CN102222350A - Image compressing device and method based on wavelet transformation and arithmetic coding - Google Patents
Image compressing device and method based on wavelet transformation and arithmetic coding Download PDFInfo
- Publication number
- CN102222350A CN102222350A CN201110143696XA CN201110143696A CN102222350A CN 102222350 A CN102222350 A CN 102222350A CN 201110143696X A CN201110143696X A CN 201110143696XA CN 201110143696 A CN201110143696 A CN 201110143696A CN 102222350 A CN102222350 A CN 102222350A
- Authority
- CN
- China
- Prior art keywords
- unit
- low frequency
- wavelet
- coefficient
- computing
- 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 Or Coding Systems Of Tv Signals (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention discloses an image compressing device and method based on wavelet transformation and arithmetic coding. The device comprises a reading unit, an operation pool unit, a wavelet transformation unit, a coefficient write-back unit, a low-frequency prediction unit, a vector quantification unit and an arithmetic coding unit. After wavelet transformation of the same level is performed, the lossless compression rate is 3 percent higher than that of JPEG2000 (Joint Photographic Experts Group 2000); and a system framework which combines vector quantification with wavelet transformation is adopted for the device, so that high complexity of JPEG2000 entropy coding is avoided in the system, and the compression performance is enhanced.
Description
Technical field
What the present invention relates to is a kind of apparatus and method of technical field of image processing, specifically is a kind of device and method of the compression of images based on wavelet transformation and arithmetic coding.
Background technology
At present, wavelet transformation is a kind of mathematic(al) manipulation method that can be applicable to image transformation, and encoding with JPEG2000 standard code method based on the digital image compression of wavelet transformation is representative, but because its computational complexity is higher, not widespread use at present.The purpose of image transformation is to change the distribution of view data or can separate all kinds of important informations, so that carry out entropy coding.Wavelet transformation can separate picture high and low frequency information, make low-frequency information concentrate on low frequency sub-band (LL subband), laterally, high-frequency information vertical and diagonal is distributed in respectively in LH, HL and the HH subband, these subbands are set of coefficient, and follow-up computing is carried out at these coefficients.This helps encoding at the character of different sub-band.And use 5/3 small echo can carry out the integer non-loss transformation, the various overheads that do not have floating-point operation to bring.
The low frequency prediction is the computing of doing for the redundancy of eliminating the small echo low frequency coefficient.Because the value span of low frequency coefficient is bigger, is unsuitable for follow-up entropy coding, after the low frequency prediction, obtain the residual error of coefficient, the distribution of residual error is more suitable for entropy coding.
Vector quantization is earlier vector to be formed in packet then, at last the method that quantizes at vector.At first require data to be encoded to be in desired scope, in code book, find code vector for these data then, with this code vector vector to be encoded of encoding near this vector.Since codebook size much smaller than vector to be encoded the combination that may occur, therefore, the redundancy behind the coding can be very little.
Arithmetic coding can be encoded to a string symbol sebolic addressing the decimal between 0 to 1, and code efficiency is generally than Huffman encoding height.
Finding through the retrieval to prior art, is that Columbus encodes at the used entropy coding of scrambler described in the VLSI image compression encoder (2009.04.08 of China national Department of Intellectual Property) of wavelet transformation.Under lower computational complexity, obtained compression effectiveness preferably.
But the entropy coding method that the prior art adopted can not be obtained and be not so good as arithmetic coding efficient height, only is an alternative entropy coding method for fear of JPEG2000 entropy coding computing complexity, and its code efficiency is far away from the arithmetic coding height.
Summary of the invention
The present invention is directed to the prior art above shortcomings, a kind of device and method of the compression of images based on wavelet transformation and arithmetic coding is provided, make it have higher compression efficiency and lower computational complexity.
