CN108921807A - A kind of compression sensing method based on adaptive-filtering - Google Patents
A kind of compression sensing method based on adaptive-filtering Download PDFInfo
- Publication number
- CN108921807A CN108921807A CN201810925896.2A CN201810925896A CN108921807A CN 108921807 A CN108921807 A CN 108921807A CN 201810925896 A CN201810925896 A CN 201810925896A CN 108921807 A CN108921807 A CN 108921807A
- Authority
- CN
- China
- Prior art keywords
- adaptive
- filtering
- measurement result
- resolution ratio
- sensing method
- 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
- 238000001914 filtration Methods 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 23
- 230000006835 compression Effects 0.000 title claims abstract description 19
- 238000007906 compression Methods 0.000 title claims abstract description 19
- 238000005259 measurement Methods 0.000 claims abstract description 39
- 239000011159 matrix material Substances 0.000 claims abstract description 38
- 238000004364 calculation method Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 15
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 7
- 230000006798 recombination Effects 0.000 claims description 3
- 238000005215 recombination Methods 0.000 claims description 3
- 229910017435 S2 In Inorganic materials 0.000 claims 1
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000012730 cephalometry Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of compression sensing methods based on adaptive-filtering, set total observation frequency M according to the image resolution ratio of target image to be detected first, then generate M calculation matrix by algorithm routine；Its secondary control signal generates a calculation matrix using Digital Micromirror Device, obtains a two-dimensional measured value；Last step is repeated again M times, obtain M two-dimensional measured values, and form measurement result；Then adaptive-filtering processing carried out to each two-dimensional measured value, and adaptive-filtering treated measurement result sums to pixel value according to the image resolution ratio of Digital Micromirror Device, obtain revised measurement result；Finally target image is reconstructed using revised measurement result and calculation matrix.By carrying out adaptive-filtering analysis processing to each two-dimensional measurement, the accuracy of measurement result is improved, improves the precision of reconstruction signal, or even reconstruct original signal in the measured value of original signal from that can not reconstruct originally.
Description
Technical field
The present invention relates to field of signal processing, and in particular to a kind of compression sensing method based on adaptive-filtering.
Background technique
Compressive sensing theory is relatively one of quantum jump that people obtain in field of signal processing in recent years.It is will have it is dilute
The signal for dredging characteristic does dimensionality reduction linear projection by calculation matrix, and leads to too small amount of Cephalometry and calculation matrix recovers
A kind of theory of original signal.It breaches the limitation of nyquist sampling theorem to a certain extent, adopts to reduce to data
Collect the requirement of hardware, provides new thinking for the acquisition of signal, transmission, storage and detection.
During compressed sensing technology is from theory into action, to accurately obtain measured signal, in addition to guaranteeing to measure
Except the reasonability of matrix and restructing algorithm, it should also reduce hardware system bring error to the greatest extent, make measurement data obtained
With notional result as close possible to.
In the actual treatment system of compressed sensing technology, if passing through CCD using binary sparse matrix as calculation matrix
Directly acquire measured value, in matrix the number of element " 0 " far fewer than element " 1 " number, so its effective information is mainly concentrated
In the higher position of pixel value, and its pixel value of the lower position of pixel value ought to all 0, but due to CCD and environment
The influence of effect itself, actual pixel value is usually 1~10 etc., to keep measurement result bigger than normal, greatly reduces signal weight
The precision of structure.
Summary of the invention
The present invention aiming at the problems existing in the prior art, provides a kind of base of precision for being greatly improved reconstruction signal
In the compression sensing method of adaptive-filtering.
In order to solve the above-mentioned technical problem, the technical scheme is that：A kind of compressed sensing based on adaptive-filtering
Method includes the following steps：
S1:Set total observation frequency M according to the image resolution ratio of target image to be detected, image resolution ratio be n ×
N, M, n are natural number, × represent product；
S2:According to total observation frequency M of setting, M calculation matrix is generated by algorithm routine；
S3:It controls signal and generates a calculation matrix using Digital Micromirror Device, light beam is transmitted by Digital Micromirror Device
Or object region is refracted to, observation signal is received by receiver, obtains a two-dimensional measured value；
S4:Step S3 is repeated M times, obtains M two-dimensional measured values, and form measurement result；
S5:Adaptive-filtering processing carried out to each two-dimensional measured value that step S4 is obtained, and by adaptive-filtering
Measurement result that treated sums to pixel value according to the image resolution ratio of Digital Micromirror Device, obtains revised measurement
As a result；
S6:Target image is reconstructed using revised measurement result and calculation matrix.
