CN108989817A - A kind of radar data compression method based on reference frame dislocation prediction - Google Patents
A kind of radar data compression method based on reference frame dislocation prediction Download PDFInfo
- Publication number
- CN108989817A CN108989817A CN201811021057.4A CN201811021057A CN108989817A CN 108989817 A CN108989817 A CN 108989817A CN 201811021057 A CN201811021057 A CN 201811021057A CN 108989817 A CN108989817 A CN 108989817A
- Authority
- CN
- China
- Prior art keywords
- data
- frame
- difference
- reference frame
- dislocation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
-
- 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
- H04N19/17—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 an image region, e.g. an object
- H04N19/172—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 an image region, e.g. an object the region being a picture, frame or field
-
- 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
-
- 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
- H04N19/593—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving spatial prediction techniques
-
- 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
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention discloses a kind of radar data compression methods based on reference frame dislocation prediction, the strongest optimal reference frame of each original frame data correlation is determined first, optimal reference frame is subjected to location dislocation structure forecast frame in frame, present frame is predicted that the difference data of frame does compression and the decompression of Block Adaptive Quantization algorithm with it, the difference data being restored, again the location information of reference frame data dislocation coding is sent to decoding end after Huffman lossless coding, the difference data of recovery is then added the data reconstructed with prediction frame.The present invention determines frame data reference frame, interframe dislocation prediction prediction, Block-adaptive quantization algorithm, Huffman encoding, frame data difference and reconfiguration technique using frame-to-frame correlation, under the premise of guaranteeing certain calculation amount, higher Y-PSNR and data reconstruction degree are realized, has the characteristics that inter-prediction correlation is strong, differential process quantization error is small, reconstruct data restoring degree is high.
Description
Technical field
The invention belongs to Image Compression fields, and in particular to a kind of radar data pressure based on reference frame dislocation prediction
Contracting method.
Background technique
With succeeding in sending up for " Chang'e I " moon exploration program, indicate that the process forward direction deep space exploration hair of space is explored by China
Exhibition.Deep space exploration refers to the detection to the moon and translunar celestial body and space progress, and main purpose is to develop and utilize
Space resources, development space technology carry out scientific research, explore the origin of the solar system and universe, extend the living space of the mankind,
Develop service for the long-term sustainable of human society.Deep space exploration has important military and political meaning, makes the mankind in new generation
One of three key airline companies day activity of discipline.
The main tool for carrying out deep space exploration research includes radio science, radar and radio astronomy, wherein radar by
In that no matter its unique feature plays the role of in DSN deep space network or in target property detects is especially important.It is real
Shi Xingqiang, metrical information are abundant, can initiatively, round-the-clock ground-to-air target detected.
In deep space exploration task, secondary surface layer detection radar is equipped in device as important modern radar load,
The information such as the following geologic structure in secondary surface layer and layering for obtaining detected celestial body.As Radar Technology is to ultrahigh resolution side
To development, radar bandwidth is increasing, so that radar data amount sharply increases.One side is full, the average data rate of high resolution radar
It is very big, typically much deeper than 2Mbps；On the other hand, by taking mars exploration as an example, the effective downlink data transmission ability of satellite has
Limit, about 1-2Mbps.Deep space data downstream transmittability is limited, once the important bottleneck to influence science data acquisition.Therefore,
There is an urgent need to design reasonable compression algorithm for deep space time surface layer detection radar data characteristics, solves data and store and transmit
The problem of.
In a certain range, the celestial body time following geologic structure in surface layer and hierarchical information are gradual along earth's surface, this makes phase
The radar return of adjacent secondary surface region reflection is similar, and the data of radar consecutive frame road (radar return of adjacent area) have phase
Guan Xing.Using this correlation, value is made the difference to adjacent frame data, the dynamic range of difference data can substantially subtract relative to former data
It is small.
Under conditions of quantized interval quantity is certain, the dynamic range of data is smaller, and quantized interval length is smaller, caused
Quantization error is also just smaller.Therefore, dynamic range can be reduced by difference, thus lower quantization error.
In the stronger region of radar return, difference data dynamic range is substantially reduced, and most related to current frame data
Be not necessarily its previous frame data, so before carrying out calculus of differences, need to carry out frame data several before present frame pre-
It surveys, finds out certain maximally related frame data and carry out difference.A certain data and the data of corresponding reference frame same position in present frame
Difference is not necessarily minimum compared with the difference of the data of reference frame other positions.For this purpose, the position of reference frame data can be adjusted
It is whole, m are moved up or down, selection carrys out structure forecast frame with the smallest data of current frame data difference in 2m+1 data.