The present invention is achieved by the following technical solutions:
The present invention relates to a kind of device of the compression of images based on wavelet transformation and arithmetic coding, comprising: reading unit, computing pool unit, wavelet transform unit, coefficient write back unit, low frequency predicting unit, vector quantization unit and arithmetic coding unit, wherein:
The computing pool unit receive from the outside and buffer memory original image data and output image raw data to wavelet transform unit, wavelet transform unit carries out that the wavelet transformation computing obtains and writes back wavelet coefficients at different levels or the low frequency that the unit writes back the computing pool unit by coefficient predicting the outcome to reading unit, reading unit respectively with wavelet transform unit, the low frequency predicting unit link to each other with the vector quantization unit and the output image raw data and through the wavelet transform unit wavelet transformation computing obtains and writes back the unit by coefficient writing back the wavelet coefficients at different levels of computing pool unit to wavelet transform unit, reading unit is exported five-star low frequency wavelet coefficient tremendously low frequency predicting unit and output low frequency predicts the outcome to arithmetic coding unit, wavelet transform unit receives the original image data of reading unit output or removes the low frequency wavelet coefficients at different levels of highest low frequency, the height of the corresponding input of output wavelet coefficient grade behind wavelet transformation, low frequency wavelet coefficient to coefficient writes back the unit, coefficient writes back that the unit links to each other with the low frequency predicting unit with wavelet transform unit respectively and the wavelet coefficient and the low frequency predicting unit output ground low frequency that receive wavelet transform unit output predicts the outcome, predict the outcome to the computing pool unit through map addresses output wavelet coefficient and low frequency, output low frequency predicted the outcome to coefficient and writes back the unit after the low frequency predicting unit received the low frequency wavelet coefficient of reading unit output and carries out low frequency prediction computing, the output vector quantification symbol is to the coding unit that counts after the row vector quantization operations that the low frequency of vector quantization unit reception reading unit output predicts the outcome and other wavelet coefficients are gone forward side by side, and the coding unit that counts carries out the arithmetic coding computing to the vector quantization symbol and exports the code word of arithmetic coding.
Described reading unit is controlled wavelet transform unit, low frequency predicting unit and vector quantization unit respectively by three the address generator that reads the address of computing pool unit is formed, wherein:
First address generator links to each other with the computing pool unit with the image wavelet arithmetic unit respectively, and: when the one-level wavelet transformation according to from top to bottom to image inside, from left to right read the order calculated address; According to from top to bottom, ranks are 1 at interval between the neighbor that reads order and read from left to right when the secondary wavelet transformation; By that analogy, the ranks that adopt when the third level and follow-up wavelet transformation are spaced apart 2
K-1-1, wherein k is a wavelet transformation level several;
Second address generator links to each other with the computing pool unit with the low frequency predicting unit respectively and calculates low frequency sub-band last cell address, according to image inside from top to bottom, from right to left read the address that order produces each low frequency sub-band element according to conversion progression;
The three-address generator link to each other with the computing pool unit with the vector quantization unit respectively and according to from senior to rudimentary, image is from top to bottom inner, the order that reads from left to right produces LL, LH, HL and the HH address of totally four subbands.
Described computing pool unit is made up of to store the data of two width of cloth images two memory blocks, and the data that write by the outside are YCbCr components of image, and each works alone according to the order of lining by line scan and two memory blocks.
Described wavelet transform unit is calculated the small echo one-level conversion of continuous 5 row images, and wherein: preceding 2 row are to calculate to finish the result in every continuous 5 row images, and back 3 row are to calculate intermediate result.
Described coefficient writes back the unit and writes back address generator and low frequency and write back address generator and form by controlling small echo that wavelet transform unit and low frequency predicting unit write back the address of computing pool unit with wavelet coefficient respectively, this coefficient writes back the unit and exports the operation result of wavelet transform unit and low frequency predicting unit to the computing pool unit, wherein:
Small echo writes back address generator and connects image wavelet arithmetic unit and computing pool unit, writes back order with the wavelet transformation address generator;
Low frequency writes back address generator and connects low frequency predicting unit and computing pool unit, writes back order with low frequency predicted address generator.
Described low frequency predicting unit with each low frequency coefficient under with the condition of location computing adjacent with left and above the difference of coefficient mean value of adjacent 2 low frequency sub-band LL and LH as the predicted value of current coefficient the predicted value that reads all low frequency sub-band coefficients of order computation corresponding to low frequency predicted address generator.
Described vector quantization unit is divided into coefficient sets with coefficient, and coefficient sets is carried out vector quantization, and the result is directly sent into arithmetic coding unit.
Described arithmetic coding unit is carried out arithmetic coding to the coefficient sets of vector quantization unit output, produces coding result.