Further, the image resolution ratio of the target image to be detected is determined by Digital Micromirror Device.
Further, in the step S1, total observation frequency M<N, N=n × n.
Further, the range of total observation frequency M is 0.1N~0.6N.
Further, in the step S2, the image resolution ratio of each calculation matrix is also determined by Digital Micromirror Device,
And the image resolution ratio of calculation matrix is consistent with the image resolution ratio of target image.
Further, it in the step S3, obtains a two-dimensional measured value and planar array detector is utilized to receive from target
The 2D signal of image-region reflection.
Further, in the step S5, adaptive-filtering processing is specially：For each two-dimensional measured value, root
According to its pixel value distribution histogram, the threshold value T of valid pixel is calculated, all pixels value less than threshold value T is set 0.
Further, specifically, selecting a numberical range, calculating threshold value T meets the threshold value T for calculating valid pixel
Meet in numberical range 0 to the sum of the number of pixels of pixel value between T.
Further, in the step S6, the formula for reconstructing target image is：
O=Φ I
Wherein：I∈RN, it is that N × 1 ties up original signal, i.e., by target image by column recombination acquired results；Φ∈RM×N, it is M
× N-dimensional calculation matrix, O ∈ RMIt is the measurement result that M × 1 is tieed up.
Compression sensing method provided by the invention based on adaptive-filtering carries out the two-dimensional measurement obtained each time
Adaptive-filtering analysis processing, can adaptively obtain filtering threshold, threshold value pixel value below is filtered, the threshold is utilized
Pixel value more than value carries out target image reconstruct, improves the accuracy of measurement result, so that the precision of reconstruction signal is improved,
Or even original signal is reconstructed in the measured value of original signal from that can not reconstruct originally.The present invention passes through at adaptive-filtering
Reason, the improvement that the effect of target image reconstruct may be significantly.
Detailed description of the invention
Fig. 1 is the flow chart of the compression sensing method the present invention is based on adaptive-filtering；
Fig. 2 is present invention original image signal figure to be detected；
Fig. 3, Fig. 5 are in compressed sensing real system of the present invention respectively for different calculation matrix and without adaptive-filtering
Handle recovered original signal figure；
Fig. 4, Fig. 6 are in compressed sensing real system of the present invention respectively for different calculation matrix and by adaptive-filtering
Handle recovered original signal figure.
Specific embodiment
The present invention is described in detail with reference to the accompanying drawing：
As shown in Figure 1, including the following steps the present invention provides a kind of compression sensing method based on adaptive-filtering：
S1:Total observation frequency M, image resolution ratio N=n are set according to the image resolution ratio of target image to be detected
× n, M, n are natural number, × represent product；Specifically, the image resolution ratio of target image is by Digital Micromirror Device (Digital
Micromirror Device, DMD) it determines, the resolution ratio of actual compression sensory perceptual system, total observation frequency M can be regarded as<N, N=n
× n can be selected, it is furthermore preferred that the value range of M is 0.1N~0.6N such as efficiency and accuracy rate according to actual needs
Deng.
S2:According to total observation frequency M of setting, M calculation matrix is generated by algorithm routine；The figure of each calculation matrix
As resolution ratio is also determined by Digital Micromirror Device, and the image resolution ratio of calculation matrix and the image resolution ratio of target image one
It causes, is n × n, i.e., M calculation matrix forms matrix Φ, the Φ ∈ R that a dimension is M × N hereinM×N, N=n × n.
S3:It controls signal and generates a calculation matrix using Digital Micromirror Device, light beam is transmitted by Digital Micromirror Device
Or object region is refracted to, observation signal is received by receiver, obtains a two-dimensional measured value, i.e., is visited using face battle array
It surveys device and receives the 2D signal reflected from object region.Planar array detector can be CCD camera or CMOS camera, for connecing
The signal that receipts are reflected from object region, i.e. image resolution ratio are the 2D signal of n × n.