Summary of the invention
In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is that providing a kind of based on reference
The radar data compression method of frame dislocation prediction, with inter-prediction correlation is strong, compression performance is high, differential process quantization error
Small, the advantages that quantitative graphs are high, reconstructed image restoring degree is high.
The invention adopts the following technical scheme:
A kind of radar data compression method based on reference frame dislocation prediction, it is first determined each original frame data correlation is most
Optimal reference frame is carried out location dislocation structure forecast frame in frame, present frame is predicted to the difference of frame with it by strong optimal reference frame
Value Data does compression and the decompression of Block Adaptive Quantization algorithm, the difference data being restored, then reference frame data is misplaced and is compiled
The location information of code is sent to decoding end after Huffman lossless coding, is then added the difference data of recovery with prediction frame
To the data of reconstruct.
Specifically, the following steps are included:
S1, the data for taking certain surface layer detection radar to acquire carry out compression emulation experiment, choose effective echo data and carry out
Compression emulation, and quantize data as 16 storing datas as original frame data；
S2, for the present frame in original frame data, it is done into difference with multinomial original frame data before；
S3, its 1- norm and small data accounting are asked to each difference data obtained in step S2, by 1- norm and decimal
Index according to the product of accounting as evaluation reference frame；
S4, repetition step S3 obtain the corresponding optimal reference frame data of each frame data；
S5, the reference frame data for choosing a certain data corresponding position in present frame, by the reference frame data and its front and back m
Data are compared with the specified data of present frame, frame data number of the most similar numerical value of selection as the prediction frame position
Value, the prediction frame of present frame is successively constructed according to the method, and save the location information of dislocation；
S6, the location information of the reference frame data dislocation coding in step S5 is transmitted to solution after Huffman encoding decodes
Code end, is used for recovery and rebuilding data；
S7, the difference data that difference obtains is carried out to present frame and prediction frame obtained in step s5；
S8, it step S7 is obtained to difference data is transferred to decoding after Block-adaptive quantization algorithm is compressed, decompressed
End, be restored difference；
S9, the recovery difference value of the predicted frame data of step S5 and step S8 are obtained into reconstructed frame data；
S10, the reconstructed frame data obtained by initial data and the step S9 of step S1 calculate Y-PSNR PSNR and force
True degree K judges to reconstruct the reconstruct degree of data, completion data reconstruction jointly.
Further, in step S2, difference traversal is done to the n-1 frame data before present frame, each frame data are most
The maximum value d of positional distance of the excellent reference frame away from present frame.
Further, in step S3,1- norm and small data accounting are asked to difference respectively, 1- norm judges data dynamic model
It encloses, small data accounting judges quantization error；
For array A=[a1,a2,a2,...,an], 1- norm is defined as follows:
Norm1 (A)=| a1|+|a2|+...+|an|
Wherein, a indicates that the element in array A, n indicate the element number in array A；
Difference data is normalized, taking the half of the maxima and minima of data after normalizing is threshold
The data definition for being less than threshold value after normalization is small data, it is as follows to define small data accounting by value:
Small_data_ratio=N1/N2
Wherein, N1 is the quantity of small data, and N2 is total amount of data.
Further, in step S5, misplace prediction technique are as follows:
It moves up or moves down m for the position of reference frame data to adjust, selection and current frame data in 2m+1 data
The smallest data of difference carry out structure forecast frame.
Further, in step S6, Huffman encoding method is used to the location information of reference frame dislocation, uses elongated volume
Code table encodes source symbol, and wherein variable length coding table is obtained by a kind of method for assessing source symbol occurrence probability
, occurrence probability it is high letter use shorter coding, otherwise occurrence probability it is low then use longer coding.
Further, in step S6, to the specific coding of location information, steps are as follows:
S601, source symbol, that is, misalignment position data probability is arranged in descending order；
S602, two the smallest probability is added, and continues this step, higher probability branch is placed on the right always,
To the last become probability 1；
S603, it draws by probability 1 to the path of each source symbol, sequentially writes down along the 0 of path and 1, obtain the symbol
Number Huffman code word；
S604, each pair of one combined, left side is appointed as 0, one, the right is appointed as 1.