The present invention relates to the method for compressing image of said apparatus, comprise the steps:
Reads image data is to wavelet transform unit from the computing pool unit for the first step, reading unit, and the computing pool unit is an internal memory of storage two width of cloth images;
Second step, wavelet transform unit are carried out wavelet transformation to the view data of reading in and are obtained wavelet coefficient; Be the economize on hardware expense, the RAM of wavelet transform unit inside only can buffer memory 5 row view data, read the update image data by constantly writing, finish until the entire image conversion;
The 3rd step, coefficient write back the unit wavelet coefficient are write back the computing pool unit;
According to from top to bottom, order from right to left reads low frequency coefficient to the low frequency predicting unit from the computing pool unit for the 4th step, reading unit, low frequency sub-band is carried out low frequency predict computing, and simultaneously, coefficient writes back the unit operation result is write back in the computing pool unit;
The 5th step, reading unit are read each subband to the vector quantization unit according to from senior to rudimentary and the order of LL, LH, HL, HH, and coding result is sent with coefficient grouping, quantification in the vector quantization unit;
The 6th step, coding result enter arithmetic coding unit, and arithmetic coding is encoded to the vector quantization result, output code flow.
JPEG2000 apparatus of the present invention computational complexity that lossless compress performance of the present invention is better than wavelet transformation at the same level is lower, based on the JPEG2000 of small echo and arithmetic coding, greatly reduces computational complexity with respect to equally; The present invention simultaneously adopts pipeline organization, has improved treatment effeciency.
Description of drawings
Fig. 1 is a system chart of the present invention.
Fig. 2 is a system flow waterline synoptic diagram.
Fig. 3 is reading of small echo computing and writes back order.
The order that low frequency sub-band read and writes back when Fig. 4 was the low frequency prediction.
Embodiment
Below embodiments of the invention are elaborated, present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.
Embodiment
As shown in Figure 1, this device comprises: reading unit, computing pool unit, wavelet transform unit, coefficient write back unit, low frequency predicting unit, vector quantization unit and arithmetic coding unit.
The computing pool unit receives from the outside and buffer memory original image data and read address output image raw data to reading unit according to what the wavelet transformation address generator provided, reading unit respectively with wavelet transform unit, the low frequency predicting unit links to each other with the vector quantization unit, at first can the output image raw data to wavelet transform unit, wavelet transform unit receives the original image data of reading unit output, output one-level height behind wavelet transformation, the wavelet coefficient of low frequency to coefficient writes back the unit, coefficient writes back the unit and links to each other with the low frequency predicting unit with wavelet transform unit respectively, at this moment coefficient writes back the wavelet coefficient that the unit receives wavelet transform unit output, write back address generator according to small echo and export wavelet coefficient to the computing pool unit according to the address that produces with the location operation rule, so far the one-level wavelet transformation has been finished.
Low frequency wavelet coefficient that last conversion obtains is exported to reading unit in the address that second level conversion is at first provided according to wavelet transformation address generator in the reading unit by the computing pond, reading unit exports these low frequency wavelet coefficients to wavelet transform unit, wavelet transform unit receives the wavelet coefficient of reading unit output, behind wavelet transformation, export to and write back the unit, write back the wavelet coefficient that the unit receives wavelet transform unit output, write back address generator according to the write back address computing pool unit that produces with the location operation rule according to small echo.The process of similar second level conversion is constantly carried out conversion, and till reaching desired progression, this progression is specified by input reading unit by the user.
After finishing the wavelet transformation of appointment progression of piece image, the computing pond begins to export according to the address that reading unit medium and low frequency predicted address generator produces the highest low frequency coefficient tremendously low frequency predicting unit of this width of cloth image, write back address output low frequency operation result to the coefficient that address generator produces according to low frequency after the low frequency predicting unit receives the highest low frequency wavelet coefficient of reading unit output and carries out low frequency prediction computing and write back the unit, the low frequency computing of this width of cloth image so far finishes.
After finishing the low frequency prediction computing of piece image, the computing pool unit begins the address output low frequency operation result according to the generation of the vector quantization address generator in the reading unit, the low frequency that the vector quantization unit receives reading unit output predicts the outcome and goes forward side by side after the row vector quantization operations output vector quantification symbol to the coding unit that counts, the coding unit that counts carries out the arithmetic coding computing to the vector quantization symbol and exports the code word of arithmetic coding, and so far the compression encoding process of piece image finishes
Each arithmetic element as shown in Figure 1 can be carried out next width of cloth treatment of picture after handling piece image, realize streamline in view of the above.
