CN102186069A - Remote sensing image data compression method capable of maintaining measurement performance - Google Patents
Remote sensing image data compression method capable of maintaining measurement performance Download PDFInfo
- Publication number
- CN102186069A CN102186069A CN 201110007292 CN201110007292A CN102186069A CN 102186069 A CN102186069 A CN 102186069A CN 201110007292 CN201110007292 CN 201110007292 CN 201110007292 A CN201110007292 A CN 201110007292A CN 102186069 A CN102186069 A CN 102186069A
- Authority
- CN
- China
- Prior art keywords
- mrow
- coefficient
- remote sensing
- wavelet decomposition
- wavelet
- 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
- 238000000034 method Methods 0.000 title claims abstract description 93
- 238000005259 measurement Methods 0.000 title claims abstract description 21
- 238000013144 data compression Methods 0.000 title claims abstract description 19
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 135
- 230000008569 process Effects 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 13
- 230000009466 transformation Effects 0.000 claims abstract description 13
- 238000005311 autocorrelation function Methods 0.000 claims description 8
- 238000013139 quantization Methods 0.000 claims description 6
- 238000009826 distribution Methods 0.000 claims description 5
- 238000012986 modification Methods 0.000 claims description 2
- 230000004048 modification Effects 0.000 claims description 2
- 238000007906 compression Methods 0.000 abstract description 71
- 230000006835 compression Effects 0.000 abstract description 69
- 230000008859 change Effects 0.000 abstract description 3
- 238000011084 recovery Methods 0.000 abstract description 2
- 239000003550 marker Substances 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 12
- 238000000605 extraction Methods 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013500 data storage Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011031 large-scale manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000004393 visual impairment Effects 0.000 description 1
Images
Landscapes
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention discloses a remote sensing image data compression method capable of maintaining measurement performance. A remote sensing image is processed by adopting wavelet transform, so the method is suitable for the compression of the remote sensing image. The remote sensing image is converted and coded by adopting integer wavelet transform, so transformation from an integer to another integer can be realized, the inverse transformation of the transformation from the integer to another integer can completely reconstruct the image, the uncontrollable loss of image information is avoided, and integer-wavelet-transform-based lossless image compression becomes possible. The compression efficiency is improved, and a compression ratio is increased by improved embedded zerotree wavelet (EZW) coding methods for the establishment of a zerotree marker table, the integer updating processing of a wavelet decomposition coefficient, the change of a scanning direction of the wavelet decomposition coefficient, the independent processing of the lowest frequency sub-band coefficient and the like. Characteristic points of important characteristics in the remote sensing image are extracted, and are identified in a wavelet transform process, and an integer wavelet decomposition coefficient in a high frequency sub-band is predicted and compensated by utilizing an integer wavelet decomposition coefficient recovery value in a low frequency sub-band to improve the performance of measuring the decompressed image.
Description
Technical Field
The invention belongs to the technical field of remote sensing image data compression and digital signal processing, and relates to a wavelet transform of a high-resolution remote sensing image and an EZW (embedded zerotree wavelet coding) compression method for the remote sensing image, in particular to a remote sensing image data compression method for keeping measurement performance.
Background
With the continuous development of photoelectric technology, computer technology and remote sensing technology, more and more information acquired by remote sensing is data which exists and is processed in a digital form, and a more direct and richer data source is brought to digital photogrammetry. However, while the development of digital photogrammetry production technology, product quality and quantity is promoted, the mass remote sensing information data (mainly remote sensing image data) puts higher requirements on storage and transmission conditions, and also brings severe tests to the existing limited bandwidth and hardware equipment. Particularly, when the digital photogrammetry is developed in a networked and large-scale production mode, a large amount of original image data and intermediate image data need to be stored on a client workstation, and a large amount of remote sensing image data needs to be transmitted between different clients and between the clients and a server. Therefore, remote sensing image compression technology is receiving more and more attention.
Data compression is the representation of a signal from a source with a minimum of numbers to reduce the signal space that holds a given set of information or data samples, the signal space being the space, time, and frequency space occupied by a set of signals. The image compression is a compression coding technology which is adopted for coding an image source and trying to reduce the necessary digital code rate by deleting redundant or unnecessary information on the premise of ensuring the expected image quality. Image compression is also often referred to as image coding.
Digital image compression encoding technology dates back to the digitization of television signals proposed in 1948, and has been in history for over 50 years today. In the meantime, a plurality of image compression coding methods appeared, and particularly, in the late 80 s, due to the establishment of a wavelet transform theory, a fractal theory, an artificial neural network theory and a visual simulation theory, an image compression technology has not been developed before, wherein the image compression technology based on the wavelet transform is one of the hot spots of the current research.
Image compression provides an effective way to solve the problem of large amount of digital image data. The image compression can reduce the storage space of the image, improve the network transmission efficiency, reduce the storage and transmission cost, and play a decisive role in the real-time transmission of the digital image.
Generally, a compression ratio is higher in lossy compression than in lossless compression, and in common multimedia technologies, digital image contents are mostly human images, and a lossy compression technology is mostly adopted, that is, on the premise of not generating obvious image visual loss, the information content of the images is reduced as much as possible, so that the image compression with a high compression ratio is realized. And the remote sensing image is not compressed or adopts a lossless compression scheme, so that a higher compression ratio cannot be obtained. The prior remote sensing digital image is not compressed with loss mainly because:
1. the remote sensing image is difficult to obtain, and the existing aerospace remote sensing images all need to invest large manpower, financial resources and material resources to establish a remote sensing data acquisition system, so that the cost of the remote sensing image is high, and the information loss in the lossy compression process of the remote sensing image cannot be accepted;
2. the precision required for processing and applying the remote sensing image is high, the remote sensing image is not only used for observation of human eyes, but also used for searching, identifying and measuring of a computer, and if a general lossy compression mode is adopted, the application effect of the remote sensing image is influenced due to information loss.
Because the digital surveying and mapping guarantee information and the production and operation of products have the unique requirement on the data compression technology, if the remote sensing image data is compressed, the remote sensing image data compression method has a better visual effect, more importantly, the measurement performance of the decompressed image is kept, the cost spent on data storage and transfer can be reduced while the operation precision is guaranteed, the production efficiency is further improved, and the method has important significance particularly on the production of high-timeliness surveying and mapping guarantee products.
Disclosure of Invention
The invention aims to provide a method for compressing remote sensing image data with a measurement performance maintained, so as to meet the technical requirements in the fields of full-digital photogrammetry equipment, data storage and transmission in networked full-digital photogrammetry operation, surveying and mapping guarantee and the like.