S4:Step S3 is repeated M time, obtains a two-dimensional measured values of M, and form measurement result, dimension can regard as M ×
1, wherein the dimension of each measured value is n × n.
S5:Adaptive-filtering processing carried out to each two-dimensional measured value that step S4 is obtained, and by adaptive-filtering
Measurement result that treated sums to pixel value according to the image resolution ratio of Digital Micromirror Device, obtains revised measurement
As a result；Wherein, adaptive-filtering, which is handled, is specially：For each two-dimensional measured value, according to its pixel value distribution histogram,
All pixels value less than threshold value T is set 0 by the threshold value T for calculating valid pixel.The threshold value for calculating valid pixel is specific
To select a numberical range, calculating the sum of the quantity of pixel value of threshold value T satisfaction between 0~T in the numberical range.
The detailed process of adaptive-filtering processing is described below with specific example：
Each measured value is got by planar array detector reception, two dimension letter corresponding for each measured value
Number, it is in the nature the grayscale image of the object region of measured Matrix cover.
Therefore, the measurement result O ∈ R tieed up for M × 1M, O=[O1,O2,O3……OM], optional wherein certain one-shot measurement knot
Fruit Oi∈Rn×n(wherein [1, M] i ∈) obtains its pixel value distribution histogram (the distribution number of i.e. each pixel value), horizontal seat
It is designated as pixel value size, ordinate is the corresponding pixel number of each pixel value size.
Based on pixel value distribution histogram, the sum of the corresponding pixel number of each pixel value size sum is found out, i.e.,
The sum of all ordinates in histogram, and the resolution ratio of measurement result, i.e. sum=N each time, N=n × n, due to selected survey
Moment matrix is binary sparse matrix, and above-mentioned numberical range is typically chosen 0.60~0.97sum, more preferably 0.75~
0.97sum, certain numberical range can also be adaptively adjusted according to the actual situation, in the protection scope of the application
It is interior, filtering threshold T is then found out in pixel value distribution histogram, meets threshold value T from 0 to the corresponding pixel number of T pixel value
Mesh summation meets the numberical range.
For the corresponding 2D signal of each measured value in measurement result O, it will wherein be less than the pixel value of filtering threshold T
Corresponding point sets 0, to obtain adaptive-filtering treated measurement result O ', by adaptive-filtering treated measurement result
O ' sums to pixel value according to the image resolution ratio of Digital Micromirror Device, obtains revised measurement result O ".
S6:Target image is reconstructed using revised measurement result O " and calculation matrix.
It is described reconstruct target image formula be：
O=Φ I
Wherein：I∈RN, it is that N × 1 ties up original signal, N=n × n, i.e., by target image by column recombination acquired results；Φ∈
RM×N, it is M × N-dimensional calculation matrix, O ∈ RMIt is the measurement result that M × 1 is tieed up.Specifically, in step s 6, O herein is represented most
Whole measurement result O ", due to Φ, O " it is known that so as to calculate the matrix I of target image, to reconstruct target figure
Picture.
For specific manifestation superiority of the invention, this example will be drawn using binary system random measurement matrix with based on augmentation
The restructing algorithm that the full variation of Ge Langfa and alternating direction method minimizes algorithm carries out description of test.Point of experimental system DMD
Resolution is 100 × 100, sample rate 0.3, i.e. sampling number M=0.3N, is 3000 times, and light source is laser, triggering in experiment
DMD generates binary measurement matrix and exposes to target area, and is believed by the two dimension that CCD camera receives object region reflection
Number, form measurement result.
Fig. 2 is original image signal to be processed；Fig. 3, Fig. 5 be in compressed sensing real system without adaptive-filtering at
Recovered original signal is managed, Fig. 4, Fig. 6 are that recovered original signal is handled by adaptive-filtering；Fig. 3 with
Fig. 5 or Fig. 4 is that calculation matrix is different from the difference of Fig. 6.
For two dimensional image signal, we measure reconstruct with reconstruction signal and the Y-PSNR (PSNR) of original signal
Effect.It is not difficult to find out that handled for echo signal by adaptive-filtering analysis, can eliminate in measurement result because environment with
The Y-PSNR of reconstruction signal is promoted to 20.49dB from 15.52dB, even by the problem bigger than normal of measured value caused by hardware
It cannot be reconstructed from script and reconstruct original signal in the measurement result of original signal.Comparison diagram 3 and Fig. 4 and Fig. 5 and Fig. 6,
This it appears that being greatly improved by adaptive-filtering analysis processing to reconstructed image quality.