Further, in step S8, piecemeal is carried out to difference data, mean value, variance in calculation block, by each data block into
It is 0 that row normalized, which meets mean value, the Gaussian Profile that variance is 1, to be quantified based on (0,1) Gaussian Profile, is calculated
Output level is encoded and is transmitted by output level, recovers reconstruct difference by data block, data block variance and output level
Data.
Further, in step S9, it is as follows to define Y-PSNR PSNR for the data stored for 16:
PSNR=s/e
Wherein, s=655352,
Further, it is as follows to define fidelity K:
Wherein, f (i, j) is data after compression, and g (i, j) is unpressed initial data.
Compared with prior art, the present invention at least has the advantages that
It is adaptive with traditional piecemeal the present invention provides a kind of radar data compression method based on reference frame dislocation prediction
It answers quantization algorithm to compare, radar initial data consecutive frame track data is utilized with correlation and related frame data residual error data
The characteristics of dynamic range can substantially reduce relative to former data devises the radar data pressure based on reference frame data dislocation prediction
Frame data are passed through the dynamic model of reduction difference data by contracting new method, the algorithm after its optimal related frame data is compared with misplacing
It encloses to construct prediction frame, carries out compression and the decompression of BAQ algorithm by the difference of prediction frame and original frame data, and by the position of dislocation
Confidence breath is transferred to decoding end by Huffman encoding method, finally by the method recovery and rebuilding data of iteration, by side of the invention
The data that method constructs not only Y-PSNR with higher, reconstruct degree also with higher.
Further, present frame does the process that difference itself is a traversal with multinomial frame data before, but due to this
Invention needs to seek present frame the strongest frame data of its correlation, referred to as optimal reference frame, therefore after primary traversal
The position that optimal frames can be grasped, after its location information record, when needing to calculate correlation again, location information can be direct
It uses, to reduce calculation amount.
Further, in present frame a certain data and the difference of the data of corresponding reference frame same position and reference frame other
The difference of the data of position is minimum compared to not necessarily.For this purpose, the position of reference frame data can adjust, m is moved up or down
Position, selection carrys out structure forecast frame with the smallest data of current frame data difference in 2m+1 data, pre- after position adjusts
1 norm value of difference data for surveying frame and present frame gradually becomes smaller, and can reduce the calculation amount of BAQ algorithm, and increase recovery accuracy.
Further, Huffman encoding method is a kind of lossless coding method, and the misalignment position information of reference frame is passed through
Huffman encoding decoding is transmitted to after decoding end the accuracy that can guarantee misalignment position information, due to the misalignment position of decoding end
Information is mainly used for the reconstruct of frame data, to also improve the accuracy of frame data reconstruct.
Further, BAQ algorithm generally applies to the quantization compression of radar initial data, because of Deep space radar data consecutive frame
The data in road have correlation, make the difference value to adjacent frame data using this correlation, the dynamic range of difference data relative to
Former data can substantially reduce, thus to difference data be BAQ compression can not only reduce amount of calculation, compression can also be improved
The decompression accuracy of data.
In conclusion the present invention determines frame data reference frame, interframe dislocation prediction prediction, piecemeal certainly using frame-to-frame correlation
Quantization algorithm (BAQ), Huffman encoding, frame data difference and reconfiguration technique are adapted to, it is real under the premise of guaranteeing certain calculation amount
Higher Y-PSNR and data reconstruction degree are showed, with inter-prediction correlation is strong, differential process quantization error is small, reconstruct
The features such as data restoring degree is high.
Below by drawings and examples, technical scheme of the present invention will be described in further detail.
Detailed description of the invention
Fig. 1 is general flow chart of the invention；
Fig. 2 is the Compress softwares flow chart of piecemeal BAQ (adaptive quantizing) algorithm.
Specific embodiment
The present invention provides a kind of radar data compression methods based on reference frame dislocation prediction, certain surface layer is detected
The data of radar acquisition, first determine the strongest optimal reference frame of each original frame data correlation, and optimal reference frame is carried out frame
The dislocation of interior position carrys out structure forecast frame, and present frame is predicted that the difference data of frame is BAQ (Block Adaptive Quantization algorithm) with it
Compression and decompression, the difference data being restored, then the location information of dislocation is sent to solution after Huffman lossless coding
Then the difference data of recovery is added the data reconstructed with prediction frame by code end.This method has inter-prediction correlation
By force, the advantages that compression performance is high, differential process quantization error is small, quantitative graphs are high, reconstructed image restoring degree is high.