A data compression method for maintaining the measurement performance of a digital remote sensing image is disclosed, wherein: the method comprises the following specific steps:
step 1), collecting digital remote sensing image signals, and counting parameters of the digital remote sensing image signals: mean, variance, entropy, mean energy, sharpness, autocorrelation coefficients;
let the digital remote sensing image be f (x, y), and its pixel f (i, j) be akK is {1, 2, 3, … L }, i and j are respectively the horizontal and vertical coordinates of the pixel, and the parameter L is the maximum gray value in the digital remote sensing image;
any pixel in digital remote sensing imageThe probability of occurrence is p (a)k) The method comprises the following steps:
mean value of
Refers to the average gray value of the digital remote sensing image signal, such as the formula (2):
variance σ2: the variance of the mean gray value of the digital remote sensing image signal reflects the discrete condition of the gray distribution, such as the formula (3):
energy E: the intensity of the overall gray scale of the digital remote sensing image is also called as a second moment, and the intensity is expressed by the formula (4):
average gradient g: the average gradient of the digital remote sensing image reflects the expressive force of the details of the digital remote sensing image and reflects the definition of the digital remote sensing image, and the formula (5) is as follows:
autocorrelation coefficient R (Δ x, Δ y): reflecting the correlation of the digital remote sensing image signals, and adopting a normalized autocorrelation function, wherein the expression is as shown in formula (6):
where, when Δ x is 1 or Δ y is 1, R (Δ x, Δ y) values represent autocorrelation function values between adjacent pixels, which are referred to as autocorrelation coefficients;
information entropy h (u): the average information content carried by each pixel in the digital remote sensing image is represented by the physical significance of the code length required by each coding symbol under the condition of lossless coding:
step 2), performing multi-level integer wavelet transform on the digital remote sensing image, establishing a wavelet tree structure, and determining integer wavelet decomposition coefficients after the wavelet transform of the digital remote sensing image; integer wavelet transform refers to discrete signals with wavelet decomposition coefficients being integers obtained after wavelet transform is carried out on discrete signals of an integer set; the integer wavelet decomposition coefficient has the formula:
wherein,
represents a pair d1，lA/2 rounding operation, wherein the coefficient d is transformed1，lAnd s1，lAre all integer sets;
after the digital remote sensing image is subjected to integer wavelet transform, a low-frequency sub-band and three high-frequency sub-bands with different resolutions are formed from left to right and from top to bottom; the integer wavelet decomposition coefficient of the low-frequency sub-band is called as a parent coefficient, all integer wavelet decomposition coefficients corresponding to the high-frequency sub-bands in all spatial directions are called as descendant coefficients of the parent coefficient, and thus, three tree structures with the parent coefficient as a vertex are formed;
step 3), carrying out embedded zerotree wavelet coding:
(31) selecting a group of wavelet decomposition coefficients XiThreshold T for judging importance0，T1…Tn-1The wavelet decomposition coefficients X are individually subjected toiPerforming importance judgment, wherein the threshold value is selected according to Ti＝T i-12 and an initial threshold value T0Satisfy | Xi|＜2To；
(32) Significant coefficient and insignificant coefficient: given a wavelet decomposition coefficient XiIf there is | X for a given threshold TiL > T, where T ∈ T0，T1…Tn-1Then the wavelet decomposition coefficient is an important coefficient; otherwise, the coefficient is an unimportant coefficient; in the embedded zerotree wavelet coding, the non-important coefficient is marked as zero; the corresponding non-important coefficient is a zero node in the wavelet tree, and the important coefficient is a non-zero node;
(33) and a zero tree: the zero tree represents the correlation based on the inter-subband wavelet decomposition coefficients: if the wavelet decomposition coefficient on the low-frequency sub-band is a non-important coefficient for the threshold T, all the wavelet decomposition coefficients in the high-frequency sub-band in each space direction corresponding to the low-frequency sub-band are also non-important coefficients for the threshold T, and the non-important coefficients are represented by a tree structure, namely, a zero tree;
(34) and a zero tree root: the zero node of the zero tree positioned in the lowest frequency sub-band is the zero tree root, and the zero tree root is represented by a symbol ZTR;
(35) main table and auxiliary table: during the encoding process, two separate, constantly updated tables are maintained: a primary table and a secondary table; the main table corresponding to a non-important set or fraction of the codeWave decomposition coefficients, and the sub-table is the effective information of the coding; storing important coefficients in each wavelet decomposition coefficient into a secondary table, storing non-important coefficients into a main table, and setting a threshold value T0；
(36) Main scanning and sub scanning:
main scanning: for all wavelet decomposition coefficients X in the main tableiMake a judgment if | XiIf is greater than threshold TiIf the node is an important coefficient; if XiOutputting a code ZTR for the zero tree root, and not scanning all the descendant coefficients of the zero tree root; if XiContinuously scanning all wavelet decomposition coefficients subsequent to the isolated zero point for the isolated zero point;
sub-scanning: wavelet decomposition coefficient Y in sublistiAll satisfy Ti＜Yi＜2Ti(ii) a The sub-scanning is performed by scanning all wavelet decomposition coefficients Y in the sub-tableiA different code "0" or "1" is output; satisfy Ti＜Yi＜3TiWavelet decomposition coefficient Y of/2i0 is output as the non-significant coefficient, and 3T is satisfiedi/2＜Yi＜2TiWavelet decomposition coefficient Y ofi Outputting 1 for the important coefficient;
step 4), compressing the digital remote sensing image by using improved embedded zerotree wavelet coding:
(41) establishing a zerotree mark table ZT, and initially setting all mark bits of the zerotree mark table ZT to be 1 before scanning; each element of the wavelet decomposition coefficient has a corresponding marking bit in a marking table, when an element is determined to be a zero tree root, all descendant coefficients of the zero tree are marked as '0' in the marking table, and meanwhile, the corresponding position of each non-important coefficient is marked as '0';
when each wavelet decomposition coefficient is scanned, firstly checking the sign of the wavelet decomposition coefficient in the zero tree mark table ZT, if the sign is '0', the wavelet decomposition coefficient is a descendant coefficient of a zero tree root, judging the wavelet decomposition coefficient is not carried out, and if the sign is 1, scanning is carried out;
marking bits of important coefficients in the scanning are all set to be 0, after the scanning is completed, all symbols of the zero tree marking table ZT are subjected to inverse logic operation, and the obtained new zero tree marking table ZT is used for the next scanning; skipping the important coefficient marked as '0' in the current scanning process without judging;
(42) recording the maximum value of the wavelet decomposition coefficient in each sub-band: completing extreme value statistics of wavelet decomposition coefficients in a sub-band of a new zero tree marking table ZT in the scanning process; counting the extreme values of the wavelet decomposition coefficients in each sub-band for determining the scanning threshold T0If the threshold value T of the scanning isiIf the value is larger than the extreme value of the wavelet decomposition coefficient of the sub-band, the sub-band does not need to be scanned;
(43) and independently processing the wavelet decomposition coefficients in the lowest frequency sub-band: because the lowest frequency sub-band contains most of the energy of the original digital remote sensing image, the wavelet decomposition coefficients in the lowest frequency sub-band after wavelet transformation are independently coded and do not participate in embedded zerotree wavelet coding scanning; forming three tree structures for the rest high-frequency sub-bands after wavelet transformation;
(44) changing the coding symbol:
in the successive approximation quantization process of the embedded zerotree wavelet coding, main scanning and sub-scanning are carried out simultaneously; for wavelet decomposition coefficient YiAnd a threshold value TiIf Y isi∈[Ti+Ti/2，2Ti) Output the coded symbol "111"; if Y isi∈[Ti，Ti+Ti/2), the symbol "110" is output if Yi∈[-Ti，-Ti-Ti/2), output symbol "100"; if Y isi∈[-Ti-Ti/2，2Ti) Output symbol "101"; if Y isiIs zero tree root, outputs symbol "00", if YiFor isolated zero, output symbol "01";
(45) and integer processing of wavelet decomposition coefficient updating:
threshold value sequence T in the successive approximation quantization process of embedded zero-tree wavelet coding0，T1…Tn-1The relationship between is Ti＝Ti-1Per 2, updating important coefficient and non-important coefficient in sub-scanning process, i.e. adjusting important coefficient YiIf Y isi∈[Ti，Ti+Ti/2), modifying the value at update
If Y isi∈[Ti+Ti/2，2Ti) Modifying the value at update
If the wavelet decomposition coefficient is an integer, when TiWhen 1 is satisfied, Y is inevitably 1. ltoreq. YiIf Y is < 2iIs a positive significant coefficient, then YiUpdated modification factor of 1
By TiAfter the threshold value of 1 is scanned, all wavelet decomposition coefficients are zero, and the integer wavelet transform-based embedded zero-tree wavelet coding is completed.