Reconstruct can be obviously improved from the foregoing, it will be observed that analyzing with adaptive-filtering and carrying out compressed sensing processing in real system
The precision of signal.Demonstrate practicability and reliability of the method for optimization compressed sensing reconstruction signal.
In conclusion the compression sensing method provided by the invention based on adaptive-filtering, to the two dimension obtained each time
Measured value carries out adaptive-filtering analysis processing, can adaptively obtain filtering threshold, threshold value pixel value below is set 0
It filters, carries out target image reconstruct using the pixel value more than threshold value, improve the accuracy of measurement result, to improve weight
The precision of structure signal, or even original signal is reconstructed in the measured value of original signal from that can not reconstruct originally.The present invention passes through
Adaptive-filtering processing, the improvement that the effect of target image reconstruct may be significantly.
Although embodiments of the present invention are illustrated in specification, these embodiments are intended only as prompting,
It should not limit protection scope of the present invention.It is equal that various omission, substitution, and alteration are carried out without departing from the spirit and scope of the present invention
It should be included within the scope of the present invention.
Claims (9)
1. a kind of compression sensing method based on adaptive-filtering, which is characterized in that include the following steps：
S1:Total observation frequency M is set according to the image resolution ratio of target image to be detected, image resolution ratio is n × n, M, n
For natural number, × represent product；
S2:According to total observation frequency M of setting, M calculation matrix is generated by algorithm routine；
S3:It controls signal and generates a calculation matrix using Digital Micromirror Device, light beam is transmitted or rolled over by Digital Micromirror Device
It is incident upon object region, observation signal is received by receiver, obtains a two-dimensional measured value；
S4:Step S3 is repeated M times, obtains M two-dimensional measured values, and form measurement result；
S5:Adaptive-filtering processing is carried out to each two-dimensional measured value that step S4 is obtained, and adaptive-filtering is handled
Measurement result afterwards sums to pixel value according to the image resolution ratio of Digital Micromirror Device, obtains revised measurement knot
Fruit；
S6:Target image is reconstructed using revised measurement result and calculation matrix.
2. the compression sensing method according to claim 1 based on adaptive-filtering, which is characterized in that the mesh to be detected
The image resolution ratio of logo image is determined by Digital Micromirror Device.
3. the compression sensing method according to claim 1 based on adaptive-filtering, which is characterized in that the step S1
In, total observation frequency M<N, N=n × n.
4. the compression sensing method according to claim 3 based on adaptive-filtering, which is characterized in that total observation time
The range of number M is 0.1N~0.6N.
5. the compression sensing method according to claim 1 based on adaptive-filtering, which is characterized in that the step S2
In, the image resolution ratio of each calculation matrix is also determined by Digital Micromirror Device, and the image resolution ratio and target of calculation matrix
The image resolution ratio of image is consistent.
6. the compression sensing method according to claim 1 based on adaptive-filtering, which is characterized in that the step S3
In, it obtains a two-dimensional measured value and planar array detector is utilized to receive the 2D signal reflected from object region.
7. the compression sensing method according to claim 1 based on adaptive-filtering, which is characterized in that the step S5
In, adaptive-filtering processing is specially：For each two-dimensional measured value, according to its pixel value distribution histogram, calculating has
All pixels value less than threshold value T is set 0 by the threshold value T for imitating pixel.
8. the compression sensing method according to claim 7 based on adaptive-filtering, which is characterized in that described to calculate effectively
For the threshold value T of pixel specifically, selecting a numberical range, calculating threshold value T meets the number of pixels 0 to the pixel value between T
The sum of meet the numberical range.
