CN109151482A - Spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method - Google Patents
Spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method Download PDFInfo
- Publication number
- CN109151482A CN109151482A CN201811269511.8A CN201811269511A CN109151482A CN 109151482 A CN109151482 A CN 109151482A CN 201811269511 A CN201811269511 A CN 201811269511A CN 109151482 A CN109151482 A CN 109151482A
- Authority
- CN
- China
- Prior art keywords
- value
- lossless
- environment
- compression
- spaceborne
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/004—Predictors, e.g. intraframe, interframe coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/182—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a pixel
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
Abstract
Hybrid compression system and method are damaged the invention discloses a kind of spaceborne spectrum picture spectral coverage is lossless, solve the problems, such as that spaceborne spectrum picture compression can not guarantee that compression ratio and important spectral coverage information are not lost simultaneously, specific steps have: spectral image data and parameter input;Input data and control signal;Initialization of variable;Partial gradient calculates and quantization;Merge gradient vector;Predict current pixel and pixel correction;Residual computations quantify, remap and pixel reconstruction;Compression module context parameters are mixed to update;Columbus's coding parameter k is calculated;Damage lossless mixing Columbus coding;The output of compression of images result.The present invention improves on the basis of JPEG_LS algorithm, realizes that spectral coverage selectively compresses to spectrum picture, to specified spectral coverage lossless compression, other spectral coverage lossy compressions guarantee that important information is completely lossless, and reach certain compression ratio, occupancy resource is less, and compression speed is faster.Have been used for mars exploration spectroanalysis instrument.
Description
Technical field
The invention belongs to spectrum picture processing technology fields, further relate to spectrum picture and carry out lossless by spectral coverage and have
Damage mixing compression.Specifically a kind of spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method.For to spectrum picture difference
Spectral coverage carries out lossless and close lossless mixing compression.
Background technique
JPEG-LS is a kind of new lossless/close lossless compression standard for continuous tone still image, ISO-
14495-1/ITU-T.87.Core algorithm is LOCO-I (LOw Complexity Lossless Compression for
Images).LOCO-I algorithm, which can be obtained, similar with current many compression algorithms based on arithmetic coding even preferably presses
Contracting efficiency, while it being able to maintain lower complexity again.
LOCO-I algorithm combines simple (relative to the arithmetic coding) of Huffman coding and the huge pressure of context modeling
Contracting potentiality, therefore taken the length of many families.The algorithm has used a kind of nonlinear prediction device with edge self-checking function, and
The context model simply determined by quantization gradient very much based on one.This method is realized multiple with a few statistical parameter
Effect achieved by miscellaneous universal model technology, and there is no " context dilution " effect.In the method, certainly by one
The Golomb-Rice coding of adaptation makes the corresponding one-parameter probability distribution of each context, and uses in flat site embedding
Enter formula alphabet extension mechanism.Another complicated algorithm FELICS be also based on Golomb-Rice coding, LOCO-I and it
The difference is that LOCO-I more follows traditional prediction-modeling-coding mode, using a kind of novel and simple specific
Golomb-Rice method for parameter estimation.In the low complex degree maintained like, the compression efficiency of LOCO-I ratio FELICS is mentioned
It is high very much.Equally, LOCO-I is higher than the compression ratio of current many methods, and complexity is relatively low.In fact, ISO/
The IEC JTC1/SC29WG1 committee is considering whether to replace current low complex degree lossless compression standard with JPEG-LS.
Xian Electronics Science and Technology University its patent application " JPEG_LS routine coded hardware implementation method " (number of patent application:
201210198818.X, publication number: CN102724506A) in disclose JPEG_LS routine coded hardware implementation method.This side
Method uses level Four feedback loop, real-time update and feedback parameter updated value, corrects ginseng with the first two point tolerance of current pixel point
The method of numerical prediction error amount, improves compression speed.But remain unfortunately, this method may be only available for whole
Width compression of images can not accomplish to damage spectrum picture and compress with lossless mixing, it is difficult to realize in the case where guaranteeing compression ratio
Important spectral coverage information is kept not lose.
Summary of the invention
It is an object of the invention to overcome the deficiencies in the prior art, propose a kind of speed of service faster, to occupy resource less
Spaceborne spectrum picture spectral coverage lossless damage mixing compression method.
The present invention, which is that a kind of spaceborne spectrum picture spectral coverage is lossless first, damages hybrid compression system, successively according to data flow
It is connected to input module, control module, compression module and output module, which is characterized in that compression module is mixing compression module, is mixed
It closes in compression module comprising damaging environment submodule, lossless environment submodule and Columbus's encoding submodule, damages environment submodule
Block and lossless environment submodule are switched over by the way of table tennis, damage the calculating of environment submodule and lossless environment submodule
As a result it is input to Columbus's encoding submodule, mixing compression result is obtained after being encoded and is input in output module.
The present invention or a kind of spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method, described in claim 1-2
Any spaceborne spectrum picture spectral coverage lossless damage realized in hybrid compression system, which is characterized in that comprise the following steps that
Step 1, spectral image data and parameter input: it is slow that spaceborne spectral image data is input to the front end in input module
It deposits in FIFO and caches;The parameter of spaceborne spectrum picture is inputted to input module, parameter includes frame start signal, spectrogram image width
Degree, spectrum picture height, spectral image data and enable signal, limits of error near value;Near value determines present image
The environment of compression is to damage or lossless, indicates to indicate damaging when near value is not 0 in lossless environment when near value is 0
Environment;Near value value range is 0-15;Frame synchronizing signal, line synchronising signal are generated according to the parameter of input.
Step 2, input data and control signal: control module by spectral image data, control signal in enable signal,
Frame start signal, near value, frame synchronizing signal, line synchronising signal are input to mixing compression module.
Step 3, initialization of variable: to the variable in the initialization unit of mixing compression module according to the corresponding near of every row
Value assign initial value, variable include the first quantized interval threshold values T1, the second quantized interval threshold values T2, third quantized interval threshold values T3,
Bit number qbpp needed for predicting error range range and mapping error value;At the beginning of damaging environment submodule and lossless environment submodule
The value of beginningization element variable is different.
Step 4, partial gradient calculates and quantifies: first, in accordance in the calculating of cause and effect formwork calculation partial gradient and quantifying unit
Local gradient vectors, including first partial gradient value D1, the second local gradient value D2, third partial gradient value D3, and
According to each quantized interval threshold values T1, T2, T3 and near value of setting, 9 partial gradient quantized intervals, partial gradient quantization are obtained
Section value be -4 to 4,9 local quantized intervals be respectively (bear infinite,-T3], (- T3,-T2], (- T2,-T1], (- T1, -
Near), [- near, near], (near, T1], (T1, T2], (T2, T3], (T3, just infinite), by judging local gradient vectors
In partial gradient value Di, i value be 1 to 3, fall in which local quantized interval obtains corresponding quantized value Qi, after obtaining quantization
Local gradient vectors, respectively first partial gradient quantized value Q1, the second local gradient amount value Q2, third partial gradient amount
Change value Q3.
Step 5, merge gradient vector: in three partial gradient quantized values, the opposite partial gradient amount of the same symbol will be worth
Change value merges, and sign bit sign is arranged to the gradient vector after merging, and according to Q1, Q2, Q3 computation index value Q.
Step 6, current pixel and pixel correction are predicted: spaceborne spectrum picture current pixel predicted value is calculated according to fallout predictor
Px;Current pixel predicted value Px is modified using corrected parameter C [Q], and current pixel predicted value Px is limited in [0,
MAXVAL] in, MAXVAL is possible maximum image pixel value, obtains amendment predicted value Px_c.