The data compression method for maintaining the measurement performance of the digital remote sensing image comprises the following steps: lower left corner LH of four sub-bands after wavelet decompositioniWhen wavelet decomposition coefficients in the high-frequency sub-band are scanned, scanning is performed from left to right in the order from top to bottom; the wavelet decomposition coefficients in the remaining one low frequency subband and two high frequency subbands are scanned sequentially from top to bottom in left-to-right order.
By adopting the technical scheme, the invention achieves the following technical effects:
the remote sensing image data compression method for maintaining the measurement performance has the following advantages:
1) the remote sensing image is processed by adopting the wavelet transformation, and the wavelet transformation has multi-resolution analysis capability and is beneficial to selection and judgment of a compression scheme, so the technical scheme of the invention is very suitable for compression of the remote sensing image;
2) the remote sensing image transform coding is carried out by adopting integer wavelet transform, so that the transform from integer to integer can be realized, the image can be completely reconstructed by the inverse transform, the uncontrollable loss of image information is avoided, and the lossless compression of the image based on the integer wavelet transform becomes possible;
3) the improved EZW coding method is realized by establishing a zero tree marking table, updating and processing wavelet decomposition coefficients by integer, changing the scanning direction of the wavelet decomposition coefficients, independently processing lowest frequency sub-band coefficients and the like, so that the compression efficiency and the compression ratio are improved;
4) and by extracting the point characteristics of the important characteristics in the remote sensing image, identifying the characteristic points in the wavelet transformation process, and predicting and compensating the high-frequency coefficient by using the integer wavelet low-frequency coefficient recovery value, the measurement performance of the decompressed image is improved.
Drawings
FIG. 1 is a flow chart of a method for compressing remote sensing image data with measurement capability maintained according to the present invention;
FIG. 2 is a tree structure of wavelet decomposition coefficients;
FIG. 3 is a grid scan of wavelet decomposition coefficients of a modified EZW method;
FIG. 4 is a remote sensing image of an original carrier;
FIG. 5 is a comparison graph of peak SNR of two reconstructed images with different compression ratios;
FIG. 6 is a graph comparing the feature numbers of the image (a) and the reconstructed image points in FIG. 4;
FIG. 7 is a comparison of the feature numbers of the image (b) and reconstructed image points of FIG. 4;
FIG. 8 is a diagram illustrating the relationship between the effective feature point and the compression ratio;
FIG. 9 is a diagram illustrating the relationship between the statistics of valid feature point errors and compression ratios;
FIG. 10 is a statistical value of two original images shown in FIG. 4;
FIG. 11 is a table comparing the efficiency of compression coding of the standard EZW method and the improved EZW method (unit: size, pixel; time, second);
FIG. 12 is a comparison table of peak SNR of reconstructed images (unit: PSNR, dB);
FIG. 13 is a table of compression ratio comparisons (unit: size, pixel) for different sizes.
Detailed Description
The invention provides a data compression method for maintaining the measurement performance of a remote sensing image, which comprises the following specific steps as shown in figure 1:
step 1), collecting remote sensing image data with different resolution sizes, such as a digitized aerial remote sensing image of an urban area with the actual size of 230 multiplied by 230mm and the scanning resolution of 25 mu m shown in a graph (a) in fig. 4, and a digitized aerial remote sensing image which is mainly vegetation with the actual size of 180 multiplied by 180mm and the scanning resolution of 25 mu m shown in a graph (b) in fig. 4. Acquiring some common characteristics of the remote sensing image, such as an average value, a variance, an entropy value, average energy, definition, an autocorrelation coefficient, information entropy and the like, as shown in fig. 10;
let the digital remote sensing image be f (x, y), and its pixel f (i, j) be akK is {1, 2, 3, … L }, i and j are respectively the abscissa and ordinate of the pixel, and L is the maximum gray-scale value in the image; the probability that any pixel in the remote sensing image appears in the remote sensing image is p (a)k) The method comprises the following steps:
mean value
Refers to the average gray scale value of the image signal, as in equation (2);
variance σ2: the variance of the gray values of the remote sensing image reflects the discrete condition of gray distribution, such as formula (3);
energy E: the intensity of the whole gray scale of the remote sensing image is expressed by the second moment, and the energy can be calculated for the block image, such as the formula (4);
average gradient g: the average gradient of the image can reflect the expressive force of image details, and shows whether the image is clear or not, such as formula (5);
autocorrelation coefficient R (Δ x, Δ y): reflecting the correlation of image signals, a normalized autocorrelation function is commonly used, and the expression is as the following expression (6):
where, when Δ x is 1 or Δ y is 1, R (Δ x, Δ y) values represent autocorrelation function values between adjacent pixels, which are referred to as autocorrelation coefficients;
information entropy h (u): the average information quantity carried by each symbol in the image signal is indicated, and the physical significance of the average information quantity represents the code length required by each coding symbol under the condition of lossless coding;
step 2), performing multi-level integer wavelet transform on the remote sensing image, wherein integer wavelet transform refers to discrete signals of an integer set, and discrete signals with wavelet decomposition coefficients being integers are obtained after wavelet transform; the integer wavelet decomposition coefficient has the formula:
wherein
Represents a pair d1，lA/2 rounding operation to transform the coefficient d1，lAnd s1，lAre all integer sets;
after the remote sensing image is subjected to integer wavelet decomposition, a low-frequency sub-band and three high-frequency sub-bands with different resolutions are formed in the sequence from left to right and from top to bottom; the wavelet decomposition coefficients in the low frequency sub-band are called parent coefficients, all wavelet decomposition coefficients in the high frequency sub-band corresponding to each spatial direction are called descendant coefficients of the parent coefficients, and thus, three tree structures with the parent coefficients as vertexes are formed, as shown in fig. 2; step 3), compressing the remote sensing image by using the improved EZW code:
(31) selecting a group of wavelet decomposition coefficients XiThreshold T for judging importance0，T1…Tn-1The wavelet decomposition coefficients X are individually subjected toiPerforming importance judgment, wherein the threshold value is selected according to Ti＝T i-12 and an initial threshold value T0Satisfy | Xi|＜2To；
(32) Significant coefficient and insignificant coefficient: given a wavelet decomposition coefficient XiIf there is | X for a given threshold TiL > T, where T ∈ T0，T1…Tn-1Then the wavelet decomposition coefficient is an important coefficient; otherwise, the coefficient is an unimportant coefficient; in the embedded zerotree wavelet coding, the non-important coefficient is marked as zero; the corresponding non-important coefficient is a zero node in the wavelet tree, and the important coefficient is a non-zero node;
(33) and a zero tree: the zero tree represents the correlation based on the inter-subband wavelet decomposition coefficients: if the wavelet decomposition coefficient on the low-frequency sub-band is a non-important coefficient for the threshold T, all the wavelet decomposition coefficients in the high-frequency sub-band in each space direction corresponding to the low-frequency sub-band are also non-important coefficients for the threshold T, and the non-important coefficients are represented by a tree structure, namely, a zero tree;
(34) and a zero tree root: the zero node of the zero tree positioned in the lowest frequency sub-band is the zero tree root, and the zero tree root is represented by a symbol ZTR;
(35) main table and auxiliary table: during the encoding process, two separate, constantly updated tables are maintained: a primary table and a secondary table. The primary table corresponds to a non-significant set or non-significant coefficients of the code, while the secondary table is the significant information of the code. Storing important coefficients in each wavelet decomposition coefficient into a secondary table, storing non-important coefficients into a main table, and setting a threshold value T0；
(36) Main scanning and sub scanning:
main scanning: for all wavelet decomposition coefficients X in the main tableiMake a judgment if | XiIf is greater than threshold TiIf the node is an important coefficient; if XiOutputting a code ZTR for the zero tree root, and not scanning all the descendant coefficients of the zero tree root; if XiContinuously scanning all wavelet decomposition coefficients subsequent to the isolated zero point for the isolated zero point;
sub-scanning: wavelet decomposition coefficient Y in sublistiAll satisfy Ti＜Yi＜2Ti(ii) a The sub-scanning is performed by scanning all wavelet decomposition coefficients Y in the sub-tableiA different code "0" or "1" is output; satisfy Ti＜Yi＜3TiWavelet decomposition coefficient Y of/2i0 is output as the non-significant coefficient, and 3T is satisfiedi/2＜Yi＜2TiWavelet decomposition coefficient Y ofiFor the significant coefficient, 1 is output.
Step 4), compressing the digital remote sensing image by using improved embedded zerotree wavelet coding:
(41) establishing a marking table ZT (zero Tree):
namely, establishing a marking table of the zero tree; each element of the wavelet decomposition coefficient has a corresponding mark bit in a mark table, when an element is determined to be a zero tree root, all descendant coefficients of the zero tree are marked with '0' in the mark table, and meanwhile, the corresponding position of each non-important coefficient is marked with '0';
before scanning, all mark bits of a zerotree mark table ZT are initially set to be 1, when each wavelet decomposition coefficient is scanned, the sign of the wavelet decomposition coefficient in the zerotree mark table ZT is checked, if the sign is 0, the wavelet decomposition coefficient is a descendant coefficient of a zerotree root, the wavelet decomposition coefficient is not judged, and if the sign is 1, scanning is carried out;
at this time, the marking bits of the important coefficients in the current scanning are all set to be 0, the important coefficients marked as 0 in the current scanning are skipped over in the next scanning process, and no judgment is made;
after the scanning is finished, performing inverse logic operation on all symbols of the zero tree marking table ZT to obtain a new zero tree marking table ZT for the next scanning;
(42) recording the maxima of each sub-band wavelet decomposition coefficient: completing extreme value statistics of wavelet decomposition coefficients in a sub-band of a new zero tree marking table ZT in the scanning process; the extreme values of the wavelet decomposition coefficients in the counted sub-bands can be directly used for determining the scanning threshold T0If the threshold value T of the scanning isiIf the maximum value of the wavelet decomposition coefficient in the sub-band is larger than the maximum value of the wavelet decomposition coefficient in the sub-band, the sub-band does not need to be scanned;
(43) and (3) independently processing the small wavelength resolution coefficients in the lowest frequency sub-band: because the lowest frequency sub-band contains most of the energy of the original image, the wavelet decomposition coefficient in the lowest frequency sub-band after wavelet transformation is independently coded and does not participate in EZW scanning; the rest high-frequency sub-bands can form three tree structures;
(44) changing the coding sign:
in the successive approximation quantization process of the embedded zerotree wavelet coding, main scanning and sub-scanning are carried out simultaneously; for wavelet decomposition coefficient YiAnd a threshold value TiIf Y isi∈[Ti+Ti/2，2Ti) Output the coded symbol "111"; if Y isi∈[Ti，Ti+Ti/2), the symbol "110" is output if Yi∈[-Ti，-Ti-Ti/2), output symbol "100"; if Y isi∈[-Ti-Ti/2，2Ti) Output symbol "101"; if Y isiIs zero tree root, outputs symbol "00", if YiFor isolated zero, output symbol "01";
(45) integer processing of wavelet decomposition coefficient updating:
threshold value sequence T in the successive approximation quantization process of embedded zero-tree wavelet coding0，T1…Tn-1The relationship between is Ti＝Ti-1Per 2, updating important coefficient and non-important coefficient in sub-scanning process, i.e. adjusting important coefficient YiIf Y isi∈[Ti，Ti+Ti/2), modify value Y at update
If Y isi∈[Ti+Ti/2，2Ti) Modifying the value at update
If the wavelet decomposition coefficient is an integer, when TiWhen 1 is satisfied, Y is inevitably 1. ltoreq. YiIf Y is < 2iIs a positive significant coefficient, then YiUpdated modified wavelet decomposition coefficient Y1
By TiAfter the threshold value of 1 is scanned, all wavelet decomposition coefficients are zero, and the integer wavelet transform-based embedded zero-tree wavelet coding is completed.