Referring to Fig. 1, a kind of radar data compression method based on reference frame dislocation prediction of the present invention, including following step
It is rapid:
S1, the data for taking certain surface layer detection radar to acquire carry out compression emulation experiment, choose wherein size be 2000 ×
24576 effective echo data carries out compression emulation, and data are quantified as 16 storing datas by A/D, as the present invention
The initial data used；
S2, for present frame, it is done into difference with multinomial original frame data before；
First to the n-1 frame data before present frame do a difference traversal, because after the step of in only consider selection and work as
Thus the preceding strongest data of frame correlation reach the optimum efficiency for restoring data, so actual step S2 can cancel time
The process gone through reduces the calculation amount of this process with this, this just needs one group of location information, it may be assumed that the optimal ginseng of each frame data
Examine the maximum value d of the positional distance of frame pitch present frame；
Therefore, it is done difference with d item frame data before for calculating optimal reference frame by the calculating process of step S2；
S3, its 1 norm and small data accounting are asked to each difference data obtained in step S2, by multiplying for this two term coefficient
Index of the product as evaluation reference frame；
The purpose of inter-prediction is to find optimal reference frame, that is, is found out and the strongest frame data of current frame correlation.Prediction
The design of index should can reflect the dynamic range of difference data, make the quantization error to difference minimum again.Here it integrates
Accuracy and computation complexity, ask 1- norm and small data accounting to difference respectively, and 1- norm judges data dynamic range, decimal
Quantization error is judged according to accounting, finally using two indices product as final prediction index.
Wherein:
(1) 1- norm: norm1 (A)
For array A=[a1,a2,a2,...,an], 1- norm is defined as:
Norm1 (A)=| a1|+|a2|+...+|an|
1- norm numerical value is smaller, and dynamic range is smaller.
(2) small data accounting: Small_data_ratio (A)
Difference data is normalized, taking the half of the maxima and minima of data after normalizing is threshold
Value will be small data, quantity N1, total amount of data N2, then small data accounting less than the data definition of threshold value after normalization
Is defined as:
Small_data_ratio=N1/N2
Obviously, the small data of numerical value are more, and quantization error is smaller；Determine that each frame data are corresponding according to two above index
Optimal reference frame data.
S4, step S3 is repeated, the corresponding optimal reference frame data of each frame data is successively obtained by the above index；
S5, the reference frame data for choosing a certain data corresponding position in present frame, by the reference frame data and its front and back m
Data are compared with the specified data of present frame, frame data number of the most similar numerical value of selection as the prediction frame position
Value, the prediction frame of present frame is successively constructed according to the method, and save the location information of dislocation；
The prediction technique that misplaces is as follows:
The number of a certain data and the difference of the data of corresponding reference frame same position and reference frame other positions in present frame
According to difference it is minimum compared to not necessarily.For this purpose, the position of reference frame data can adjust, m are moved up or down, in 2m+1
Selection carrys out structure forecast frame with the smallest data of current frame data difference in a data, and frame and current is predicted after position adjusts
1 norm value of difference data of frame gradually becomes smaller, and can reduce the calculation amount of BAQ algorithm, and increase recovery accuracy.
S6, the location information of the reference frame data dislocation coding in step S5 is transmitted to solution after Huffman encoding decodes
Code end, is used for recovery and rebuilding data；
The Huffman encoding method of lossless compression is used to the location information of reference frame dislocation:
Huffman encoding encodes source symbol using variable length coding table, and wherein variable length coding table is by a kind of assessment
What the method for source symbol occurrence probability obtained, the high letter of occurrence probability uses shorter coding, and breeding occurrence probability is low
Longer coding is then used, this just reduces the average length of the character string after coding, desired value, to reach lossless compression
The purpose of data.
To the specific coding of location information, steps are as follows:
S601, source symbol, that is, misalignment position data probability is pressed into reduced decision queue；
S602, the step for two the smallest probability being added, and being continued, is placed on the right side for higher probability branch always
Side to the last becomes probability 1；
S603, it draws by probability 1 to the path of each source symbol, sequentially writes down along the 0 of path and 1, gained is exactly
The Huffman code word of the symbol；
S604, each pair of one combined, left side is appointed as to 0, one, the right is appointed as 1 (or opposite).
Wherein, the probability of source symbol is identical.