Step 7, residual computations, quantify, remap and pixel reconstruction: current pixel value is obtained by spaceborne spectral image data
Ix obtains current residue value Errval with current pixel value Ix, amendment predicted value Px_c and sign bit sign;To residual values
Errval carries out quantization and is non-negative according to currently damaging environment or lossless environment and remapping current residue value Errval
Residual values MErrval;In pixel reconstruction, if use is reconstructed to pixel according to current near value in damaging under environment
In subsequent prediction;If under lossless environment, using current pixel value Ix as reconstructed pixel value Rx.
Step 8, mixing compression module context parameters update: damage under lossless environment, respectively to current residue value
Accumulation calculating is carried out, the accumulated value of residual absolute value, the cumulative frequency of residual values, residual error accumulated value and residual prediction amendment are obtained
Value respectively writes into these context parameters variables the hereafter parameter update list for damaging environment submodule and lossless environment submodule
In the RAM of member, completes context parameters and update.
Step 9, Columbus's coding parameter k is calculated: in the case where damaging environment and lossless environment respectively according to the accumulation of residual values
The accumulated value of number and residual absolute value calculates Golomb coding parameter k;Parameter k is output to taxi driver brother's human relations in mixing compression module
In cloth encoding submodule.
Step 10, damage lossless mixing Columbus coding: Columbus's coding module is by MErrval divided by 2k, obtain divisor
And remainder, output export remainder into output module for 11 except several 0 and then output again, complete to compile Columbus of MErrval
Code, using Columbus's coding module to the output result for damaging environment submodule and lossless environment submodule above
MErrval carries out mixing Columbus's coding, completes spaceborne spectrum picture spectral coverage and damages lossless mixing compression.
Step 11, compression of images result exports: the rear end that the data for mixing Columbus's coding write into output module is cached
FIFO, and by 16 bit wide outputs, output is the mixing compression result of spaceborne spectrum picture.
Compared with the prior art, the present invention has the following advantages:
First, the present invention may be implemented lossless damage of spectrum picture spectral coverage and mix compression, i.e., realizes waveband selection to spectrum
Property compression, to designated band carry out lossless compression, all band carry out lossy compression, thus both guaranteed important information entirely without
Damage, and reach certain compression ratio, reduce the data volume for needing to transmit.
Second, the spectrum picture spectral coverage of realization of the present invention in FPGA is lossless to damage Mixing compression algorithm and the prior art
It is less compared to the resource of occupancy, it is used in slice, trigger, look-up table and uses number few than the prior art in number, so that
The scope of application of the invention is wider.
Third, the spectrum picture spectral coverage of realization of the present invention in FPGA is lossless to damage Mixing compression algorithm and the prior art
It is higher compared to maximum clock frequency, and can 7 clocks handle a pixels, so that compression speed of the invention compares the prior art
Faster.
Detailed description of the invention
Fig. 1 is the lossless flow chart for damaging mixing compression method of the spaceborne spectrum picture spectral coverage of the present invention;
Fig. 2 is cause and effect Prototype drawing of the invention;
Fig. 3 is system construction drawing of the invention;
Fig. 4 is mixing compression module structure chart of the invention.
Specific embodiment
Realization of the invention is described in detail with example with reference to the accompanying drawing.
Embodiment 1
Current method for compressing image can only compress whole picture spectrum picture, can not accomplish to damage spectrum picture and
Therefore lossless mixing compression, the present invention expand research regarding to the issue above, propose that spaceborne spectrum picture spectral coverage is lossless to be damaged
Hybrid compression system and method had not only guaranteed that important spectral coverage information was not lost, but also have reached certain compression ratio, realized mixing compression.
The present invention, which is that a kind of spaceborne spectrum picture spectral coverage is lossless first, damages hybrid compression system, successively according to data flow
It is connected to input module, control module, compression module and output module, referring to Fig. 3, compression module is mixing compression module, mixing
Comprising damaging environment submodule, lossless environment submodule and Columbus's encoding submodule in compression module, environment submodule is damaged
It is switched over by the way of table tennis with lossless environment submodule, damages the calculating knot of environment submodule and lossless environment submodule
Fruit is input to Columbus's encoding submodule, and mixing compression result is obtained after being encoded and is input in output module.
The transmission pressure for passing ground back in order to reduce spaceborne spectrum picture needs to compress spectrum picture, and current
Method can only accomplish lossy compression or lossless compression to spectrum picture, if only doing lossless compression to image, although can guarantee
The information of image is not lost, and certain compression ratio is unable to reach, if only doing lossy compression to image, though it can guarantee compression ratio,
But it not can guarantee important spectral coverage information not lose, therefore the invention proposes lossless damage of spaceborne spectrum picture spectral coverage to mix compression
System does lossless compression other spectral coverages to important spectral coverage and does lossy compression, both ensure that important spectral coverage information is not lost or can be reached
To the requirement of compression ratio.
The spaceborne lossless input module for damaging hybrid compression system of spectrum picture spectral coverage inputs spaceborne spectral image data
It caches in FIFO and caches to front end, while including frame start signal, spectrum picture according to the spaceborne spectrum picture parameter of input
Width, spectrum picture height, spectral image data and enable signal, limits of error near value generate frame synchronizing signal, row
Synchronization signal;Control module judges that current spectral coverage if important spectral coverage, then will make in current light spectrum image data, control signal
Energy signal, frame start signal, near value, frame synchronizing signal, line synchronising signal are input to the lossless environment of mixing compression module
Submodule plays the enable signal in current light spectrum image data, control signal, frame if current spectral coverage is insignificant spectral coverage
Beginning signal, near value, frame synchronizing signal, line synchronising signal are input to mixing compression module and damage environment submodule;Mixing
The output that compression module would detract from environment submodule and lossless environment submodule is input to Columbus's encoding submodule and encodes
Obtain afterwards mixing compression as a result, being output to output module.
The lossless hybrid compression system that damages of spaceborne spectrum picture spectral coverage in this example is in xc4vsx55-10ff1148 model
FPGA in realize, had been used in Mars mineral spectra analyzer.
Embodiment 2
The lossless overall composition for damaging hybrid compression system of spaceborne spectrum picture spectral coverage is the same as embodiment 1, referring to fig. 4, this hair
The environment submodule that damages of bright middle mixing compression module is connected with initialization unit, partial gradient calculating in turn according to data flow
With quantifying unit, merge gradient vector unit, prediction current pixel and pixel correction unit, residual computations and quantifying unit, on
Hereafter parameter updating unit and calculating Columbus's parameter K unit.Lossless environment submodule is identical as environment sub-modular structure is damaged,
Wherein initialization unit, partial gradient calculate with quantifying unit, residual computations and quantifying unit, context parameters updating unit and
The calculating parameter calculated in Columbus's parameter K unit is different, and merging gradient vector unit therein, prediction current pixel
Time-multiplexed mode is used with pixel correction unit.
Mixing compression module of the invention would detract from environment and lossless environment and separate, and need to carry out lossless pressure in current spectral coverage
Enter lossless environment when contracting, enter when needing to carry out lossy compression and damage environment, realizes the function of mixing compression.
The partial gradient meter of environment submodule is calculated with quantifying unit and damaged in the partial gradient of lossless environment submodule
Calculate and merge gradient vector unit with being connected between quantifying unit, lossless environment submodule residual computations and quantifying unit and
It damages and is connected with prediction current pixel and pixel correction unit between the residual computations and quantifying unit of environment submodule, merge ladder
The data of degree vector location are directly output to prediction current pixel and pixel correction unit, merge the reception of gradient vector unit and come from
The partial gradient of lossless environment submodule calculates with quantifying unit and damages the partial gradient calculating of environment submodule and quantifies single
The data of the data of member, prediction current pixel and pixel correction unit are sent to the residual computations and quantization of lossless environment submodule
Unit and the residual computations and quantifying unit for damaging environment submodule.
