffman coding is done directly to image；Yasuo function is completed after carrying out adjacent difference processing to image Huffman coding；Zigzag function is the function that function is ranked up to the image block of 8*8, function specific code and in detail note It releases and sees annex one.

According to that can set comparative experiments with superior function, steps are as follows.

(1) image is imported；

(2) Huffman coding, [a, b]=Hyasuo (original image) are carried out to image using Hyasuo compression function；% A is code table, and b is coding result.

(3) compression of adjacent difference approach is carried out to image using yasuo function

[h, e, o, p, w, M, N, Z]=yasuo (original image)；

(4) adjacent difference approach compression is carried out to image after arranging using Zig-Zag

I1=zigzag (original image)；[e, o, p, w, M, N, Z]=yasuo (I1)；

(5) tri- groups of image blocks of A, B, C are tested, with the form of table come statistical experiment data；

(6) experimental data is analyzed, sums up the advantage and disadvantage of method.

2) experimental data and analysis

According to experimental design procedure 1), the experiment to image block A, B, C is completed, statistical table is as shown in table 1.

Table 1: analysis of experimental data table

As shown in table 1, the image data of direct Huffman coding is compared, and the coding method carried out with adjacent difference From the point of view of experimental data, adjacent difference arithmetic is outstanding in the compression processing of the image block of 8*8 and directly carries out Huffman coding； And different alignment sequences, large effect can be generated to the compression effectiveness of adjacent difference arithmetic, but arrangement mode is to direct It the use of Huffman coding compression is ineffective.From the point of view of the statistical data of C image block, the three kinds of compress modes used, Encoding efficiency is all undesirable.It illustrates in the correlation of image in the case where cannot correctly map, in other words image The effect of adjacent difference arithmetic compression can have a greatly reduced quality when correlation is poor.Outstanding compression effectiveness should have reasonable image to close It is mapping function, reasonable arrangement mode and the coding method for meeting images relations.

The compression experiment of 4 different pictures

1) different images are carried out with the experimental design of adjacent difference compression

(1) compression of Huffman coding and decompression are carried out to image with custom function Hyasuo.

Compression: image data is imported

[a, b]=Hyasuo (original image)；% carries out Huffman coding

[z, x, s]=size (original image)；% obtains image and respectively ties up size

Obtain return value, a, b, z, x, s

Decoding: huffmandeco function is the decoding functions of the included Huffman coding of matlab

C=huffmandeco (b, a)；% return value c is one-dimensional decompression code value

I=reshape (c, z, x, s)；% returns to the dimension of original image

I=uint8 (I)；Return to the storage of uint8 format.

subplot(1,2,1)；imshow(lena0x2Ebmp)；Title (' original image ')；

subplot(1,2,2)；imshow(I)；Title (' decompressed image ')；

(2) it uses custom function yasuo, jieya to realize and adjacent difference approach compression, decompression is carried out to image.

Import original image

[h, e, o, p, w, M, N, Z]=yasuo (original image)；

%h is huffman table, and e is coding result；

%o is the one-dimensional sequence picture element matrix post-processed by adjacent difference；

%p is the minimum value of o matrix；

It is gauge outfit value that %w, which is original matrix,；

%M, N, Z are the ranks numbers of pages of original matrix.

[I]=jieya (h, e, p, w, M, N, Z)；% images relations decompression function

subplot(1,2,1)；Imshow (original image)；Title (' original image name')；% shows original image

subplot(1,2,2)；imshow(I)；Title (' decompressed image ')；% shows decompressed image

2) experimental result and comparison map analysis

The Huffman of pout image, circit image, lena image, 4 width picture of peppers image is completed according to experiment The compression of coding, decompression procedure.The compression of adjacent error image compression method, decompression procedure.Obtained experimental image such as Fig. 6, Shown in Fig. 7, Fig. 8, Fig. 9, a therein is original image, and b is the decompressed image of Huffman coding method, and c is adjacent difference Decompressed image.Experimental data is counted into table, detailed analysis two methods compression effectiveness obtains an accurate experiment conclusion.