S7, differential process: the difference data of present frame and prediction frame obtained in step s5 is done；
The purpose of differential process is to reduce data dynamic range, and differential mode is generally two frame initial data and carries out difference,
But the quantizing noise of this model can accumulate, and influence reconstruction result.For the accumulation of error, basic difference model is improved, with original
Beginning data and former frame reconstruct data carry out difference, using this differential mode as difference model used herein.
Step S7 is obtained difference data after BAQ process compresses, decompression and transmitted by S8, Block-adaptive quantization algorithm
To decoding end, be restored difference；
S9, restructuring procedure: the recovery difference value of step S5 prediction frame and step S8 is obtained into reconstructed frame data；
Reconstructing method is that reconstruct data are obtained by iterative formula below:
…
Reconstruct index:
(1) Y-PSNR is defined as:
PSNR=s/e
Wherein, the data 16 stored: s=655352, e is noise power caused by quantifying；
(2) fidelity is defined as:
Wherein, f (i, j) is data after compression, and g (i, j) is unpressed initial data；K value illustrates to reconstruct closer to 1
Data and initial data similarity are higher.
S10, Y-PSNR and fidelity are calculated by the reconstructed frame data that the initial data and step S9 of step S1 obtain,
Judge the reconstruct degree of reconstruct data jointly by two above index.
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.The present invention being described and shown in usually here in attached drawing is real
The component for applying example can be arranged and be designed by a variety of different configurations.Therefore, below to the present invention provided in the accompanying drawings
The detailed description of embodiment be not intended to limit the range of claimed invention, but be merely representative of of the invention selected
Embodiment.Based on the embodiments of the present invention, those of ordinary skill in the art are obtained without creative efforts
The every other embodiment obtained, shall fall within the protection scope of the present invention.
It takes the frame track data of deep space exploration radar as initial data used in the present invention, takes wherein frame conduct current
It is done difference with multinomial frame data before by frame, does the process that this part of difference itself is a traversal, but due to this hair
It is bright to need to seek present frame the strongest frame data of its correlation, referred to as optimal reference frame, therefore after primary traversal i.e.
The position that optimal frames can be grasped, after its location information record, when needing to calculate correlation again, location information can directly make
With to reduce calculation amount；
Wherein, the design of prediction index should can reflect the dynamic range of difference data, make the quantization to difference again
Error is minimum.Here accuracy and computation complexity are integrated, 1- norm and small data accounting, the judgement of 1- norm are asked to difference respectively
Data dynamic range, small data accounting judge quantization error, finally using two indices product as final prediction index.
The number of a certain data and the difference of the data of corresponding reference frame same position and reference frame other positions in present frame
According to difference it is minimum compared to not necessarily.
For this purpose, the position of reference frame data is adjusted, m are moved up or down, selects and works as in 2m+1 data
The smallest data of previous frame data difference carry out structure forecast frame, and 1 model of difference data of frame and present frame is predicted after position adjusts
Numerical value gradually becomes smaller, and can reduce the calculation amount of BAQ algorithm, and increases recovery accuracy.
The Huffman encoding method of lossless compression is used to the location information of reference frame dislocation.It next is exactly difference mistake
Journey: obtaining present frame and predicts the difference data of frame, and most basic differential mode is exactly to carry out difference with two frame initial data,
But the quantizing noise of this model can accumulate, and influence reconstruction result.
For the accumulation of error, basic difference model is improved, carries out difference with initial data and former frame reconstruct data, it will
This differential mode can achieve the effect for reducing quantization error as difference model used in the present invention.
Piecemeal is carried out to difference data, each data block is normalized, makes it by mean value, variance in calculation block
Meet mean value be 0, variance be 1 Gaussian Profile.
To be quantified based on (0,1) Gaussian Profile, output level is calculated, output level is encoded and transmitted,
Reconstruct difference data is recovered by data block, data block variance and output level.
Finally, prediction frame and recovery difference value are obtained reconstructed frame data, calculated by reconstructed frame data and initial data
Y-PSNR and fidelity are judged the reconstruct degree for reconstructing data by two above index jointly, and compression ratio is by the original before compressing
Beginning data and the total amount of data of location information obtain compared with its respectively compressed total amount of data.