The wavelet transformation address generator links to each other with the computing pool unit with the image wavelet arithmetic unit respectively, and: when the one-level wavelet transformation according to from top to bottom to image inside, from left to right read the order calculated address; Suppose that a certain figure image width is X, the figure image height is Y, the position of a certain raw data of this image in image be v (1...Y) OK, h (1...X) row, the memory address in the computing pool unit of first pixel of image is addr, and then the memory address of this raw image data in the computing pool unit is
addr+w＊(v-1)+h-1
Simultaneously, this also is institute's address stored after this original image data process arbitrary number of level wavelet transformation computing.
According to from top to bottom, ranks are 1 at interval between the neighbor that reads order and read from left to right when the secondary wavelet transformation; By that analogy, the ranks that adopt when the third level and follow-up wavelet transformation are spaced apart 2
K-1-1, wherein k is a wavelet transformation progression, therefore, the hypothesis during according to the one-level conversion, during the secondary wavelet transformation, the address that the wavelet transformation address generator produces is
addr+w＊(v-1)+h-1，v＝2＊n+1，h＝2＊m+1，m，n＝0，1，2...
During the conversion of N level, the address that the wavelet transformation address generator produces is
addr+w＊(v-1)+h-1，v＝2
n+2
N-1-1，h＝2
m+2
N-1-1，m，n＝1，2...
Low frequency predicted address generator links to each other with the computing pool unit with the low frequency predicting unit respectively and calculates low frequency sub-band last cell address, according to image inside from top to bottom, from right to left read the address that order produces each low frequency sub-band element according to conversion progression, what suppose to finish is N level wavelet transformation, and then the address that produces of low frequency predicted address generator is
addr+w＊(v-1)+h-1，v＝2
n+2
N-1，h＝2
m+2
N-1，m，n＝1，2...
The vector quantization address generator link to each other with the computing pool unit with the vector quantization unit respectively and according to image inside from senior to rudimentary, from top to bottom, the order that reads from left to right produces LL, LH, HL and the HH address of totally four subbands, and the address generation rule is with the wavelet transformation address generator.
Described computing pool unit is made up of with the data of storing two width of cloth images and two memory blocks by picture portion two memory blocks, work alone, when system start-up, at first to define the address subregion, defining first width of cloth image memory block first address is first_image_addr, defining second width of cloth image storage first address is second_image_addr, in the course of work, two width of cloth images write respectively in the space that above two addresses begin, after a certain width of cloth Flame Image Process in the space finished, new images just can write in the space of this width of cloth image.
Described wavelet transform unit is calculated the small echo one-level conversion of continuous 5 row images, 2 row are to calculate the result who finishes in the continuous 5 row images, other 3 row are the intermediate result of calculating, need new data could continue to calculate, after result's output that 2 row calculating are finished, after reading in 2 new line data, 2 guilds among the 3 interline results calculate and finish, and two line data that newly read in become intermediate result after as calculated.
Described coefficient writes back the unit and writes back address generator and low frequency and write back address generator and form by controlling small echo that wavelet transform unit and low frequency predicting unit write back the address of computing pool unit with wavelet coefficient respectively, this coefficient writes back the unit and exports the operation result of wavelet transform unit and low frequency predicting unit to the computing pool unit, wherein:
Small echo writes back address generator and connects image wavelet arithmetic unit and computing pool unit, writes back order with the wavelet transformation address generator.
Low frequency writes back address generator and connects low frequency predicting unit and computing pool unit, writes back order with low frequency predicted address generator.
Described low frequency predicting unit with each low frequency coefficient under with the condition of location computing adjacent with left and above the difference of coefficient mean value of adjacent 2 low frequency sub-band LL and LH as the predicted value of current coefficient the predicted value that reads all low frequency sub-band coefficients of order computation corresponding to low frequency predicted address generator.
Described vector quantization unit is divided into a coefficient sets with per 4 coefficients, and coefficient sets is carried out vector quantization, and the result is directly sent into arithmetic coding unit.
Described arithmetic coding unit is carried out arithmetic coding to the coefficient sets of vector quantization unit output, produces coding result.