As shown in fig. 6, pout compression of images experimental comparison figure.

As shown in fig. 7, circuit compression of images experimental comparison figure.

As shown in figure 8, lean compression of images experimental comparison figure.

As shown in figure 9, peppers compression of images experimental comparison figure.

Fig. 2-Fig. 5 is the image procossing especially common test image in compression of images field, is used in the embodiment of the present invention They compare test, to verify the validity of the method for the present invention.Fig. 6-Fig. 9 is respectively that test image is pressed by Huffman Later image is compressed and decompressed to contracting method and the method for the present invention.Comparison diagram 6, Fig. 7, Fig. 8, Fig. 9, original image therein, Huffman decompressed image, adjacent difference arithmetic decompressed image, do not lack on visual perception.And original image and decompression are schemed Picture number of pixels is consistent, illustrates that adjacent difference compression algorithm proposed by the present invention belongs to lossless compression.

As obtained by the experimental data of table 2, adjacent difference arithmetic belongs to lossless compression and passes through comparison diagram 6, Fig. 7, Fig. 8, Fig. 9. Original image therein, Huffman decompressed image, adjacent difference arithmetic decompressed image, does not lack on visual perception.And it is former Figure is consistent with decompression image pixel number, and it is reasonable in experimental design to illustrate, does not occur mistake in experimental calculation.Adjacent difference Algorithm belongs to lossless compression.

Adjacent difference arithmetic compression speed is slower, since adjacent difference arithmetic needs first first carry out adjacent difference meter to image Calculation processing, when image is larger, the speed calculated necessarily will affect the experience of the compression with image.

Adjacent difference arithmetic compression of images effect want it is excellent with directly to image carry out Huffman coding: from table 2 count Data from the point of view of, the adjacent difference compression multiple and mean code length of 4 width images will be better than directly carrying out Huffman volume to image The compression effectiveness of code, has biggish promotion for compression effectiveness.

Table 2:4 width compression of images contrast table

Annex one；

Realization of the 1.Zig-zag function in matlab

Function zz=zigzag (a)

[n, m]=size (a)；

If (n~=8&m~=8) % checks whether matrix belongs to 8*8 matrix

error('input array is not 8by 8')；

end

Zigzag=[0,1,8,16,9,2,3,10 ...

17,24,32,25,18,11,4,5,...

12,19,26,33,40,48,41,34....

27,20,13,6,7,14,21,28,...

35,42,49,56,47,50,43,36,...

29,22,15,23,30,37,44,51,...

58,59,52,45,38,31,39,46,...

53,60,61,54,47,55,62,63]；% enumerative technique is listed under 8*8zigzag sequence Mark

Zigzag=zigzag+1；

Aa=reshape (a, 1,64)；

Zz=aa (zigzag)；

2. directly carrying out Huffman compression function Hyasuo to image

Function [a, b]=Hyasuo (I)

% realizes the Huffman compression process to original image；

%a is the Huffman table of output；

%b is the Huffman table of output；

[M, N]=size (I)；The dimension of % acquisition image

I1=I (:)；Image is become one-dimensional sequence by %

P=zeros (1,256)；The null matrix of % building 1-256

% obtains the probability of each symbol；

For i=0:255

P (i+1)=length (find (I1==i))/(M*N)；

end

K=0:255；

Dict=huffmandict (k, P)；% generates Huffman table

A=dict；% exports Huffman table

Enco=huffmanenco (I1, dict)；% generates Huffman coding

B=enco；% exports Huffman coding

end

3. image compression algorithm function yasuo

[h, e, o, p, w, M, N, Z]=yasuo (circuit)；

Function [h, e, o, p, w, M, N, Z]=yasuo (II)

% is mainly the compression for doing image；

%h is Huffman table；

%e is coding result；

%o is by adjacent difference treated one-dimensional sequence picture element matrix；

%p is the minimum value of o matrix；

It is gauge outfit value that %w, which is original matrix,；

%M, N are the ranks numbers of original matrix.