Main thought based on the radar data compression method of AR model (autoregression model) frame prediction is determining each original
The optimal reference frame and suboptimum reference frame of frame data obtain prediction frame after reference frame is carried out AR model inter-prediction, will be current
Frame predicts the difference data that the difference of frame does the compression of BAQ and decompression is restored with it, then by the difference data of recovery with
Prediction frame is added the data reconstructed.This method is strong with inter-prediction correlation, differential process quantization error is small, reconstruct image
The advantages that high as restoring degree.
The present invention is only it needs to be determined that the optimal reference frame of original frame data, misplace in frame to structure to optimal reference frame
Prediction frame is built, AR model is eliminated, increases misalignment position information, and encoded with Huffman lossless transmission method, we
The prediction frame of method building and the difference data dynamic range of present frame are smaller, after BAQ compression algorithm decompression, the method for the present invention structure
The data built out have higher Y-PSNR, and specific experiment data are shown in Table 1:
Table 1 is the present invention compared with the performance of BAQ algorithm and AR model frame prediction algorithm
The compression effectiveness of compression algorithm ratio BAQ algorithm it can be seen from experimental result and upper table by dislocation is well very
It is more, and the digit that misplaces is more, and Y-PSNR is bigger under same compression ratio, and compression effectiveness is better.
The above content is merely illustrative of the invention's technical idea, and this does not limit the scope of protection of the present invention, all to press
According to technical idea proposed by the present invention, any changes made on the basis of the technical scheme each falls within claims of the present invention
Protection scope within.
Claims (10)
1. a kind of radar data compression method based on reference frame dislocation prediction, which is characterized in that determine each original frame number first
According to the strongest optimal reference frame of correlation, optimal reference frame is subjected to location dislocation structure forecast frame in frame, by present frame and its
The difference data of prediction frame does compression and the decompression of Block Adaptive Quantization algorithm, the difference data being restored, then by reference frame
The location information of data dislocation coding is sent to decoding end after Huffman lossless coding, then by the difference data of recovery and in advance
It surveys frame and is added the data reconstructed.
2. a kind of radar data compression method based on reference frame dislocation prediction according to claim 1, which is characterized in that
The following steps are included:
S1, the data for taking certain surface layer detection radar to acquire carry out compression emulation experiment, choose effective echo data and are compressed
Emulation, and quantize data as 16 storing datas as original frame data；
S2, for the present frame in original frame data, it is done into difference with multinomial original frame data before；
S3, its 1- norm and small data accounting are asked to each difference data obtained in step S2,1- norm and small data is accounted for
Index of the product of ratio as evaluation reference frame；
S4, repetition step S3 obtain the corresponding optimal reference frame data of each frame data；
S5, the reference frame data for choosing a certain data corresponding position in present frame, by the reference frame data and its front and back m-bit data
It is compared with the specified data of present frame, frame data numerical value of the most similar numerical value of selection as the prediction frame position, root
Method successively constructs the prediction frame of present frame accordingly, and saves the location information of dislocation；
S6, the location information of the reference frame data dislocation coding in step S5 is transmitted to decoding after Huffman encoding decodes
End is used for recovery and rebuilding data；
S7, the difference data that difference obtains is carried out to present frame and prediction frame obtained in step s5；
S8, it step S7 is obtained to difference data is transferred to decoding end after Block-adaptive quantization algorithm is compressed, decompressed, obtain
To recovery difference；
S9, the recovery difference value of the predicted frame data of step S5 and step S8 are obtained into reconstructed frame data；
S10, Y-PSNR PSNR and fidelity K is calculated by the reconstructed frame data that the initial data and step S9 of step S1 obtain
The reconstruct degree of common judgement reconstruct data, completes data reconstruction.
3. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S2, difference traversal is done to the n-1 frame data before present frame, the optimal reference frame of each frame data is away from present frame
The maximum value d of positional distance.
4. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S3,1- norm and small data accounting are asked to difference respectively, 1- norm judges data dynamic range, the judgement of small data accounting
Quantization error；
For array A=[a1,a2,a2,...,an], 1- norm is defined as follows:
Norm1 (A)=| a1|+|a2|+...+|an|
Wherein, a indicates that the element in array A, n indicate the element number in array A；
Difference data is normalized, taking the half of the maxima and minima of data after normalizing is threshold value,
It is small data by the data definition for being less than threshold value after normalization, it is as follows defines small data accounting:
Small_data_ratio=N1/N2
Wherein, N1 is the quantity of small data, and N2 is total amount of data.
5. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S5, misplace prediction technique are as follows:
It moves up or moves down m for the position of reference frame data to adjust, selection and current frame data difference in 2m+1 data
The smallest data carry out structure forecast frame.
6. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S6, Huffman encoding method is used to the location information of reference frame dislocation, source symbol is carried out using variable length coding table
Coding, wherein variable length coding table is obtained by a kind of method for assessing source symbol occurrence probability, the high word of occurrence probability
Mother use shorter coding, otherwise occurrence probability it is low then use longer coding.
7. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S6, to the specific coding of location information, steps are as follows:
S601, source symbol, that is, misalignment position data probability is arranged in descending order；
S602, two the smallest probability is added, and continues this step, higher probability branch is placed on the right always, until
Eventually become probability 1；
S603, it draws by probability 1 to the path of each source symbol, sequentially writes down along the 0 of path and 1, obtain the symbol
Huffman code word；
S604, each pair of one combined, left side is appointed as 0, one, the right is appointed as 1.
8. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S8, piecemeal is carried out to difference data, each data block is normalized and meets by mean value, variance in calculation block
Mean value is 0, the Gaussian Profile that variance is 1, to be quantified based on (0,1) Gaussian Profile, calculates output level, by output electricity
It is flat to be encoded and transmitted, reconstruct difference data is recovered by data block, data block variance and output level.
9. a kind of radar data compression method based on reference frame dislocation prediction according to claim 2, which is characterized in that
In step S9, it is as follows to define Y-PSNR PSNR for the data stored for 16:
PSNR=s/e
Wherein, s=655352,
10. a kind of radar data compression method based on reference frame dislocation prediction according to claim 9, feature exist
In definition fidelity K is as follows:
Wherein, f (i, j) is data after compression, and g (i, j) is unpressed initial data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811021057.4A CN108989817A (en) | 2018-09-03 | 2018-09-03 | A kind of radar data compression method based on reference frame dislocation prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811021057.4A CN108989817A (en) | 2018-09-03 | 2018-09-03 | A kind of radar data compression method based on reference frame dislocation prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108989817A true CN108989817A (en) | 2018-12-11 |
Family
ID=64548672
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811021057.4A Pending CN108989817A (en) | 2018-09-03 | 2018-09-03 | A kind of radar data compression method based on reference frame dislocation prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108989817A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109799483A (en) * | 2019-01-25 | 2019-05-24 | 中国人民解放军空军研究院战略预警研究所 | A kind of data processing method and device |
| CN112633122A (en) * | 2020-12-17 | 2021-04-09 | 厦门大学 | Front-end mileage calculation method and system of monocular VIO system |
| CN113075624A (en) * | 2021-03-31 | 2021-07-06 | 华中科技大学 | Lossless compression method and device for radar echo sequence and radar system |
| CN115664426A (en) * | 2022-12-27 | 2023-01-31 | 深圳安德空间技术有限公司 | Ground penetrating radar data real-time lossless compression method and system |
| CN116582889A (en) * | 2023-07-14 | 2023-08-11 | 武汉能钠智能装备技术股份有限公司四川省成都市分公司 | High-throughput data transmission and storage method, electronic equipment and storage medium |
| CN116996076A (en) * | 2023-09-27 | 2023-11-03 | 湖北华中电力科技开发有限责任公司 | Intelligent management method for electrical energy consumption data of campus equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102273206A (en) * | 2008-10-31 | 2011-12-07 | Sk电信有限公司 | Method and apparatus for encoding a motion vector, and method and apparatus for encoding/decoding image using same |
| US20120106645A1 (en) * | 2009-06-26 | 2012-05-03 | Huawei Technologies Co., Ltd | Method, apparatus and device for obtaining motion information of video images and template |
| CN108390871A (en) * | 2018-02-09 | 2018-08-10 | 西安交通大学 | A kind of radar data compression method based on the prediction of autoregression model frame |
-
2018
- 2018-09-03 CN CN201811021057.4A patent/CN108989817A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102273206A (en) * | 2008-10-31 | 2011-12-07 | Sk电信有限公司 | Method and apparatus for encoding a motion vector, and method and apparatus for encoding/decoding image using same |
| US20120106645A1 (en) * | 2009-06-26 | 2012-05-03 | Huawei Technologies Co., Ltd | Method, apparatus and device for obtaining motion information of video images and template |