Merge gradient vector unit, prediction current pixel and pixel correction unit in the way of time-multiplexed by damaging ring
Border submodule and lossless environment submodule are used in conjunction with, that is, the partial gradient for damaging environment submodule calculates and quantifying unit and nothing
The partial gradient calculating of damage environment submodule all enters after coming out with quantifying unit spectral image data and merges gradient vector unit,
It, will prediction current pixel and pixel correction if being currently lossy compression subsequently into prediction current pixel and pixel correction unit
The result of unit output is input to the residual computations and quantifying unit for damaging environment submodule, will be pre- if being currently lossless compression
The result of survey current pixel and pixel correction unit is input to the residual computations and quantifying unit of lossless environment submodule.Damage ring
The output of border submodule and lossless environment submodule is also all input in Columbus's encoding submodule, also utilizes time-multiplexed side
Formula is used in conjunction with Columbus's encoding submodule.This time-multiplexed mode makes spaceborne spectrum picture spectral coverage is lossless to damage mixing
The resource occupation amount of compressibility is less.
Embodiment 3
The present invention or a kind of spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method, is related to input and output mould
Block, control module and compression module usually generate control signal in control module and are input to compression module, any of the above-described
Spaceborne spectrum picture spectral coverage lossless damage realize that spaceborne spectrum picture spectral coverage is lossless to damage mixing compression in hybrid compression system
System is with embodiment 1-2, and referring to Fig. 1, compression module is mixing compression module in the present invention, and spaceborne spectrum picture spectral coverage is lossless to be had
Damage mixing compression method comprises the following steps that
Step 1, spectral image data and parameter input: spaceborne spectral image data be input to spaceborne spectrum picture spectral coverage without
Damage damages to be cached in the front end caching FIFO in the input module of hybrid compression system;User inputs the parameter of spaceborne spectrum picture
To the lossless input module for damaging hybrid compression system of spaceborne spectrum picture spectral coverage, parameter includes frame start signal, spectrogram
Image width degree, spectrum picture height, spectral image data and enable signal, limits of error near value;Near value determines current
The environment of compression of images is to damage or lossless, indicates to indicate to exist when near value is not 0 in lossless environment when near value is 0
Damage environment;Near value value range is 0-15;It is generated according to the frame start signal of input, picture traverse, picture altitude information
Frame synchronizing signal, line synchronising signal.
Data 1a) are read from FIFO, one data of every reading, data counts data_cnt adds 1, until data counts are equal to
0 is set when picture traverse;
When 1b) and data_cnt effective in frame synchronizing signal is 0, line synchronising signal is set 1, is equal to image in data_cnt
Line synchronising signal is set 0 after width, while providing EOL mark;
1c) when frame start signal frame_start signal is effective, frame synchronizing signal is set 1, and starts to count row
It counts, effectively then to row, count is incremented for each EOL mark, and until row counting is equal to, picture altitude subtracts 1 while EOL mark has
When effect, frame end mark is set to 1, while frame synchronizing signal is set to 0.
Step 2, input data and control signal: the lossless control mould for damaging hybrid compression system of spaceborne spectrum picture spectral coverage
Block by spectral image data, control signal in enable signal, frame start signal, near value, frame synchronizing signal, line synchronising signal
It is input to mixing compression module.
Step 3, initialization of variable: to the variable in the initialization unit of mixing compression module according to the corresponding near of every row
Value assign initial value, variable include the first quantized interval threshold values T1, the second quantized interval threshold values T2, third quantized interval threshold values T3,
Bit number qbpp needed for predicting error range range and mapping error value;At the beginning of damaging environment submodule and lossless environment submodule
The value of beginningization element variable is different.
Step 4, partial gradient calculates and quantifies: first, in accordance in the calculating of cause and effect formwork calculation partial gradient and quantifying unit
Local gradient vectors, including first partial gradient value D1, the second local gradient value D2, third partial gradient value D3, and
According to each quantized interval threshold values T1, T2, T3 and near value being arranged in step 3,9 partial gradient quantized intervals are obtained, part
Gradient quantized interval value be -4 to 4,9 local quantized intervals be respectively (bear infinite,-T3], (- T3,-T2], (- T2, -
T1], (- T1 ,-near), [- near, near], (near, T1], (T1, T2], (T2, T3], (T3, just infinite), pass through judgement office
Partial gradient value Di, i value in portion's gradient vector is 1 to 3, falls in which local quantized interval obtains corresponding quantized value Qi,
To the local gradient vectors after quantify, respectively first partial gradient quantized value Q1, the second part gradient amount value Q2,
Third partial gradient quantized value Q3.
Step 5, merge gradient vector: in three partial gradient quantized values, the opposite partial gradient amount of the same symbol will be worth
Change value merges, and sign bit sign is arranged to the gradient vector after merging, and according to Q1, Q2, Q3 computation index value Q.
If 5a) first nonzero element of partial gradient quantized value Q1, Q2, Q3 are negative values, by the value of all elements
It negates;
1 5b) is set by sign bit if being inverted, is otherwise provided as 0;
Index value Q 5c) is calculated by partial gradient quantized value Q1, Q2, Q3.
Step 6, current pixel and pixel correction are predicted: spaceborne spectrum picture current pixel predicted value is calculated according to fallout predictor
Px;Current pixel predicted value Px is modified using corrected parameter C [Q], and current pixel predicted value Px is limited in [0,
MAXVAL] in, MAXVAL is possible maximum image pixel value, obtains amendment predicted value Px_c.
Step 7, residual computations, quantify, remap and pixel reconstruction: current pixel value is obtained by spaceborne spectral image data
Ix obtains current residue value Errval with current pixel value Ix, amendment predicted value Px_c, determines residual values according to sign bit sign
Whether Errval negates, and sign bit is 1 and negates that sign bit does not negate for 0;Residual values Errval is quantified and according to working as
Before damage environment or lossless environment and remap current residue value Errval for non-negative residual values MErrval.In pixel weight
When structure, if being reconstructed pixel for subsequent prediction according to current near value in damaging under environment;If being in lossless ring
Under border, then using current pixel value Ix as reconstructed pixel value Rx.
Step 8, mixing compression module context parameters update: referring to fig. 4, damage under lossless environment, respectively to working as
Preceding residual values carry out accumulation calculating, obtain the accumulated value of residual absolute value, the cumulative frequency of residual values, residual error accumulated value and residual error
Forecast value revision value is write into these context parameters variables in two groups of RAM, completes context parameters and updates;Residual absolute value
Accumulated value first takes absolute value for residual values to be accumulated again, and residual error accumulated value is that residual values are directly accumulated, and forecast value revision value is by residual
Poor accumulated value and the cumulative frequency of residual values are calculated.
Step 9, Columbus's coding parameter k is calculated: in the case where damaging environment and lossless environment respectively according to the accumulation of residual values
The accumulated value of number and residual absolute value calculates Golomb coding parameter k;Parameter k is output to the compression of mixing described in step 10
In Columbus's encoding submodule in module.
Step 10, damage lossless mixing Columbus coding: Columbus's coding module is by MErrval divided by 2k, obtain divisor
And remainder, output export remainder into output module for 11 except several 0 and then output again, complete to compile Columbus of MErrval
Code, in order to save resource, mixing compression module finally only uses Columbus's coding module and damages environment submodule to above
Mixing Columbus's coding is carried out with the output result MErrval of lossless environment submodule, spaceborne spectrum picture spectral coverage is completed and damages
Lossless mixing compression.
Step 11, compression of images result exports: the rear end that the data for mixing Columbus's coding write into output module is cached
FIFO, and by 16 bit wide outputs, output is the mixing compression result of image.
The present invention may be implemented lossless damage of spectrum picture spectral coverage and mix compression, i.e., realizes band selective pressure to spectrum
Contracting carries out lossless compression to designated band, and all band carries out lossy compression, thus not only guaranteed that important information was completely lossless, but also
Reach certain compression ratio, reduces the data volume for needing to transmit.