Scanning direction of the change coefficient:
the traditional scanning mode of EZW coding for wavelet decomposition coefficients is that each subband is scanned in the same direction, namely, the upper left corner is taken as a starting point, and the wavelet decomposition coefficients are scanned one by one from left to right and from top to bottom; until all sub-bands are scanned;
the invention changes the scanning mode of wavelet decomposition coefficient, i.e. the lower left corner LH of four sub-bands after wavelet decompositioniWhen scanning the wavelet decomposition coefficients in the sub-bands, see fig. 3, the wavelet decomposition coefficients in the other sub-bands are scanned in the order from top to bottom from left to right, and the wavelet decomposition coefficients in the other sub-bands are scanned in the order from top to bottom from left to rightThe decomposition coefficients are still scanned from top to bottom in left to right order; as shown in FIG. 3;
this is because LH is found by analysis of the wavelet decomposition coefficientsiThe wavelet decomposition coefficients in the sub-band have strong horizontal direction correlation and weak vertical direction correlation, and are in LHiThe sub-bands are scanned from left to right in sequence from top to bottom, the characteristic that wavelet decomposition coefficients in the high-frequency sub-bands have different correlations in different directions is fully considered, the correlation of adjacent scanning results is enhanced, and the correlation of the wavelet decomposition coefficients is further removed;
by combining the relevant characteristics of the remote sensing image, the digital remote sensing image is compressed by adopting the remote sensing image improved EZW coding compression method, and the digital remote sensing image is compared with a standard EZW coding method in the aspects of running speed and quality effect of a reconstructed image;
(1) compression speed comparison of the standard EZW method and the improved method:
the radiation resolution of the digital aerial remote sensing image is 8 bit; compressing images with different sizes and calculating the time used in the encoding process; FIG. 11 is a comparison of the compression efficiency of the conventional EZW encoding method and the improved method of the present invention for remote sensing images with different sizes, and it can be seen from FIG. 11 that the compression efficiency of the conventional EZW encoding method and the improved method of the present invention is very close to that of the remote sensing images with different sizes; in order to eliminate the influence of the system on the randomness of the compression process, two methods are alternately used for carrying out multiple compression experiments on the image with the same size, and the compression time of the image with the size of a single method is the average value of the multiple compression times;
(2) the compression effect comparison of the EZW method and the improved method:
the experimental data is an area part with the size of 4096 multiplied by 4096 pixels in the digital aerial remote sensing image, and the radiation resolution is 8 bit; the compression method used is a standard EZW coding method and an improved method;
verification of lossy compression:
the remote sensing images are compressed by two methods according to different compression ratios respectively, the image quality after the two methods are decompressed under the same compression ratio is compared, FIG. 12 is an image peak signal-to-noise ratio comparison table after the traditional EZW coding method and the improved method of the invention are compressed and reconstructed, and as seen from FIG. 12, the compression ratio is 4: 1 to 64: 1; the signal-to-noise ratio of the improved method is larger than that of the traditional EZW coding method, and when the compression ratio is 32: 1, the difference between the signal-to-noise ratio of the improved method and the traditional EZW coding method is the largest.
For the same original image data, under the condition of different compression ratios, the peak signal-to-noise ratio (PSNR) of the reconstructed image compressed by the improved method is 0.2-0.57 higher than that of the reconstructed image compressed by the traditional EZW method, and the effectiveness of the improved method is verified;
verification of lossless compression:
respectively carrying out lossless compression on remote sensing images with different sizes by using two methods, and comparing the maximum lossless compression ratios which can be obtained by using the two methods, as shown in FIG. 13; when the size of the obtained remote sensing image is small, the data size of the remote sensing image is small, the lossless compression capacity of the traditional EZW method and the improved method of the invention is poor, and the compression ratio of the two methods is increased along with the increase of the image size; the following conclusions are drawn by combining the characteristics of the remote sensing image:
(1) the remote sensing image has more details and large information amount, and the correlation is weak when the image data is small; is not beneficial to lossless compression;
(2) the larger the size of the image is, the more redundancy of image information is, and the larger the compression ratio of the image is;
(3) the compression ratio of the improved method is better than that of the standard EZW coding method on the whole, and the larger the image size is, the more obvious the superiority is.
Performing point feature extraction on the remote sensing image by using a point feature extraction operator (such as a Moravec operator, a Forstner operator and a Harris operator), recording position information of the extracted feature points, and evaluating image quality before and after compression and information loss degree according to comparison of the feature points;
the Harris operator is inspired by an autocorrelation function in signal processing, and a matrix M related to the autocorrelation function is given; the Harris method considers that the characteristic point is a pixel point corresponding to the maximum interest value in the local range; therefore, after the interest values of all points in the image are calculated, all points with the maximum local interest values are extracted from the original image; in actual operation, a certain neighborhood of a pixel, namely a window with a certain size, can be selected, and if the interest value of the pixel is larger than the interest values of other pixels in the window, the pixel can be judged to be a characteristic point;
calculating the amplitude and distribution rule of integer wavelet decomposition coefficients on different resolution layers at the positions of the feature points; determining the importance of the wavelet decomposition coefficient in the local range of 9 multiplied by 9 pixels with the characteristic point as the center according to the position information of the characteristic point or the amplitude and the distribution rule of the wavelet decomposition coefficient; the characteristic points are identified in the wavelet transform process through statistics of the importance of the decomposition coefficients of the integer wavelet of each frequency band after the integer wavelet transform of the remote sensing image, so that the characteristic points of the decompressed image have better measurement performance;
and (3) obtaining a restored image by adopting integer wavelet inverse transformation, wherein the integer wavelet inverse transformation process comprises the following steps:
analyzing the influence of image compression on image characteristics by comparing the number of point characteristic extraction, errors and other results;
scanning and digitizing the aerial image according to a resolution 256 gray level of 25 μm, and respectively selecting two urban parts and suburban parts with 1024 × 1024 pixels from the two original remote sensing images in fig. 4 as experimental data;
(1) calculating peak signal-to-noise ratios of the reconstructed images of the two remote sensing images at different compression ratios by adopting formulas (10), (11) and (12); FIG. 5 is a comparison graph of peak signal-to-noise ratios of two reconstructed remote sensing images with different compression ratios;
mean square error MSE:
wherein, M and N are respectively the image size calculated by taking the pixel as a unit, and S (i, j) and S' (x, y) are respectively the gray values of the original image and the reconstructed image at the pixel point (i, j);
signal-to-noise ratio SNR:
peak signal-to-noise ratio PSNR:
wherein
Is the average power of the original image,
the peak of the original signal.
(2) Influence of image compression on point feature extraction points
An improved EZW coding method is used as a compression tool of the remote sensing image, and a Harris operator is selected as a point feature extraction operator; the calculation window of the Harris operator is 5 × 5 pixels, σ of the gaussian template is 0.8, the suppression window of the maximum value is 15 × 15 pixels, and the threshold T is 5 × 107(ii) a Respectively extracting point characteristics of the original image and the decompressed reconstructed image by using a Harris operator, recording the quantity and position information of the point characteristics, and analyzing an experimental result;
comparing the feature points extracted from the decompressed reconstructed image with the standard by taking the number and the positions of the feature points extracted from the original image as the standard; wherein, the reconstructed image characteristic points which have the same positions with the original image characteristic points are compared and called the same points, the reconstructed image characteristic points which can not find the corresponding characteristic points in the original image are called error points, wherein, the point error distance siThe unit is a pixel, and the unit is an error value of the feature point extracted from the reconstructed image relative to the feature point extracted from the original image; the formula (13) is as follows:
if siIf the number of pixels is more than 5, the pixels are marked as error characteristic points of a reconstructed image, as shown in FIGS. 6 and 7, and FIG. 6 is a comparison graph of the characteristic numbers of the image (a) and the reconstructed image points in FIG. 4; FIG. 7 is a comparison of the feature numbers of the image (b) and the reconstructed image points of FIG. 4.