| CN108390871A (en) * | 2018-02-09 | 2018-08-10 | 西安交通大学 | A kind of radar data compression method based on the prediction of autoregression model frame |
Non-Patent Citations (1)
| Title |
|---|
| 毛小明等: "一种基于H.264/AVC的高性能快速运动估计算法", 《计算机应用研究》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109799483A (en) * | 2019-01-25 | 2019-05-24 | 中国人民解放军空军研究院战略预警研究所 | A kind of data processing method and device |
| CN112633122A (en) * | 2020-12-17 | 2021-04-09 | 厦门大学 | Front-end mileage calculation method and system of monocular VIO system |
| CN112633122B (en) * | 2020-12-17 | 2024-01-23 | 厦门大学 | Front-end mileage calculation method and system of monocular VIO system |
| CN113075624A (en) * | 2021-03-31 | 2021-07-06 | 华中科技大学 | Lossless compression method and device for radar echo sequence and radar system |
| CN113075624B (en) * | 2021-03-31 | 2024-03-19 | 华中科技大学 | Lossless compression method and device for radar echo sequence and radar system |
| CN115664426A (en) * | 2022-12-27 | 2023-01-31 | 深圳安德空间技术有限公司 | Ground penetrating radar data real-time lossless compression method and system |
| CN116582889A (en) * | 2023-07-14 | 2023-08-11 | 武汉能钠智能装备技术股份有限公司四川省成都市分公司 | High-throughput data transmission and storage method, electronic equipment and storage medium |
| CN116582889B (en) * | 2023-07-14 | 2023-11-21 | 武汉能钠智能装备技术股份有限公司四川省成都市分公司 | High-throughput data transmission and storage method, electronic equipment and storage medium |
| CN116996076A (en) * | 2023-09-27 | 2023-11-03 | 湖北华中电力科技开发有限责任公司 | Intelligent management method for electrical energy consumption data of campus equipment |
| CN116996076B (en) * | 2023-09-27 | 2023-12-19 | 湖北华中电力科技开发有限责任公司 | Intelligent management method for electrical energy consumption data of campus equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108989817A (en) | A kind of radar data compression method based on reference frame dislocation prediction | |
| CN107832837B (en) | Convolutional neural network compression method and decompression method based on compressed sensing principle | |
| CN101640803B (en) | Progressive distribution type encoding and decoding method and device for multispectral image | |
| CN108960333B (en) | Hyperspectral image lossless compression method based on deep learning | |
| CN105374054A (en) | Hyperspectral image compression method based on spatial spectrum characteristics | |
| CN104811664B (en) | mine video image compression and transmission method | |
| Huang et al. | ECG compression using the context modeling arithmetic coding with dynamic learning vector–scalar quantization | |
| CN103024392B (en) | Method and device for intra-frame mode prediction based on two-dimensional Hadamard transformation | |
| CN110248190A (en) | A kind of compressed sensing based multilayer residual error coefficient image encoding method | |
| CN105791189A (en) | Sparse coefficient decomposition method for improving reconstruction accuracy | |
| CN104952039A (en) | Distributed compressive sensing reconstruction method for images | |
| CN101099669A (en) | Electrocardiogram data compression method and decoding method based on optimum time frequency space structure code | |
| CN103686177A (en) | Image compression and decompression method, device and system | |
| CN102300095A (en) | Fast compression coding method for hyperspectral signal and image compression method | |
| CN105354867A (en) | Hyperspectral image compression algorithm research of adaptive redundant dictionary compressed sensing | |
| CN108390871A (en) | A kind of radar data compression method based on the prediction of autoregression model frame | |
| CN110930408B (en) | Semantic image compression method based on knowledge reorganization | |
| CN113132727A (en) | Scalable machine vision coding method based on image generation | |
| CN102695055B (en) | JPEG_LS (Joint Pho-tographic Experts Group-Lossless Standard) bit rate control method under high bit rate | |
| CN103985096A (en) | Hyperspectral image regression prediction compression method based on off-line training | |
| CN108093455B (en) | High-energy-efficiency wireless sensor network data transmission method based on time-space correlation | |
| CN114422802B (en) | Self-encoder image compression method based on codebook | |
| CN105578183A (en) | Compression sensing video encoding and decoding method based on Gaussian mixture model (GMM) | |
| JP2018033131A (en) | Decoder, encoder and method for decoding encoded value | |
| CN103442236A (en) | Remote sensing signal compressed encoding method of multilevel and fractal dimension vector quantization |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181211 |
|
| RJ01 | Rejection of invention patent application after publication |