Embodiment 4
Lossless hybrid compression system and the method for damaging of spaceborne spectrum picture spectral coverage is with embodiment 1-3, described in step (3)
The variable for needing to initialize will initialize primary when first data of the every row of spectrum picture arrive according to different near values;
T1 is assigned a value of 18, T2 and is assigned a value of 67, T3 being assigned a value of 276, range and be assigned a value of 4096, qbpp being assigned a value of 12 under lossless environment;?
It damages under environment, with formula T1=18+3*near, T2=67+5*near, T3=276+7*near, range=(4095+2*
Near)/(2*near+1)+1, qbpp=log2(range) value being calculated is in the initialization unit of mixing compression module
Variable carry out initialization assignment.
The present invention is when the spectral coverage of spaceborne spectrum picture needs lossless compression with the initialization unit of lossless environment submodule
Initialization assignment is carried out, when the spectral coverage of spaceborne spectrum picture needs lossy compression with the initialization unit for damaging environment submodule
Initialization assignment is carried out, the initialization that would detract from environment and lossless environment separates, this step is to realize the basis of mixing compression.
Embodiment 5
Lossless hybrid compression system and the method for damaging of spaceborne spectrum picture spectral coverage is with embodiment 1-4, described in step (7)
If can be reconstructed according to current near value to pixel in damaging under environment in pixel reconstruction, obtain reconstructed pixel Rx;
If under lossless environment, using current pixel value Ix as reconstructed pixel value Rx.
In the case where damaging environment, pixel reconstruction is carried out according to quantization residual error E, near value and sign bit sign, reconstruction formula is such as
Under:
If (sign=1) Rx=Px-E* (2*near+1)
Else Rx=Px+E* (2*near+1)
Under lossless environment, reconstruction formula is as follows:
Rx=Ix
It is single with the residual computations of lossless environment submodule and quantization when the spectral coverage of spaceborne spectrum picture needs lossless compression
Member carries out pixel reconstruction to residual error, when the spectral coverage of spaceborne spectrum picture needs lossy compression with the residual error for damaging environment submodule
It calculates and quantifying unit carries out pixel reconstruction to residual error, the pixel reconstruction that would detract from environment and lossless environment separates, and guarantees mixing
Compression damages environment submodule and lossless environment submodule and can operate normally.
Embodiment 6
Lossless hybrid compression system and the method for damaging of spaceborne spectrum picture spectral coverage with embodiment 1-5, described in step (8) on
Hereafter parametric variable is write into two groups of RAM, and completing context parameters update is that parameter is stored into two groups of ram, lossless
A1 [Q], N1 [Q], B1 [Q], C1 [Q] value enter one group of ram, and the A damaged [Q], N [Q], B [Q], C [Q] value enter another group
RAM, specifically
8.1 in the case where damaging environment, carries out accumulation calculating to current residue value, obtains the accumulated value A for damaging residual absolute value
[Q] damages the cumulative frequency N [Q] of residual values, damages residual error accumulated value B [Q] and damages residual prediction correction value C [Q], will be pre-
It surveys correction value C [Q] and constrains in realization update in [MIN_C, MAX_C] range.
8.2 updated parameters are write into the RAM for damaging environment.
8.3 under lossless environment, carries out accumulation calculating to current residue value, obtains the accumulated value A1 of lossless residual absolute value
[Q], the cumulative frequency N1 [Q] of lossless residual values, lossless residual error accumulated value B1 [Q], lossless residual prediction correction value C1 [Q].
8.4 these variables are write into the RAM of lossless environment, complete context parameters and update.
Joined respectively to store the context of the context parameters and lossless environment that damage environment submodule using two groups of RAM
Number, prevents two kinds of parameters from interfering with each other, and allows and damages environment submodule and lossless environment submodule independently works normally, and protects
Card mixing compression can operate normally.
Embodiment 7
Lossless hybrid compression system and the method for damaging of spaceborne spectrum picture spectral coverage is with embodiment 1-6, described in step (9)
Golomb coding parameter k is calculated, basis damages the cumulative frequency N [Q] for damaging residual values in environment and damages in the case where damaging environment
The accumulated value A [Q] of prediction residual absolute value calculates Golomb coding parameter k;According to lossless in lossless environment under lossless environment
The cumulative frequency N1 [Q] of residual values and accumulated value A1 [Q] the calculating parameter k of non-destructive prediction residual absolute value.
In the case where damaging environment, k, which is taken, meets N [Q] * 2kThe maximum value of < A [Q], k, which takes, under lossless environment meets N1 [Q] * 2k<A1
The maximum value of [Q].
The present invention calculates k using the context parameters for damaging environment submodule in the case where damaging environment, in lossless environment
Lower context parameters using lossless environment submodule calculate k, total to damage environment submodule and lossless environment submodule
Guarantee is provided with using Columbus's encoding submodule.
Embodiment 8
Lossless hybrid compression system and the method for damaging of spaceborne spectrum picture spectral coverage is the same as embodiment 1-7, step (5), step (6)
Described in merging gradient vector unit, Columbus compiles described in prediction current pixel and pixel correction unit and step (10)
Numeral module is all made of time-multiplexed mode.
Due to lossless environment and damages the merging gradient vector unit of environment, predicts that current pixel and pixel correction unit are complete
Exactly the same, lossless environment uses time-multiplexed mode to share the two units, and lossless environment submodule with environment is damaged
Enter Columbus's encoding submodule with the result for damaging environment submodule.
The present invention is shared in the way of time-multiplexed merges gradient vector unit, prediction current pixel and pixel correction
Unit and Columbus's encoding submodule, take full advantage of resource, reduce resource occupation amount.
Embodiment 9
For lossless hybrid compression system and the method for damaging of spaceborne spectrum picture spectral coverage with embodiment 1-8, the present invention is guaranteeing one
Under conditions of determining compression ratio, lossless compression is carried out to some important spectral coverages, remaining spectral coverage carries out lossy compression.
To achieve the above object, the method for the present invention includes following steps:
Step 1, parameter is inputted, parameter includes frame start signal, picture traverse, picture altitude, image data and enabled letter
Number, limits of error near value.
Step 2, the front end caching FIFO of input module is write data into.
Step 3, according to the frame start signal of input, picture traverse, picture altitude information generates line synchronising signal and frame is same
Walk signal.
Image data 3a) is read from FIFO, one data of every reading, data counts data_cnt adds 1, until data counts
0 is set when equal to picture traverse.
When 3b) and data_cnt effective in frame synchronizing signal is 0, line synchronising signal is set 1, is equal to image in data_cnt
Line synchronising signal is set 0 after width, while providing EOL mark.
3c) when frame start signal frame_start signal is effective, frame synchronizing signal is set 1, and starts to count row
It counts, effectively then to row, count is incremented for each EOL mark, and until row counting is equal to, picture altitude subtracts 1 while EOL mark has
When effect, frame end mark is set to 1, while frame synchronizing signal is set to 0.
Step 4, data, data are enabled, the input mixing compression of frame synchronization, line synchronising signal, frame start signal, near value
Module.
Step 5, referring to fig. 4, environment submodule initialization unit and lossless environment submodule initialization unit pair are being damaged
All variables are initialized according to the corresponding near value of every row, and variable includes quantized interval threshold values T1, T2, T3, and prediction
Bit number qbpp needed for error range range and mapping error value.
Step 6, partial gradient is calculated to current pixel
Step 7, partial gradient is quantified
According to the calculated partial gradient size of quantized interval threshold values T1, T2, T3 and step 6 of setting, by partial gradient
Value quantization obtains partial gradient quantized value Q1, Q2, Q3 to 9 sections (value is -4 to 4).