(3) Influence of image compression on extracted effective points
If the feature point P is extracted on the reconstructed image, relative to the number and position of the feature points extracted on the original imagerWith a certain feature point P on the original imageoIs a distance of error siLess than or equal to 5 pixels, can be regarded as PrIs a valid feature point; FIG. 8 is a diagram showing the relationship between the number of effective feature points and the ratio of the total number of feature points of a reconstructed image and the compression ratio, and FIG. 9 is a diagram showing the relationship between the mean value and variance of error distances obtained by counting the effective feature points and an original image; description of the analysis: when the compression ratio is increased, not only the characteristic information of the original image is influenced, but also the error characteristic expressed by introduced noise is reduced; when the compression ratio is less than 10: 1, the mean value and the variance of the error of the effective point of the reconstructed image are obviously increased, and when the compression ratio is larger, the mean value and the variance have no obvious change, which shows that the compression ratio is increased to a certain degree, the influence on the characteristic information of the effective point is reduced, and the error tends to be stable.
The invention provides an improved EZW coding method aiming at large-format remote sensing image data on the basis of improving the insufficient standard EZW coding, and realizes a remote sensing image data compression method keeping the measurement performance by combining the point characteristic extraction of the remote sensing image; through practical application, the decompressed image has better measurement accuracy; the method is used for production and operation of digital surveying and mapping guarantee information and products, the operation precision is guaranteed, the production efficiency is further improved, and the method has good practical value.
Claims (6)
1. A data compression method for maintaining the measurement performance of a digital remote sensing image is characterized in that: the method comprises the following specific steps:
step 1), collecting digital remote sensing image signals, and counting parameters of the digital remote sensing image signals: mean, variance, entropy, mean energy, sharpness, autocorrelation coefficients;
step 2), performing multi-level integer wavelet transform on the digital remote sensing image, establishing a wavelet tree structure, and determining integer wavelet decomposition coefficients after the wavelet transform of the digital remote sensing image;
after the digital remote sensing image is subjected to integer wavelet transform, a low-frequency sub-band and three high-frequency sub-bands with different resolutions are formed from left to right and from top to bottom; the integer wavelet decomposition coefficient of the low-frequency sub-band is called as a parent coefficient, all integer wavelet decomposition coefficients corresponding to the high-frequency sub-bands in all spatial directions are called as descendant coefficients of the parent coefficient, and thus, three tree structures with the parent coefficient as a vertex are formed;
step 3), carrying out embedded zerotree wavelet coding:
and 4) compressing the digital remote sensing image by using the improved embedded zerotree wavelet coding.
2. The data compression method for maintaining the measurement performance of the digital remote sensing image according to claim 1, characterized in that: lower left corner LH of four sub-bands after wavelet decompositioniWhen wavelet decomposition coefficients in the high-frequency sub-band are scanned, scanning is performed from left to right in the order from top to bottom; the wavelet decomposition coefficients in the remaining one low frequency subband and two high frequency subbands are scanned sequentially from top to bottom in left-to-right order.
3. The data compression method for maintaining the measurement performance of the digital remote sensing image according to claim 1 or 2, characterized in that: in step 1), the average value, the variance, the entropy value, the average energy, the definition and the autocorrelation coefficient are specifically calculated as follows:
let the digital remote sensing image be f (x, y), and its pixel f (i, j) be akK is {1, 2, 3, … L }, i and j are respectively the horizontal and vertical coordinates of the pixel, and the parameter L is the maximum gray value in the digital remote sensing image;
the probability that any pixel in the digital remote sensing image appears in the digital remote sensing image is p (a)k) The method comprises the following steps:
mean value of
Refers to the average gray value of the digital remote sensing image signal, such as the formula (2):
variance σ2: the variance of the mean gray value of the digital remote sensing image signal reflects the discrete condition of the gray distribution, such as the formula (3):
energy E: the intensity of the overall gray scale of the digital remote sensing image is also called as a second moment, and the intensity is expressed by the formula (4):
average gradient g: the average gradient of the digital remote sensing image reflects the expressive force of the details of the digital remote sensing image and reflects the definition of the digital remote sensing image, and the formula (5) is as follows:
autocorrelation coefficient R (Δ x, Δ y): reflecting the correlation of the digital remote sensing image signals, and adopting a normalized autocorrelation function, wherein the expression is as shown in formula (6):
where, when Δ x is 1 or Δ y is 1, R (Δ x, Δ y) values represent autocorrelation function values between adjacent pixels, which are referred to as autocorrelation coefficients;
information entropy h (u): the average information content carried by each pixel in the digital remote sensing image is represented by the physical significance of the code length required by each coding symbol under the condition of lossless coding:
4. the data compression method for maintaining the measurement performance of the digital remote sensing image according to claim 3, wherein: the integer wavelet transform in the step 2) means that discrete signals of an integer set are still discrete signals with wavelet decomposition coefficients being integers after the discrete signals are subjected to wavelet transform;
the integer wavelet decomposition coefficient has the formula:
wherein,
represents a pair d1，lA/2 rounding operation, wherein the coefficient d is transformed1，lAnd s1，lAre all integer sets.
5. The data compression method for maintaining the measurement performance of the digital remote sensing image according to claim 4, wherein: the specific steps of performing the embedded zerotree wavelet coding in the step 3) are as follows:
(31) selecting a group of wavelet decomposition coefficients XiThreshold T for judging importance0，T1…Tn-1The wavelet decomposition coefficients X are individually subjected toiPerforming importance judgment, wherein the threshold value is selected according to Ti＝Ti-12 and an initial threshold value T0Satisfy | Xi|＜2To；
(32) Significant coefficient and insignificant coefficient: given a wavelet decomposition coefficient XiIf there is | X for a given threshold TiL > T, where T ∈ T0，T1…Tn-1If not, the wavelet decomposition coefficient is judged to be a non-important coefficient; in the embedded zerotree wavelet coding, the non-important coefficient is marked as zero; the corresponding non-important coefficient is a zero node in the wavelet tree, and the important coefficient is a non-zero node;
(33) and a zero tree: the zero tree represents the correlation based on the inter-subband wavelet decomposition coefficients: if the wavelet decomposition coefficient on the low-frequency sub-band is a non-important coefficient for the threshold T, all the wavelet decomposition coefficients in the high-frequency sub-band in each space direction corresponding to the low-frequency sub-band are also non-important coefficients for the threshold T, and the non-important coefficients are represented by a tree structure, namely, a zero tree;
(34) and a zero tree root: the zero node of the zero tree positioned in the lowest frequency sub-band is the zero tree root, and the zero tree root is represented by a symbol ZTR;
(35) main table and auxiliary table: during the encoding process, two separate, constantly updated tables are maintained: a primary table and a secondary table; the primary table corresponds to the unimportant set of the code or wavelet decomposition coefficients, while the secondary table is the valid information of the code; storing important coefficients in each wavelet decomposition coefficient into a secondary table, storing non-important coefficients into a main table, and setting a threshold value T0；
(36) Main scanning and sub scanning:
main scanning: for all wavelet decomposition coefficients X in the main tableiMake a judgment if | XiIf is greater than threshold TiIf the node is an important coefficient; if XiOutputting a code ZTR for the zero tree root, and not scanning all the descendant coefficients of the zero tree root; if XiContinuously scanning all wavelet decomposition coefficients subsequent to the isolated zero point for the isolated zero point;
sub-scanning: wavelet decomposition coefficient Y in sublistiAll satisfy Ti＜Yi＜2Ti(ii) a The sub-scanning is performed by scanning all wavelet decomposition coefficients Y in the sub-tableiA different code "0" or "1" is output; satisfy Ti＜Yi＜3TiWavelet decomposition coefficient Y of/2i0 is output as the non-significant coefficient, and 3T is satisfiedi/2＜Yi＜2TiWavelet decomposition coefficient Y ofiFor the significant coefficient, 1 is output.