9. the compression sensing method according to claim 1 based on adaptive-filtering, which is characterized in that the step S6
In, the formula for reconstructing target image is：
O=Φ I
Wherein：I∈RN, it is that N × 1 ties up original signal, i.e., by target image by column recombination acquired results；Φ∈RM×N, it is M × N-dimensional
Calculation matrix, O ∈ RMIt is the measurement result that M × 1 is tieed up.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810925896.2A CN108921807A (en) | 2018-08-15 | 2018-08-15 | A kind of compression sensing method based on adaptive-filtering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810925896.2A CN108921807A (en) | 2018-08-15 | 2018-08-15 | A kind of compression sensing method based on adaptive-filtering |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108921807A true CN108921807A (en) | 2018-11-30 |
Family
ID=64404723
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810925896.2A Pending CN108921807A (en) | 2018-08-15 | 2018-08-15 | A kind of compression sensing method based on adaptive-filtering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108921807A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369638A (en) * | 2020-05-27 | 2020-07-03 | 中国人民解放军国防科技大学 | Laser reflection tomography undersampled reconstruction method, storage medium and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120038798A1 (en) * | 2010-08-11 | 2012-02-16 | Woods Gary L | Techniques for Removing Noise in a Compressive Imaging Device |
| CN103475875A (en) * | 2013-06-27 | 2013-12-25 | 上海大学 | Image adaptive measuring method based on compressed sensing |
| US8783874B1 (en) * | 2012-01-18 | 2014-07-22 | Nusensors, Inc. | Compressive optical display and imager |
-
2018
- 2018-08-15 CN CN201810925896.2A patent/CN108921807A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120038798A1 (en) * | 2010-08-11 | 2012-02-16 | Woods Gary L | Techniques for Removing Noise in a Compressive Imaging Device |
| US8783874B1 (en) * | 2012-01-18 | 2014-07-22 | Nusensors, Inc. | Compressive optical display and imager |
| CN103475875A (en) * | 2013-06-27 | 2013-12-25 | 上海大学 | Image adaptive measuring method based on compressed sensing |
Non-Patent Citations (1)
| Title |
|---|
| 邓超等: "《数字图像处理与模式识别研究》", 30 June 2018, 地质出版社 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111369638A (en) * | 2020-05-27 | 2020-07-03 | 中国人民解放军国防科技大学 | Laser reflection tomography undersampled reconstruction method, storage medium and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101991343B1 (en) | Method of estimating optical flow on the basis of an asynchronous light sensor | |
| Kopsiaftis et al. | Vehicle detection and traffic density monitoring from very high resolution satellite video data | |
| Vicari et al. | New estimates of leaf angle distribution from terrestrial LiDAR: Comparison with measured and modelled estimates from nine broadleaf tree species | |
| US8125549B2 (en) | Methods and apparatus to capture compressed images | |
| Lavreniuk et al. | Regional retrospective high resolution land cover for Ukraine: Methodology and results | |
| EP2511680A2 (en) | Optimized orthonormal system and method for reducing dimensionality of hyperspectral Images | |
| JP2008292449A (en) | Automatic target identifying system for detecting and classifying object in water | |
| US20130011051A1 (en) | Coded aperture imaging | |
| Traganos et al. | Cubesat-derived detection of seagrasses using planet imagery following unmixing-based denoising: Is small the next big? | |
| WO2011145040A1 (en) | Edge-preserving noise filtering | |
| CN108288256A (en) | A kind of multispectral mosaic image restored method | |
| CN113870132A (en) | Noise elimination method and system in ghost imaging sampling calculation process and related components | |
| CN109087267A (en) | A kind of compressed sensing based object detection method | |
| Yu et al. | Cloud removal in optical remote sensing imagery using multiscale distortion-aware networks | |
| CN108921807A (en) | A kind of compression sensing method based on adaptive-filtering | |
| CN116739958B (en) | Dual-spectrum polarization super-resolution fusion detection method and system | |
| US20220414844A1 (en) | Motion artifact correction for phase-contrast and dark-field imaging | |
| CN109102551A (en) | A kind of time compressed sensing reconstructing method based on ray tracing | |
| CN117422619A (en) | Training method of image reconstruction model, image reconstruction method, device and equipment | |
| CN109596069A (en) | Object phase restoration methods based on distortion grating and code aperture | |
| CN104992456B (en) | Multiple dimensioned matrix coder method | |
| CN109523466A (en) | A kind of compressed sensing image reconstructing method based on zero padding operation | |
| US20240135694A1 (en) | Change detection device and related methods | |
| Gupta et al. | Object recognition based on template matching and correlation method in hyperspectral images | |
| Zhang et al. | An adaptive infrared image preprocessing method based on background complexity descriptors |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181130 |