The present invention relates to the method for compressing image of said apparatus, comprise the steps:
Reads image data is to wavelet transform unit from the computing pool unit for the first step, reading unit, and the computing pool unit is an internal memory of storage two width of cloth images;
Second step, wavelet transform unit are carried out wavelet transformation to the view data of reading in and are obtained wavelet coefficient; Be the economize on hardware expense, the RAM of wavelet transform unit inside only can buffer memory 5 row view data, read the update image data by constantly writing, finish until the entire image conversion;
The 3rd step, coefficient write back the unit wavelet coefficient are write back the computing pool unit;
According to from top to bottom, order from right to left reads low frequency coefficient to the low frequency predicting unit from the computing pool unit for the 4th step, reading unit, low frequency sub-band is carried out low frequency predict computing, and simultaneously, coefficient writes back the unit operation result is write back in the computing pool unit;
The 5th step, reading unit are read each subband to the vector quantization unit according to from senior to rudimentary and the order of LL, LH, HL, HH, and coding result is sent with coefficient grouping, quantification in the vector quantization unit;
The 6th step, coding result enter arithmetic coding unit, and arithmetic coding is encoded to the vector quantization result, output code flow.
In Fig. 1, image at first is written into computing pool unit (1), by reading unit (2) image or wavelet data is read, and is write harmless wavelet transform unit (4), low frequency predicting unit (5) and vector quantization unit (6); The result of wavelet transform unit and low frequency predicting unit writes back unit (3) by coefficient and writes back the computing pool unit; The output of vector quantization unit is connected to the input end of arithmetic coding unit (7), through the processing of arithmetic coding unit, and final output image output code flow.
In Fig. 2, the process that data stream is crossed four devices has been thought of as streamline.Because the computing pool unit is divided into two zones, each loads piece image, and load image is taken turns in two zones.In Fig. 2, laterally represent the time, 1 represents the treatment of picture process in zone 1, and 2 represent the treatment of picture process in zone 2.Each arrow elapsed time section is represented this width of cloth treatment of picture process.Vertically the dotted line place represents the image update in the computing pool unit.Each regional image is all read after vector quantization is finished to finish, and at this moment can be written into down piece image.
In wavelet transform unit, RAM can hold 5 row images, and after the wavelet transformation of two row was finished, unwanted two row coefficients of computing can be write back the computing pool unit, and new two guilds that need computing are read the unit and read in.Because the computing pond has the function that reads while write, therefore, can read in two row new in the limit, two row that limit output computing is finished with computing.
Through experimental verification, this device is behind the wavelet transformation that carries out same levels, and the lossless compress rate is higher by 3% than JPEG2000; This device has adopted the system architecture of vector quantization and wavelet transformation combination, makes native system promptly avoid the high complexity of JPEG2000 entropy coding, has strengthened compression performance simultaneously again.
Claims (8)
1. device based on the compression of images of wavelet transformation and arithmetic coding, it is characterized in that, comprise: reading unit, the computing pool unit, wavelet transform unit, coefficient writes back the unit, the low frequency predicting unit, vector quantization unit and arithmetic coding unit, wherein: the computing pool unit receive from the outside and buffer memory original image data and output image raw data to wavelet transform unit, wavelet transform unit carries out that the wavelet transformation computing obtains and writes back wavelet coefficients at different levels or the low frequency that the unit writes back the computing pool unit by coefficient predicting the outcome to reading unit, reading unit respectively with wavelet transform unit, the low frequency predicting unit link to each other with the vector quantization unit and the output image raw data and through the wavelet transform unit wavelet transformation computing obtains and writes back the unit by coefficient writing back the wavelet coefficients at different levels of computing pool unit to wavelet transform unit, reading unit is exported five-star low frequency wavelet coefficient tremendously low frequency predicting unit and output low frequency predicts the outcome to arithmetic coding unit, wavelet transform unit receives the original image data of reading unit output or removes the low frequency wavelet coefficients at different levels of highest low frequency, the height of the corresponding input of output wavelet coefficient grade behind wavelet transformation, low frequency wavelet coefficient to coefficient writes back the unit, coefficient writes back that the unit links to each other with the low frequency predicting unit with wavelet transform unit respectively and the wavelet coefficient and the low frequency predicting unit output ground low frequency that receive wavelet transform unit output predicts the outcome, predict the outcome to the computing pool unit through map addresses output wavelet coefficient and low frequency, output low frequency predicted the outcome to coefficient and writes back the unit after the low frequency predicting unit received the low frequency wavelet coefficient of reading unit output and carries out low frequency prediction computing, the output vector quantification symbol is to the coding unit that counts after the row vector quantization operations that the low frequency of vector quantization unit reception reading unit output predicts the outcome and other wavelet coefficients are gone forward side by side, and the coding unit that counts carries out the arithmetic coding computing to the vector quantization symbol and exports the code word of arithmetic coding.