I=double (II)；Image is carried out Data Transform by %

A=numel (II)；

I2=reshape (I, 1, a)；% is converted to one-dimensional pixel matrix

B=I2 (1)；

I3=[b, I2 (1:a-1)]；I4=I2-I3；O=I4；C=min (I4)；P=c；

I=I4-c；After % obtains adjacent difference, in addition the minimum value of matrix solves the problems, such as negative

[M, N]=size (I)；

I1=I (:)；

P=zeros (1,301)；%% obtains the probability of each symbol；

For i=0:300

P (i+1)=length (find (I1==i))/(M*N)；

End

T=0:300；

Dict=huffmandict (t, P)；% generates dictionary using huffmandict function

H=dict；

Enco=huffmanenco (I1, dict)；% generates huffman coding using huffmanenco function.

E=enco；W=I2 (1)；[M, N]=size (II)；The size of % original matrix returns to original dimension when for decoding Several images

M=M；% return value M, N N=N；Z=Z；

End

Annex two

1. image algorithm decompression function jieya

Function [I]=jieya (a, b, d, f, M, N)

%a is the Huffman table generated, and b is the Huffman coding of compression of images

%d original is matrix minimum value, and f is gauge outfit pixel value, and I is the image after decompression

Deco=huffmandeco (b, a)；

I5=deco+d；I5=I5'；I7=[f, I5]；I8=cumsum (I7)；

I9=[I8 (2:M*N+1)]；I10=reshape (I9, M, N)；

I10=uint8 (I10)；I=I10；

end

The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.

Claims (4)

1. one kind is based on the compression of adjacent error image and decompressing method, which is characterized in that described based on adjacent error image pressure The method of contracting carries out adjacent difference processing to image first, then carries out Huffman coding to difference processing result again.

2. the method compressed and decompressed based on adjacent error image as described in claim 1, which is characterized in that described to be based on phase The method of the compression of adjacent error image and decompression specifically includes the following steps:

Step 1 imports image；

Step 2, the compression of adjacent difference approach is carried out using custom function yasuo to image, and custom function yasuo is first It first resequences to the pixel in image according to certain sequential manner, by two dimensional image or all pixels of 3-D image It is arranged as one-dimensional sequence array；The pixel value adjacent to this one-dimension array carries out difference and sequence of differences and independent is calculated Sequence of differences is carried out Huffman coding, obtains coding schedule by two array values for retaining image sequence table head and table tail；According to Coding schedule encodes sequence of differences to obtain compressed data；

Step 3 is realized using custom function jieya and carries out adjacent difference approach decompression, custom function jieya to image It decodes compressed data progress Huffman to obtain sequence of differences based on coding schedule；By sequence of differences combination image sequence gauge outfit and Two array values of table tail obtain the one-dimensional sequence array of original image, finally reconfigure to one-dimension array, restore former Beginning image.

3. the method as claimed in claim 2 based on the compression of adjacent error image, which is characterized in that in the step 2, with certainly Defined function yasuo, jieya, which is realized, carries out adjacent difference approach compression, decompression to image, specifically:

Import original image；

[h, e, o, p, w, M, N, Z]=yasuo (original image)；

%h is Huffman coding schedule, and e is coding result；

%o is the one-dimensional sequence picture element matrix post-processed by adjacent difference；

%p is the minimum value of o matrix；

It is gauge outfit value that %w, which is original matrix,；

%M, N, Z are the ranks numbers of pages of original matrix；

[I]=jieya (h, e, p, w, M, N, Z)；% images relations decompression function；

subplot(1,2,1)；Imshow (original image)；Title (' original image name')；% shows original image；

subplot(1,2,2)；imshow(I)；Title (' decompressed image ')；% shows decompressed image.

4. a kind of image procossing using the method based on the compression of adjacent error image described in claims 1 to 3 any one is flat Platform.