6. The data compression method for maintaining the measurement performance of the digital remote sensing image according to claim 5, wherein: the specific steps of compressing the digital remote sensing image by using the improved embedded zerotree wavelet coding in the step 4) are as follows:
(41) establishing a zerotree mark table ZT, and initially setting all mark bits of the zerotree mark table ZT to be 1 before scanning; each element of the wavelet decomposition coefficient has a corresponding marking bit in a marking table, when an element is determined to be a zero tree root, all descendant coefficients of the zero tree are marked as '0' in the marking table, and meanwhile, the corresponding position of each non-important coefficient is marked as '0';
when each wavelet decomposition coefficient is scanned, firstly checking the sign of the wavelet decomposition coefficient in the zero tree mark table ZT, if the sign is '0', the wavelet decomposition coefficient is a descendant coefficient of a zero tree root, judging the wavelet decomposition coefficient is not carried out, and if the sign is 1, scanning is carried out;
marking bits of important coefficients in the scanning are all set to be 0, after the scanning is completed, all symbols of the zero tree marking table ZT are subjected to inverse logic operation, and the obtained new zero tree marking table ZT is used for the next scanning; skipping the important coefficient marked as '0' in the current scanning process without judging;
(42) recording the maximum value of the wavelet decomposition coefficient in each sub-band: completing extreme value statistics of wavelet decomposition coefficients in a sub-band of a new zero tree marking table ZT in the scanning process; counting the extreme values of the wavelet decomposition coefficients in each sub-band for determining the scanning threshold T0If the threshold value T of the scanning isiIf the value is larger than the extreme value of the wavelet decomposition coefficient of the sub-band, the sub-band does not need to be scanned;
(43) and independently processing the wavelet decomposition coefficients in the lowest frequency sub-band: because the lowest frequency sub-band contains most of the energy of the original digital remote sensing image, the wavelet decomposition coefficients in the lowest frequency sub-band after wavelet transformation are independently coded and do not participate in embedded zerotree wavelet coding scanning; forming three tree structures for the rest high-frequency sub-bands after wavelet transformation;
(44) changing the coding symbol:
in the successive approximation quantization process of the embedded zerotree wavelet coding, main scanning and sub-scanning are carried out simultaneously; for wavelet decomposition coefficient YiAnd a threshold value TiIf Y isi∈[Ti+Ti/2，2Ti) Output the coded symbol "111"; if Y isi∈[Ti，Ti+Ti/2), the symbol "110" is output if Yi∈[-Ti，-Ti-Ti/2), output symbol "100"; if Y isi∈[-Ti-Ti/2，2Ti) Output symbol "101"; if Y isiIs zero tree root, outputs symbol "00", if YiFor isolated zero, output symbol "01";
(45) and integer processing of wavelet decomposition coefficient updating:
threshold value sequence T in the successive approximation quantization process of embedded zero-tree wavelet coding0，T1…Tn-1The relationship between is Ti＝Ti-1Per 2, updating important coefficient and non-important coefficient in sub-scanning process, i.e. adjusting important coefficient YiIf Y isi∈[Ti，Ti+Ti/2), modifying the value at updateIf Y isi∈[Ti+Ti/2，2Ti) Modifying the value at update
If the wavelet decomposition coefficient is an integer, when TiWhen 1 is satisfied, Y is inevitably 1. ltoreq. YiIf Y is < 2iIs a positive significant coefficient, then YiUpdated modification factor of 1
By TiAnd after the threshold value of 1 is scanned, all wavelet decomposition coefficients are zero, and the integer wavelet transform-based embedded zero-tree wavelet coding is completed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110007292 CN102186069B (en) | 2011-01-14 | 2011-01-14 | Remote sensing image data compression method capable of maintaining measurement performance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110007292 CN102186069B (en) | 2011-01-14 | 2011-01-14 | Remote sensing image data compression method capable of maintaining measurement performance |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102186069A true CN102186069A (en) | 2011-09-14 |
| CN102186069B CN102186069B (en) | 2013-05-08 |
Family
ID=44572130
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110007292 Expired - Fee Related CN102186069B (en) | 2011-01-14 | 2011-01-14 | Remote sensing image data compression method capable of maintaining measurement performance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102186069B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077543A (en) * | 2012-12-29 | 2013-05-01 | 浙江工业大学 | Embedded zerotree coding method on basis of inverse Loop subdivision |
| EP3035685A1 (en) * | 2014-12-16 | 2016-06-22 | Thomson Licensing | A device and a method for encoding an image and corresponding decoding method and decoding device |
| EP3035687A1 (en) * | 2014-12-16 | 2016-06-22 | Thomson Licensing | A device and a method for encoding an image and corresponding decoding method and decoding device |
| EP3035686A1 (en) * | 2014-12-16 | 2016-06-22 | Thomson Licensing | A device and a method for encoding an image and corresponding decoding method and decoding device |
| CN106018325A (en) * | 2016-04-29 | 2016-10-12 | 南京富岛信息工程有限公司 | Method for evaluating credibility of gasoline property modeling prediction result |
| CN107819469A (en) * | 2016-09-13 | 2018-03-20 | 达索系统公司 | To representing that the signal of physical attribute is compressed |
| CN109857709A (en) * | 2019-02-22 | 2019-06-07 | 中国人民解放军空军研究院战略预警研究所 | A kind of data compression method and apparatus |
| WO2019129032A1 (en) * | 2017-12-26 | 2019-07-04 | 北京市商汤科技开发有限公司 | Remote sensing image recognition method and apparatus, storage medium and electronic device |
| CN112257621A (en) * | 2020-10-28 | 2021-01-22 | 贵州电网有限责任公司 | Equipment image identification method for unmanned aerial vehicle inspection |
| CN113326744A (en) * | 2021-05-12 | 2021-08-31 | 哈尔滨工业大学 | Method and system for detecting on-orbit state abnormity of spacecraft |
| CN117395381A (en) * | 2023-12-12 | 2024-01-12 | 上海卫星互联网研究院有限公司 | Compression method, device and equipment for telemetry data |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1431523A (en) * | 2003-01-16 | 2003-07-23 | 上海交通大学 | Optimization method of remote sensing image by use of syncretic bilinear interpolation and wavelet transformation |
| US20040234146A1 (en) * | 1999-10-28 | 2004-11-25 | Egbert Ammicht | Coefficient computation in image compression using set partitioning in hierarchical trees |
| CN1770201A (en) * | 2004-11-05 | 2006-05-10 | 北京师范大学 | Adjustable remote sensing image fusion method based on wavelet transform |