2. the device of the compression of images based on wavelet transformation and arithmetic coding according to claim 1, it is characterized in that, described reading unit is controlled wavelet transform unit, low frequency predicting unit and vector quantization unit respectively by three the address generator that reads the address of computing pool unit is formed, wherein:
First address generator links to each other with the computing pool unit with the image wavelet arithmetic unit respectively, and: when the one-level wavelet transformation according to from top to bottom to image inside, from left to right read the order calculated address; According to from top to bottom, ranks are 1 at interval between the neighbor that reads order and read from left to right when the secondary wavelet transformation; By that analogy, the ranks that adopt when the third level and follow-up wavelet transformation are spaced apart 2
K-1-1, wherein k is a wavelet transformation level several;
Second address generator links to each other with the computing pool unit with the low frequency predicting unit respectively and calculates low frequency sub-band last cell address, according to image inside from top to bottom, from right to left read the address that order produces each low frequency sub-band element according to conversion progression;
The three-address generator link to each other with the computing pool unit with the vector quantization unit respectively and according to from senior to rudimentary, image is from top to bottom inner, the order that reads from left to right produces LL, LH, HL and the HH address of totally four subbands.
3. the device of the compression of images based on wavelet transformation and arithmetic coding according to claim 1, it is characterized in that, described computing pool unit is made up of to store the data of two width of cloth images two memory blocks, the data that write by the outside are YCbCr components of image, and each works alone according to the order of lining by line scan and two memory blocks.
4. the device of the compression of images based on wavelet transformation and arithmetic coding according to claim 1 and 2, it is characterized in that, described wavelet transform unit is calculated the small echo one-level conversion of continuous 5 row images, wherein: preceding 2 row are to calculate to finish the result in every continuous 5 row images, and back 3 row are to calculate intermediate result.
5. the device of the compression of images based on wavelet transformation and arithmetic coding according to claim 1, it is characterized in that, described coefficient writes back the unit and writes back address generator and low frequency and write back address generator and form by controlling small echo that wavelet transform unit and low frequency predicting unit write back the address of computing pool unit with wavelet coefficient respectively, this coefficient writes back the unit and exports the operation result of wavelet transform unit and low frequency predicting unit to the computing pool unit, wherein:
Small echo writes back address generator and connects image wavelet arithmetic unit and computing pool unit, writes back order with first address generator;
Low frequency writes back address generator and connects low frequency predicting unit and computing pool unit, writes back order with second address generator.
6. according to claim 1 or 5 based on the device of the compression of images of wavelet transformation and arithmetic coding, it is characterized in that, described low frequency predicting unit with each low frequency coefficient under with the condition of location computing adjacent with left and above the difference of coefficient mean value of adjacent 2 low frequency sub-band LL and LH as the predicted value of current coefficient the predicted value that reads all low frequency sub-band coefficients of order computation corresponding to second address generator.
7. the device of the compression of images based on wavelet transformation and arithmetic coding according to claim 1 and 2 is characterized in that described vector quantization unit is divided into coefficient sets with coefficient, and coefficient sets is carried out vector quantization, and the result is directly sent into arithmetic coding unit.