Step 8, merge gradient vector, and be arranged and meet a sign
If 8a) first nonzero element for quantifying vector (Q1, Q2, Q3) is negative value, the value of all elements is negated,
Merge conducive to context parameters
Position 1 8b) is arranged in sign bit if being inverted, is otherwise provided as 0.
Index value Q 8c) is calculated by partial gradient quantized value Q1, Q2, Q3.
Step 9, predicted value Px is calculated according to fallout predictor.
Step 10, Px is modified using corrected parameter C [Q], and Px is limited in [0, MAXVAL], MAXVAL is
Possible maximum image pixel value.
Step 11, true according to sign bit by current pixel value and revised predictor calculation prediction error value Errval
It is fixed whether to negate.
Step 12, it damages under mode and residual error is quantified, and pixel is reconstructed according to the near value of every row and is used for
Subsequent prediction;Current pixel value Ix is equal to for lossless mode reconstructed pixel value Rx.
Step 13, it is calculated according to the accumulated value A [Q] of the frequency of occurrence N [Q] of prediction error and prediction Error Absolute Value
Golomb coding parameter k.
Step 14, residual error is remapped, prediction error value Errval is mapped as nonnegative number MErrval.
Step 15, the mapping residual error MErrval of step 14 is encoded using Golomb-Rice coding
Step 16, to variables A, B, C, N are updated, referring to fig. 4,
Damage under mode, the context parameters updating unit for damaging environment submodule according to prediction error value Errval more
The accumulated value A [Q] of new prediction Error Absolute Value predicts the frequency of occurrence N [Q] of error, the accumulated value B of modified prediction error
[Q].Under lossless mode, in the context parameters updating unit of lossless environment submodule, the variable for needing to update is A1 [Q], N1
[Q],B1[Q].It updates forecast value revision value C [Q], correction value is constrained in [MIN_C, MAX_C] range, under lossless mode, needed
The variable of update is C1 [Q].
Step 17, the data of coding are write into the caching FIFO of output module, and by 16 bit wide outputs.
A more detailed example is given below, to the present invention carry out deeper into explanation
Embodiment 10
Spaceborne spectrum picture spectral coverage is lossless to damage hybrid compression system and method with embodiment 1-9,
The realization step of 1 pair of the method for the present invention is described in detail with reference to the accompanying drawing.
Step 1, referring to Fig. 3, parameter is inputted to input module, parameter includes frame start signal, picture traverse, image height
Degree, image data and enable signal, limits of error near value.
Step 2, the front end of spaceborne spectral image data write-in input module is cached into FIFO.
Step 3, referring to Fig. 3, frame start signal of the control module according to input, picture traverse, the generation of picture altitude information
Line synchronising signal and frame synchronizing signal, and by data, data are enabled, frame synchronization, line synchronising signal, frame start signal, near value
Input mixing compression module.
3a) from the front end of input module caching FIFO read spectral image data, one data of every reading, control module
Data counts data_cnt adds 1, sets 0 when data counts are equal to picture traverse.
When 3b) and data_cnt effective in the frame synchronizing signal of control module is 0, the line synchronising signal of control module is set
1, wait 4 clocks that the line synchronising signal of control module is set 0 again after data_cnt is equal to picture traverse, while terminating to trip
Mark.
3c) when frame start signal frame_start signal is effective, the frame synchronizing signal of control module is set 1, and start
It is counted to capable, effectively then to row, count is incremented for each EOL mark, goes simultaneously until row counting subtracts 1 equal to picture altitude
When end mark is effective, frame end mark is set to 1, while the frame synchronizing signal of control module is set to 0.
3d) for control module by data, data are enabled, and frame synchronization, line synchronising signal, frame start signal, the input of near value are mixed
Close compression module
Step 5, referring to fig. 4, to the variable in the initialization unit of mixing compression module according to the corresponding near value of every row
It is initialized, variable includes quantized interval threshold values T1, T2, T3, and prediction error range range and mapping error value institute
Bit number qbpp is needed, corresponding relationship is shown in Table 1.
1 initializing variable table of table
Near | T1 | T2 | T3 | Range | Qbpp |
0 | 18 | 67 | 276 | 4096 | 12 |
1 | 21 | 72 | 283 | 1366 | 11 |
2 | 24 | 77 | 290 | 820 | 10 |
3 | 27 | 82 | 297 | 586 | 10 |
4 | 30 | 87 | 304 | 456 | 9 |
5 | 33 | 92 | 311 | 374 | 9 |
6 | 36 | 97 | 318 | 316 | 9 |
7 | 39 | 102 | 325 | 274 | 9 |
Step 6, partial gradient is calculated to current pixel
According to cause and effect template, referring to fig. 2, wherein x is current pixel, and Ra, Rb, Rc, Rd, Re are adjacent pixel.
Wherein partial gradient D1=Rd-Rb
D2=Rb-Rc
D3=Rc-Ra
Step 7, partial gradient is quantified
Referring to fig. 4, it calculates in partial gradient with quantifying unit, partial gradient is quantized to 9 sections, according to step 5
The calculated partial gradient size of quantized interval threshold values T1, T2, T3 and step 6 of setting quantified after gradient value Q1, Q2,
The corresponding relationship of Q3, quantized interval and gradient quantized value is shown in Table 2.
The mapping table of table 2 quantized interval and gradient quantized value
Quantized interval | Gradient quantized value Q |
(-∞,-T3] | -4 |
(-T3,-T2] | -3 |
(-T2,-T1] | -2 |
(-T1,-Near) | -1 |
[-Near,Near] | 0 |
(Near,T1] | 1 |
(T1,T2] | 2 |
(T2,T3] | 3 |
(T3,+∞] | 4 |
Step 8, merge gradient vector, and be arranged and meet a sign
If 8a) first nonzero element for quantifying vector (Q1, Q2, Q3) is negative value, the value of all elements is negated.
Position 1 8b) is arranged in sign bit if being inverted, is otherwise provided as 0.
Index value Q 8c) is calculated by partial gradient quantized value Q1, Q2, Q3.
If Q1, Q2 0, Q=360+Q3;
If Q1 is that 0, Q2 is not 0, then Q=319+Q2+Q3;
If Q1, Q2 are not 0, Q=Q1+Q2+Q3-41.
Step 9, predicted value Px is calculated according to fallout predictor.
Referring to fig. 4, in prediction current pixel and pixel correction unit, Px is calculated according to the value of Ra, Rb, Rc, wherein
Px is the predicted value of x, and calculation formula is as follows
Step 10, Px is modified using corrected parameter C [Q], and Px is limited in [0, MAXVAL].
It since deviation exists, needs to be modified Px, modification method is, no if sign bit is 1, Px=Px+C [Q]
Then Px=Px-C [Q], limits Px after amendment, is limited in [0, MAXVAL], and MAXVAL is maximum image pixel
Value.
Step 11, referring to fig. 4, in residual computations and quantifying unit, by current pixel value and revised predictor calculation
Prediction error value Errval, Errval=Ix-Px determine whether to negate according to sign bit sign, negate if sign is 1, no
It does not negate then.
Step 12, it damages under mode and residual error is quantified, and pixel is reconstructed according to the near value of every row and is used for
Subsequent prediction;Current pixel value Ix is equal to for lossless mode reconstructed pixel value Rx.
Mode is being damaged, residual error is being quantified, quantified precision depends on the near value of every row, and quantitative formula is as follows
Errval=(Errval+near)/(2*near+1)
Pixel reconstruction is carried out according to quantization residual error, reconstruction formula is as follows
Rx=Px+Errval* (2*near+1)
Step 13, referring to fig. 4, in calculating Columbus's coding parameter k cell, the parameter k of golomb coding is calculated, k takes
Meet N [Q] * 2kThe maximum value of < A [Q]
Calculation formula be K [Q]=min k | 2k* N [Q] >=A [Q] }
Step 14, residual error is remapped, is changed into nonnegative number Errval.