| CN101068358A (en) * | 2007-05-24 | 2007-11-07 | 北京航空航天大学 | Image compression-oriented small wave base sorting structureal method |
| US20080037880A1 (en) * | 2006-08-11 | 2008-02-14 | Lcj Enterprises Llc | Scalable, progressive image compression and archiving system over a low bit rate internet protocol network |
| CN101835045A (en) * | 2010-05-05 | 2010-09-15 | 哈尔滨工业大学 | Hi-fidelity remote sensing image compression and resolution ratio enhancement joint treatment method |
-
2011
- 2011-01-14 CN CN 201110007292 patent/CN102186069B/en not_active Expired - Fee Related
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040234146A1 (en) * | 1999-10-28 | 2004-11-25 | Egbert Ammicht | Coefficient computation in image compression using set partitioning in hierarchical trees |
| CN1431523A (en) * | 2003-01-16 | 2003-07-23 | 上海交通大学 | Optimization method of remote sensing image by use of syncretic bilinear interpolation and wavelet transformation |
| CN1770201A (en) * | 2004-11-05 | 2006-05-10 | 北京师范大学 | Adjustable remote sensing image fusion method based on wavelet transform |
| US20080037880A1 (en) * | 2006-08-11 | 2008-02-14 | Lcj Enterprises Llc | Scalable, progressive image compression and archiving system over a low bit rate internet protocol network |
| CN101068358A (en) * | 2007-05-24 | 2007-11-07 | 北京航空航天大学 | Image compression-oriented small wave base sorting structureal method |
| CN101835045A (en) * | 2010-05-05 | 2010-09-15 | 哈尔滨工业大学 | Hi-fidelity remote sensing image compression and resolution ratio enhancement joint treatment method |
Non-Patent Citations (1)
| Title |
|---|
| 《地理空间信息》 20101231 张静等 小波变换在遥感图像压缩中的应用 56-58 1-6 第8卷, 第06期 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103077543A (en) * | 2012-12-29 | 2013-05-01 | 浙江工业大学 | Embedded zerotree coding method on basis of inverse Loop subdivision |
| EP3035685A1 (en) * | 2014-12-16 | 2016-06-22 | Thomson Licensing | A device and a method for encoding an image and corresponding decoding method and decoding device |
| EP3035687A1 (en) * | 2014-12-16 | 2016-06-22 | Thomson Licensing | A device and a method for encoding an image and corresponding decoding method and decoding device |
| EP3035686A1 (en) * | 2014-12-16 | 2016-06-22 | Thomson Licensing | A device and a method for encoding an image and corresponding decoding method and decoding device |
| US9681140B2 (en) | 2014-12-16 | 2017-06-13 | Thomson Licensing | Device and a method for encoding an image and corresponding decoding method and decoding device |
| CN106018325A (en) * | 2016-04-29 | 2016-10-12 | 南京富岛信息工程有限公司 | Method for evaluating credibility of gasoline property modeling prediction result |
| CN106018325B (en) * | 2016-04-29 | 2019-02-01 | 南京富岛信息工程有限公司 | A method of evaluation gasoline property modeling and forecasting credible result degree |
| CN107819469A (en) * | 2016-09-13 | 2018-03-20 | 达索系统公司 | To representing that the signal of physical attribute is compressed |
| CN107819469B (en) * | 2016-09-13 | 2024-03-19 | 达索系统公司 | Compression of signals representing physical properties |
| US11074445B2 (en) | 2017-12-26 | 2021-07-27 | Beijing Sensetime Technology Development Co., Ltd. | Remote sensing image recognition method and apparatus, storage medium and electronic device |
| WO2019129032A1 (en) * | 2017-12-26 | 2019-07-04 | 北京市商汤科技开发有限公司 | Remote sensing image recognition method and apparatus, storage medium and electronic device |
| CN109857709A (en) * | 2019-02-22 | 2019-06-07 | 中国人民解放军空军研究院战略预警研究所 | A kind of data compression method and apparatus |
| CN112257621A (en) * | 2020-10-28 | 2021-01-22 | 贵州电网有限责任公司 | Equipment image identification method for unmanned aerial vehicle inspection |
| CN113326744A (en) * | 2021-05-12 | 2021-08-31 | 哈尔滨工业大学 | Method and system for detecting on-orbit state abnormity of spacecraft |
| CN113326744B (en) * | 2021-05-12 | 2023-03-03 | 哈尔滨工业大学 | Method and system for detecting on-orbit state abnormity of spacecraft |
| CN117395381A (en) * | 2023-12-12 | 2024-01-12 | 上海卫星互联网研究院有限公司 | Compression method, device and equipment for telemetry data |
| CN117395381B (en) * | 2023-12-12 | 2024-03-12 | 上海卫星互联网研究院有限公司 | Compression method, device and equipment for telemetry data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102186069B (en) | 2013-05-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102186069B (en) | Remote sensing image data compression method capable of maintaining measurement performance | |
| US11494948B2 (en) | Point cloud geometric compression method based on depth auto-encoder | |
| CN110087092B (en) | Low-bit-rate video coding and decoding method based on image reconstruction convolutional neural network | |
| US20190355152A1 (en) | Point cloud attribute compression method based on kd tree and optimized graph transformation | |
| CN102148987B (en) | Compressed sensing image reconstructing method based on prior model and 10 norms | |
| CN112465846B (en) | Cloud-containing remote sensing image compression method based on filling strategy | |
| CN102945552A (en) | No-reference image quality evaluation method based on sparse representation in natural scene statistics | |
| CN110248190A (en) | A kind of compressed sensing based multilayer residual error coefficient image encoding method | |
| CN103501437A (en) | Fractal and H.264-based hyper-spectral image compression method | |
| CN103546759A (en) | Image compression coding method based on combination of wavelet packets and vector quantization | |
| CN108419083A (en) | A kind of full subband compressed sensing encryption algorithm of image multilevel wavelet | |
| Bletterer et al. | Point cloud compression using depth maps | |
| CN101720043B (en) | Imaging spectrum image compression method based on multi-mode prediction | |
| CN105528623A (en) | Imaging spectrum image sparse representation method based on ground object class classification redundant dictionary | |
| CN102316324A (en) | Image coding prediction method based on local minimum entropy | |
| Duan et al. | Content-aware deep perceptual image compression | |
| CN103065336B (en) | Directional pyramid coding method of image | |
| CN117337449A (en) | Point cloud quality enhancement method, encoding and decoding methods and devices, and storage medium | |
| Yu et al. | Analysis of JPEG2000 Compression Quality of Optical Satellite Images | |
| CN114449248B (en) | Spherical image compression method based on spherical wavelet transformation | |
| Kumari et al. | Image Compression using Clustering Techniques for Bio Medical Applications | |
| Narasimhulu et al. | Hybrid LWT and DCT based High-Quality Image Compression | |
| Shi et al. | A lossless image compression algorithm by combining DPCM with integer wavelet transform | |
| Al-Sammaraie | Medical Images Compression Using Modified SPIHT Algorithm and Multiwavelets Transformation | |
| Hu et al. | A radar data compression method based on autoencoder neural network and range encoding |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130508 Termination date: 20140114 |