8. the method for compressing image according to the described device of above-mentioned arbitrary claim is characterized in that, comprises the steps:
Reads image data is to wavelet transform unit from the computing pool unit for the first step, reading unit, and the computing pool unit is an internal memory of storage two width of cloth images;
Second step, wavelet transform unit are carried out wavelet transformation to the view data of reading in and are obtained wavelet coefficient; The RAM of wavelet transform unit inside is buffer memory 5 row view data only, read the update image data by constantly writing, and finish until the entire image conversion;
The 3rd step, coefficient write back the unit wavelet coefficient are write back the computing pool unit;
According to from top to bottom, order from right to left reads low frequency coefficient to the low frequency predicting unit from the computing pool unit for the 4th step, reading unit, low frequency sub-band is carried out low frequency predict computing, and simultaneously, coefficient writes back the unit operation result is write back in the computing pool unit;
The 5th step, reading unit are read each subband to the vector quantization unit according to from senior to rudimentary and the order of LL, LH, HL, HH, and coding result is sent with coefficient grouping, quantification in the vector quantization unit;
The 6th step, coding result enter arithmetic coding unit, and arithmetic coding is encoded to the vector quantization result, output code flow.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110143696XA CN102222350A (en) | 2011-05-31 | 2011-05-31 | Image compressing device and method based on wavelet transformation and arithmetic coding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201110143696XA CN102222350A (en) | 2011-05-31 | 2011-05-31 | Image compressing device and method based on wavelet transformation and arithmetic coding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102222350A true CN102222350A (en) | 2011-10-19 |
Family
ID=44778894
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201110143696XA Pending CN102222350A (en) | 2011-05-31 | 2011-05-31 | Image compressing device and method based on wavelet transformation and arithmetic coding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102222350A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1571514A (en) * | 2003-07-23 | 2005-01-26 | 西北工业大学 | An embedded image compression technique based on wavelet transformation |
| CN101316364A (en) * | 2008-07-15 | 2008-12-03 | 南京信息工程大学 | Image compression system |
| CN101404772A (en) * | 2008-11-19 | 2009-04-08 | 中国科学院光电技术研究所 | VLSI image compression encoder based on wavelet transformation |
-
2011
- 2011-05-31 CN CN201110143696XA patent/CN102222350A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1571514A (en) * | 2003-07-23 | 2005-01-26 | 西北工业大学 | An embedded image compression technique based on wavelet transformation |
| CN101316364A (en) * | 2008-07-15 | 2008-12-03 | 南京信息工程大学 | Image compression system |
| CN101404772A (en) * | 2008-11-19 | 2009-04-08 | 中国科学院光电技术研究所 | VLSI image compression encoder based on wavelet transformation |
Non-Patent Citations (1)
| Title |
|---|
| 张砺佳: "基于小波变换的图像压缩编码研究", 《中国优秀硕士学位论文全文数据库》, 15 May 2008 (2008-05-15) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101543037B (en) | Information processing device and method | |
| Li et al. | Learning content-weighted deep image compression | |
| KR101625548B1 (en) | Context modeling techniques for transform coefficient level coding | |
| JP2013150215A (en) | Moving image decoding device, moving image encoding device, moving image decoding method, moving image encoding method, moving image decoding program, and moving image encoding program | |
| EP3386199A1 (en) | Lossless compression method and system appled to video hard decoding | |
| KR101645544B1 (en) | Methods of encoding and decoding using multi-level prediction and apparatuses for using the same | |
| JP2017507532A (en) | Recursive block partitioning | |
| US10652541B2 (en) | Method and device for encoding video data | |
| JP2008502261A (en) | Method for storing images in memory using a cost function including compression coding and power consumption | |
| CN101707716A (en) | Video coder and coding method | |
| CN103179389A (en) | Image processing apparatus and image processing method | |
| CN101083769A (en) | Coder/decoder and coding/decoding method for processing N bit video data | |
| US8787686B2 (en) | Image processing device and image processing method | |
| CN104113759B (en) | Video system, video frame buffer recompression/decompression method and device | |
| CN105245889B (en) | A kind of reference frame compression method based on stratified sampling | |
| Jeong et al. | An overhead-free region-based JPEG framework for task-driven image compression | |
| KR102257379B1 (en) | Video encoding circuit and video encoding method therewith | |
| CN1848960B (en) | Residual coding in compliance with a video standard using non-standardized vector quantization coder | |
| KR102071459B1 (en) | Encoding apparatus and encoding method | |
| CN102222350A (en) | Image compressing device and method based on wavelet transformation and arithmetic coding | |
| KR20100044333A (en) | Video encoding apparatus, and apparatus and method of 2 dimensional ordering transform for image signal, and recording medium therefor | |
| CN104079930A (en) | Achieving method of remote-sensing image compression system | |
| CN103327340A (en) | Method and device for searching integer | |
| US8861880B2 (en) | Image processing device and image processing method | |
| CN102131093A (en) | Image processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| ASS | Succession or assignment of patent right |
Owner name: INFRASTRUCTURE TESTING INSTITUTE OF CHINA ACADEMY Effective date: 20120222 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20120222 Address after: 200240 Dongchuan Road, Shanghai, No. 800, No. Applicant after: Shanghai Jiao Tong University Co-applicant after: Infrastructure Testing Institute of China Academy of Railway Sciences Address before: 200240 Dongchuan Road, Shanghai, No. 800, No. Applicant before: Shanghai Jiao Tong University |
|
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20111019 |