When meeting the condition of near=0, k=0 and 2*B [Q]≤- N [Q], if Errval >=0, MErrval value are 2*
Errval+1, if Errval < 0, MErrval value are -2* (Errval+1);
When being unsatisfactory for above-mentioned condition, if Errval>=0, MErrval value are 2*Errval, if Errval<0, MErrval
Value is -2*Errval-1.
Step 15, referring to fig. 4, in Columbus's encoding submodule, the mapping to step 14 is encoded using Golomb-Rice
Residual error MErrval is encoded, and lossless compression and lossy compression all enter this module and encoded.
If being less than zero_num=LIMIT- according to the value M1_part that the MErrval higher bit position that k value determines is constituted
Qbpp-1 (LIMIT be coding limit value), encoding code stream are M1_part 0, add one 1, then plus MErrval k low bit
Position.Otherwise, encoding code stream is zero_num 0, adds one 1, then plus MErrval-1 qbpp low bit position.
Step 16, referring to fig. 4, in context parameters updating unit, to variables A, B, C, N is updated.
It may also be lossy compression since every row may be lossless compression, therefore two groups of RAM are set, A, B, C, N are deposited
Storage, lossless A1 [Q], N1 [Q], B1 [Q], C1 [Q] value enter one group of RAM, and the A damaged [Q], N [Q], B [Q], C [Q] value enter
Another group of RAM.
16a) update A, B, N.The value of A [Q] adds the absolute value of Errval on the basis of the original, and the value of B [Q] is original
On the basis of add Errval* (2*near+1), the value of N [Q] is equal to initial value and adds 1.If the value of N [Q], which reaches 64, A [Q] value, removes 2,
N [Q] value removes 2;2 are removed when B [Q] is more than or equal to 0, B [Q] value, otherwise B [Q] value is updated to-((1-B [Q])/2).
Forecast value revision value C [Q] 16b) is updated, when B [Q] is less than or equal to-N [Q], B [Q] is that B [Q] adds N [Q], if B at this time
[Q] is less than or equal to-N [Q], and B [Q] value is updated to-N [Q]+1;If C [Q], which is greater than -128, C [Q], subtracts 1 certainly.When B [Q] is greater than 0
When, B [Q] subtracts N [Q] in original value, if B [Q] is still set up greater than 0 at this time, B [Q] value is updated to 0;If C [Q] is less than
127, then C [Q] adds 1 certainly.
Step 17, the data of coding are write into caching FIFO, and by 16 bit wide outputs.
Effect of the invention can be further illustrated compared with the prior art by emulation.Emulation of the invention be
It is realized in 10.1 Integrated Development software environment of Xilinx ISE using Verilog HDL language.Synthesis result and simulation result are such as
Shown in table 3, table 4.
The comprehensive comparison of table 3 present invention and the prior art
From table 3 it can be seen that the present invention is compared with the prior art " JPEG_LS routine coded hardware implementation method ",
Slice uses number, and trigger uses number, and look-up table uses in number all than the prior art " JPEG_LS routine coded hardware
Implementation method " will be lacked, and the FPGA resource for illustrating that the present invention occupies is less.
Table 4 gives the comprehensive comparison of the present invention with the prior art, mainly from the target devices of realization and composite clock frequency
Two aspects of rate compare.
The comprehensive comparison of table 4 present invention and the prior art
From table 4, it can be seen that the present invention is compared with the prior art " JPEG_LS routine coded hardware implementation method ", most
High clock frequency improves a lot, and saves the runing time of entire algorithm, compression speed is faster.
In brief, spaceborne spectrum picture spectral coverage disclosed by the invention is lossless damages hybrid compression system and method, solves
Spaceborne spectrum picture compression can not guarantee the problem of compression ratio and important spectral coverage information are not lost simultaneously.Specific steps include:
(1) spectral image data and parameter input;(2) input data and control signal;(3) initialization of variable;(4) partial gradient calculates
With quantization;(5) merge gradient vector;(6) current pixel and pixel correction are predicted;(7) residual computations, quantify, remap and picture
Element reconstruct;(8) mixing compression module context parameters update;(9) Columbus's coding parameter k is calculated;(10) lossless mixing is damaged
Columbus's coding;(11) compression of images result exports.The present invention improves on the basis of JPEG_LS algorithm, can be to light
Spectrogram picture realizes that spectral coverage selectively compresses, and carries out lossless compression to specified spectral coverage, other spectral coverages carry out lossy compression, to both protect
It is completely lossless to demonstrate,prove important information, and reaches certain compression ratio, reduces the data volume for needing to transmit.The present invention realizes in FPGA,
Occupancy resource is less compared with prior art, and compression speed is faster.Have been used for mars exploration spectroanalysis instrument.
Claims (8)
1. a kind of spaceborne spectrum picture spectral coverage is lossless to damage hybrid compression system, be successively connected to according to data flow input module,
Control module, compression module and output module, which is characterized in that compression module is mixing compression module, is mixed in compression module
Comprising damaging environment submodule, lossless environment submodule and Columbus's encoding submodule, environment submodule and lossless environment are damaged
Submodule is switched over by the way of table tennis, and the calculated result for damaging environment submodule and lossless environment submodule is input to brother
Human relations cloth encoding submodule obtains mixing compression result and is input in output module after being encoded.
2. spaceborne spectrum picture spectral coverage according to claim 1 is lossless to damage hybrid compression system, which is characterized in that described
Damage environment submodule according to data flow be connected in turn initialization unit, partial gradient calculate with quantifying unit, merge ladder
Spend vector location, prediction current pixel and pixel correction unit, residual computations and quantifying unit, context parameters updating unit and
Calculate Columbus's parameter K unit;Lossless environment submodule is identical as environment sub-modular structure is damaged, wherein initialization unit, office
Portion's gradient calculates mono- with quantifying unit, residual computations and quantifying unit, context parameters updating unit and calculating Columbus's parameter K
Calculating parameter in member is different, and merging gradient vector unit therein, prediction current pixel and pixel correction unit are adopted
With time-multiplexed mode.
3. a kind of spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method, in any spaceborne light claimed in claims 1-2
Spectrogram is damaged and is realized in hybrid compression system as spectral coverage is lossless, which is characterized in that is comprised the following steps that
Step 1, spectral image data and parameter input: spaceborne spectral image data is input to the caching of the front end in input module
It is cached in FIFO;The parameter of spaceborne spectrum picture is inputted to input module, parameter includes frame start signal, spectrogram image width
Degree, spectrum picture height, spectral image data and enable signal, limits of error near value;Near value determines present image
The environment of compression is to damage or lossless, indicates to indicate damaging when near value is not 0 in lossless environment when near value is 0
Environment;Near value value range is 0-15;Frame synchronizing signal, line synchronising signal are generated according to the parameter of input;
Step 2, input data and control signal: control module plays the enable signal in spectral image data, control signal, frame
Beginning signal, near value, frame synchronizing signal, line synchronising signal are input to mixing compression module;
Step 3, initialization of variable: the variable in the initialization unit of mixing compression module is assigned according to the corresponding near value of every row
Initial value, variable include the first quantized interval threshold values T1, the second quantized interval threshold values T2, third quantized interval threshold values T3, prediction
Bit number qbpp needed for error range range and mapping error value;Damage environment submodule and lossless environment submodule initialization block
The value of element variable is different;
Step 4, partial gradient calculates and quantifies: calculating and the office in quantifying unit first, in accordance with cause and effect formwork calculation partial gradient
Portion's gradient vector, including first partial gradient value D1, the second local gradient value D2, third partial gradient value D3, and according to
Each quantized interval threshold values T1, T2, T3 and near value being arranged, obtains 9 partial gradient quantized intervals, partial gradient quantized interval
Value be -4 to 4,9 local quantized intervals be respectively (bear infinite,-T3], (- T3,-T2], (- T2,-T1], (- T1, -
Near), [- near, near], (near, T1], (T1, T2], (T2, T3], (T3, just infinite), by judging local gradient vectors
In partial gradient value Di, i value be 1 to 3, fall in which local quantized interval obtains corresponding quantized value Qi, thus the amount of obtaining
Local gradient vectors after change, respectively first partial gradient quantized value Q1, the second local gradient amount value Q2, third part ladder
Metrization value Q3;
Step 5, merge gradient vector: in three partial gradient quantized values, the opposite partial gradient quantized value of the same symbol will be worth
Merge, sign bit sign is set to the gradient vector after merging, and according to Q1, Q2, Q3 computation index value Q;
Step 6, current pixel and pixel correction are predicted: spaceborne spectrum picture current pixel predicted value Px is calculated according to fallout predictor;
Current pixel predicted value Px is modified using corrected parameter C [Q], and current pixel predicted value Px is limited in [0,
MAXVAL] in, MAXVAL is possible maximum image pixel value, obtains amendment predicted value Px_c;
Step 7, residual computations, quantify, remap and pixel reconstruction: current pixel value Ix is obtained by spaceborne spectral image data,
Current residue value Errval is obtained with current pixel value Ix, amendment predicted value Px_c and sign bit sign;To residual values Errval
Quantify and remap current residue value Errval for non-negative residual values according to environment or lossless environment is currently damaged
MErrval;In pixel reconstruction, if being reconstructed pixel for subsequent according to current near value in damaging under environment
Prediction;If under lossless environment, using current pixel value Ix as reconstructed pixel value Rx;
Step 8, mixing compression module context parameters update: damaging under lossless environment, carrying out respectively to current residue value
Accumulation calculates, and obtains the accumulated value of residual absolute value, the cumulative frequency of residual values, residual error accumulated value and residual prediction correction value,
These context parameters variables are write into two groups of RAM, completes context parameters and updates;
Step 9, Columbus's coding parameter k is calculated: in the case where damaging environment and lossless environment respectively according to the cumulative frequency of residual values
Golomb coding parameter k is calculated with the accumulated value of residual absolute value;Parameter k Columbus being output in mixing compression module is compiled
In numeral module;
Step 10, damage lossless mixing Columbus coding: Columbus's coding module is by MErrval divided by 2k, obtain divisor and remaining
Number, output export remainder into output module for 11 except several 0 and then output again, complete to encode Columbus of MErrval, make
The output result MErrval for damaging environment submodule and lossless environment submodule above is carried out with Columbus's coding module
Columbus's coding is mixed, spaceborne spectrum picture spectral coverage is completed and damages lossless mixing compression;
Step 11, compression of images result exports: the rear end that the data for mixing Columbus's coding write into output module is cached into FIFO,
And by 16 bit wide outputs, output is the mixing compression result of image.
4. spaceborne spectrum picture spectral coverage according to claim 3 is lossless to damage mixing compression method, it is characterised in that: step
(3) variable that the needs described in initialize will be according to different near values when first data of the every row of spectrum picture arrive
Initialization is primary;T1 is assigned a value of 18, T2 and is assigned a value of 67, T3 and is assigned a value of 276, range being assigned a value of 4096, qbpp under lossless environment
It is assigned a value of 12;In the case where damaging environment, with formula T1=18+3*near, T2=67+5*near, T3=276+7*near, range
=(4095+2*near)/(2*near+1)+1, qbpp=log2(range) value is calculated to the initial of mixing compression module
The variable changed in unit carries out initialization assignment.
5. spaceborne spectrum picture spectral coverage according to claim 3 is lossless to damage mixing compression method, it is characterised in that: step
(7) if described in pixel reconstruction in damaging under environment, pixel can be reconstructed according to current near value, be obtained
Reconstructed pixel Rx;If under lossless environment, using current pixel value Ix as reconstructed pixel value Rx.
6. spaceborne spectrum picture spectral coverage according to claim 3 is lossless to damage mixing compression method, it is characterised in that: step
(8) context parameters variable described in are write into two groups of RAM, and completing context parameters update is that parameter enters two groups of ram progress
Storage, lossless A1 [Q], N1 [Q], B1 [Q], C1 [Q] value enter one group of ram, the A damaged [Q], N [Q], B [Q], C [Q] value into
Enter another group of ram, prevents the interference under varying environment between parameter, specifically
8.1 in the case where damaging environment, carries out accumulation calculating to current residue value, obtains the accumulated value A [Q] for damaging residual absolute value,
The cumulative frequency N [Q] for damaging residual values damages residual error accumulated value B [Q] and damages residual prediction correction value C [Q];
Forecast value revision value C [Q] is constrained in realization in [MIN_C, MAX_C] range by 8.2 to be updated, and updated parameter has been write into
In the RAM for damaging environment;
8.3 under lossless environment, carries out accumulation calculating to current residue value, obtains the accumulated value A1 [Q] of lossless residual absolute value,
The cumulative frequency N1 [Q] of lossless residual values, lossless residual error accumulated value B1 [Q], lossless residual prediction correction value C1 [Q];
8.4 these variables are write into the RAM of lossless environment, complete context parameters and update.
7. spaceborne spectrum picture spectral coverage according to claim 3 is lossless to damage mixing compression method, it is characterised in that: step
(9) the calculating Golomb coding parameter k described in, in the case where damaging environment, basis damages the cumulative frequency that residual values are damaged in environment
N [Q] and the accumulated value A [Q] for damaging prediction residual absolute value calculate Golomb coding parameter k;According to lossless under lossless environment
Accumulated value A1 [Q] the calculating parameter k of the cumulative frequency N1 [Q] and non-destructive prediction residual absolute value of lossless residual values in environment.
8. spaceborne spectrum picture spectral coverage according to claim 3 is lossless to damage mixing compression method, it is characterised in that: step
(5), institute in merging gradient vector unit described in step (6), prediction current pixel and pixel correction unit and step (10)
Columbus's coding module stated is all made of time-multiplexed mode.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811269511.8A CN109151482B (en) | 2018-10-29 | 2018-10-29 | Lossless lossy hybrid compression method for spectral band of satellite-borne spectral image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811269511.8A CN109151482B (en) | 2018-10-29 | 2018-10-29 | Lossless lossy hybrid compression method for spectral band of satellite-borne spectral image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109151482A true CN109151482A (en) | 2019-01-04 |
CN109151482B CN109151482B (en) | 2021-03-23 |
Family
ID=64806646
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811269511.8A Active CN109151482B (en) | 2018-10-29 | 2018-10-29 | Lossless lossy hybrid compression method for spectral band of satellite-borne spectral image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109151482B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113092382A (en) * | 2021-03-16 | 2021-07-09 | 上海卫星工程研究所 | Fourier transform spectrometer on-satellite data lossless compression method and system |
CN113207004A (en) * | 2021-04-30 | 2021-08-03 | 华中科技大学 | Remote sensing image compression algorithm hardware implementation method based on JPEG-LS interframe expansion |
CN113470127A (en) * | 2021-09-06 | 2021-10-01 | 成都国星宇航科技有限公司 | Optical image effective compression method based on satellite-borne cloud detection |
CN113613001A (en) * | 2021-09-02 | 2021-11-05 | 沈阳航空航天大学 | Image high-speed compression method and system based on FPGA under CCSDS standard |
CN113727102A (en) * | 2021-11-01 | 2021-11-30 | 基石酷联微电子技术(北京)有限公司 | Low-delay lossy image encoder and decoder |
CN113766238A (en) * | 2021-09-16 | 2021-12-07 | 沈阳航空航天大学 | FPGA-based image high-speed compression method and system under JPEG-LS lossless compression standard |
CN114143552A (en) * | 2021-11-23 | 2022-03-04 | 沈阳航空航天大学 | FPGA-based image high-speed decompression system and method under CCSDS standard |
CN114693818A (en) * | 2022-04-14 | 2022-07-01 | 北京道达天际科技有限公司 | Compression method suitable for digital ortho image data |
CN116600135A (en) * | 2023-06-06 | 2023-08-15 | 广州大学 | Lossless compression-based traceability graph compression method and system |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100002946A1 (en) * | 2008-07-02 | 2010-01-07 | Texas Instruments Incorporated | Method and apparatus for compressing for data relating to an image or video frame |
CN101771874A (en) * | 2009-12-31 | 2010-07-07 | 华中科技大学 | Satellite image compression method and device for realizing satellite image compression |
CN102970531A (en) * | 2012-10-19 | 2013-03-13 | 西安电子科技大学 | Method for implementing near-lossless image compression encoder hardware based on joint photographic experts group lossless and near-lossless compression of continuous-tone still image (JPEG-LS) |
US20150043815A1 (en) * | 2011-10-30 | 2015-02-12 | Hd2 Technologies Pty. Ltd. | Perceptually lossless and perceptually enhanced image compression system & method |
CN105825530A (en) * | 2016-03-21 | 2016-08-03 | 辽宁师范大学 | Coastal zone hyperspectral image distributed lossy coding and decoding method based on region of interest |
CN105828070A (en) * | 2016-03-23 | 2016-08-03 | 华中科技大学 | Anti-error code propagation JPEG-LS image lossless/near-lossless compression algorithm hardware realization method |
-
2018
- 2018-10-29 CN CN201811269511.8A patent/CN109151482B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100002946A1 (en) * | 2008-07-02 | 2010-01-07 | Texas Instruments Incorporated | Method and apparatus for compressing for data relating to an image or video frame |
CN101771874A (en) * | 2009-12-31 | 2010-07-07 | 华中科技大学 | Satellite image compression method and device for realizing satellite image compression |
US20150043815A1 (en) * | 2011-10-30 | 2015-02-12 | Hd2 Technologies Pty. Ltd. | Perceptually lossless and perceptually enhanced image compression system & method |
CN102970531A (en) * | 2012-10-19 | 2013-03-13 | 西安电子科技大学 | Method for implementing near-lossless image compression encoder hardware based on joint photographic experts group lossless and near-lossless compression of continuous-tone still image (JPEG-LS) |
CN105825530A (en) * | 2016-03-21 | 2016-08-03 | 辽宁师范大学 | Coastal zone hyperspectral image distributed lossy coding and decoding method based on region of interest |
CN105828070A (en) * | 2016-03-23 | 2016-08-03 | 华中科技大学 | Anti-error code propagation JPEG-LS image lossless/near-lossless compression algorithm hardware realization method |
Non-Patent Citations (2)
Title |
---|
SHEN-CHUAN TAI: "A Near-Lossless Compression Method Based on CCSDS for Satellite Images", 《2012 INTERNATIONAL SYMPOSIUM ON COMPUTER, CONSUMER AND CONTROL》 * |
粘永健: "基于多波段预测的高光谱图像分布式无损压缩", 《光学精密工程》 * |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113092382A (en) * | 2021-03-16 | 2021-07-09 | 上海卫星工程研究所 | Fourier transform spectrometer on-satellite data lossless compression method and system |
CN113207004A (en) * | 2021-04-30 | 2021-08-03 | 华中科技大学 | Remote sensing image compression algorithm hardware implementation method based on JPEG-LS interframe expansion |
CN113207004B (en) * | 2021-04-30 | 2024-02-02 | 华中科技大学 | Remote sensing image compression algorithm hardware implementation method based on JPEG-LS (joint photographic experts group-LS) inter-frame expansion |
CN113613001B (en) * | 2021-09-02 | 2023-05-19 | 沈阳航空航天大学 | Image high-speed compression method and system based on FPGA under CCSDS standard |
CN113613001A (en) * | 2021-09-02 | 2021-11-05 | 沈阳航空航天大学 | Image high-speed compression method and system based on FPGA under CCSDS standard |
CN113470127A (en) * | 2021-09-06 | 2021-10-01 | 成都国星宇航科技有限公司 | Optical image effective compression method based on satellite-borne cloud detection |
CN113766238A (en) * | 2021-09-16 | 2021-12-07 | 沈阳航空航天大学 | FPGA-based image high-speed compression method and system under JPEG-LS lossless compression standard |
CN113766238B (en) * | 2021-09-16 | 2023-07-04 | 沈阳航空航天大学 | Image high-speed compression method and system based on FPGA under JPEG-LS lossless compression standard |
CN113727102A (en) * | 2021-11-01 | 2021-11-30 | 基石酷联微电子技术(北京)有限公司 | Low-delay lossy image encoder and decoder |
CN113727102B (en) * | 2021-11-01 | 2022-01-04 | 基石酷联微电子技术(北京)有限公司 | Low-delay lossy image encoder and decoder |
CN114143552A (en) * | 2021-11-23 | 2022-03-04 | 沈阳航空航天大学 | FPGA-based image high-speed decompression system and method under CCSDS standard |
CN114693818A (en) * | 2022-04-14 | 2022-07-01 | 北京道达天际科技有限公司 | Compression method suitable for digital ortho image data |
CN116600135A (en) * | 2023-06-06 | 2023-08-15 | 广州大学 | Lossless compression-based traceability graph compression method and system |
CN116600135B (en) * | 2023-06-06 | 2024-02-13 | 广州大学 | Lossless compression-based traceability graph compression method and system |
Also Published As
Publication number | Publication date |
---|---|
CN109151482B (en) | 2021-03-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109151482A (en) | Spaceborne spectrum picture spectral coverage is lossless to damage mixing compression method | |
US11935272B2 (en) | Point cloud compression | |
CN111787323A (en) | Variable bit rate generation type compression method based on counterstudy | |
US9111333B2 (en) | Predictive position encoding | |
CN100556148C (en) | Video information coding and coding/decoding method, motion compensated video encoder and corresponding decoder | |
WO2013067673A1 (en) | Predictive position decoding | |
CN102369522A (en) | Parallel, pipelined, integrated-circuit implementation of a computational engine | |
CN1843039B (en) | System and method for encoding and decoding enhancement layer data using descriptive model parameters | |
CN113258935B (en) | Communication compression method based on model weight distribution in federated learning | |
WO2022257971A1 (en) | Point cloud encoding processing method, point cloud decoding processing method, and related device | |
CN100534188C (en) | An image compression method and device | |
CN107666472B (en) | Method and apparatus for hybrid digital-analog coding | |
CN103503320B (en) | For reconstructing method and the decoder of source signal | |
WO2022062369A1 (en) | Point cloud encoding and decoding method and system, and point cloud encoder and point cloud decoder | |
CN117097906B (en) | Method and system for efficiently utilizing regional medical resources | |
CN103297776A (en) | On-board non-registering multispectral image lossless/lossy encoding system and on-board non-registering multispectral image lossless/lossy encoding method | |
CN102695055B (en) | JPEG_LS (Joint Pho-tographic Experts Group-Lossless Standard) bit rate control method under high bit rate | |
CN103227644A (en) | Compression method of automobile body small-format data | |
US6909746B2 (en) | Fast robust data compression method and system | |
CN107205151B (en) | Coding and decoding device and method based on mixed distortion measurement criterion | |
Shin et al. | RL-SPIHT: reinforcement learning-based adaptive selection of compression ratios for 1-D SPIHT algorithm | |
De Silva et al. | Exploring the Implementation of JPEG Compression on FPGA | |
CN107682017A (en) | A kind of spaceborne magnetometer magnetic field data compression processing method and device | |
CN116250008A (en) | Encoding and decoding methods, encoder, decoder and encoding and decoding system of point cloud | |
Zhang et al. | Design and Implementation of Lossless Compression System for CCSDS Hyperspectral Images |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |