CN109087246A - A kind of ultrasound image reconstructing method based on sparse reconstruct - Google Patents
A kind of ultrasound image reconstructing method based on sparse reconstruct Download PDFInfo
- Publication number
- CN109087246A CN109087246A CN201810927140.1A CN201810927140A CN109087246A CN 109087246 A CN109087246 A CN 109087246A CN 201810927140 A CN201810927140 A CN 201810927140A CN 109087246 A CN109087246 A CN 109087246A
- Authority
- CN
- China
- Prior art keywords
- dictionary
- signal
- iteration
- sample
- domain signal
- 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
- 238000002604 ultrasonography Methods 0.000 title claims abstract description 37
- 238000000034 method Methods 0.000 title claims abstract description 29
- 239000000523 sample Substances 0.000 claims abstract description 44
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 37
- 238000010276 construction Methods 0.000 claims abstract description 18
- 238000013519 translation Methods 0.000 claims description 9
- 238000009826 distribution Methods 0.000 abstract description 4
- 238000012545 processing Methods 0.000 abstract description 3
- 239000006101 laboratory sample Substances 0.000 description 9
- 238000010586 diagram Methods 0.000 description 8
- 230000007547 defect Effects 0.000 description 6
- 238000001514 detection method Methods 0.000 description 5
- WFKWXMTUELFFGS-UHFFFAOYSA-N tungsten Chemical compound [W] WFKWXMTUELFFGS-UHFFFAOYSA-N 0.000 description 5
- 239000010937 tungsten Substances 0.000 description 5
- 229910052721 tungsten Inorganic materials 0.000 description 5
- 241000208340 Araliaceae Species 0.000 description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 235000008434 ginseng Nutrition 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 229910018503 SF6 Inorganic materials 0.000 description 2
- 239000008367 deionised water Substances 0.000 description 2
- 229910021641 deionized water Inorganic materials 0.000 description 2
- 238000000151 deposition Methods 0.000 description 2
- 238000005530 etching Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- SFZCNBIFKDRMGX-UHFFFAOYSA-N sulfur hexafluoride Chemical compound FS(F)(F)(F)(F)F SFZCNBIFKDRMGX-UHFFFAOYSA-N 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 239000004411 aluminium Substances 0.000 description 1
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 1
- 229910052782 aluminium Inorganic materials 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004624 confocal microscopy Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000008021 deposition Effects 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000001755 magnetron sputter deposition Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005498 polishing Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 229960000909 sulfur hexafluoride Drugs 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
- 238000010408 sweeping Methods 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
- G06T3/4076—Super resolution, i.e. output image resolution higher than sensor resolution by iteratively correcting the provisional high resolution image using the original low-resolution image
-
- G06T5/73—
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
Abstract
The invention discloses a kind of ultrasound image reconstructing methods based on sparse reconstruct, it is related to technical field of image processing, this method comprises: obtaining the ultrasonic echo time-domain signal of sample to be tested using ultrasonic probe, complete dictionary is obtained by reference to echo-signal construction, sparse decomposition, which is carried out, using ultrasonic echo time-domain signal of the excessively complete dictionary to sample to be tested obtains sparse decomposition coefficients, according to sparse decomposition coefficients and excessively complete dictionary reconstructs to obtain ultrasonic echo time-domain signal, again combine the ultrasonic echo time-domain signal that each reconstruct obtains according to spatial position the reconstruct ultrasound image to form sample to be tested；The transverse and longitudinal resolution ratio of image can be improved, can quickly and effectively observe position, size and the distribution situation of the microdefect of sample interior.
Description
Technical field
The present invention relates to technical field of image processing, especially a kind of ultrasound image reconstructing method based on sparse reconstruct.
Background technique
High frequency ultrasound, which encounters different interfaces when component inside is propagated, to have different reflection signals (echo), utilizes this
Characteristic can scan sample to obtain the high frequency ultrasound scan image of sample, thus to micro- in sample by high frequency ultrasound
Defect is detected.
But when practical application, sample to be detected is usually more small, and the size of the microdefect in sample is more small, connects
The noise caused by the micro-structure and grain boundaries mutually mixes in the signal received, may cover returning for certain defects
Wave, to significantly limit the detection and identification of microdefect.Therefore it in order to improve the accuracy of detection and identification, needs to obtaining
The high frequency ultrasound scan image got carries out image procossing, has had more mature deblurring technology for natural image at present
And super resolution technology, but lack and be processed similarly technology for high frequency ultrasound scan image, and high frequency ultrasound scan image with from
Right image makes a big difference, and on the one hand the principle of its obscurity boundary illustrates that this is just needed to ultrasonic imaging there is no specific
Theory further conducts a research, and texture abundant is special not as natural image in another aspect high frequency ultrasound scan image
Sign, image is relatively simple, and edge variation is less, therefore also the processing method of natural image directly can not be applied to high frequency and surpassed
On sound scan image.
Summary of the invention
The present inventor regarding to the issue above and technical need, proposes a kind of ultrasound image reconstruct side based on sparse reconstruct
Method, this method are directed to high frequency ultrasound scan image, the decomposition to ultrasonic time-domain signal, B-scan and three dimensional signal may be implemented
Reconstruct, to realize that tiny flaw detection more quickly and efficiently provides method.
Technical scheme is as follows:
A kind of ultrasound image reconstructing method based on sparse reconstruct, this method comprises:
The ultrasonic echo time-domain signal of sample to be tested is obtained using ultrasonic probe；
Control echo signal is obtained, constructs the construction dictionary based on Gabor function, Control echo using Control echo signal
Signal is used to characterize the surface echo signal of ideal infinity plane；
Construction dictionary was extended to complete dictionary, includes m dictionary atom in excessively complete dictionary, m is positive integer；
Sparse decomposition is carried out using the ultrasonic echo time-domain signal of m dictionary atom pair sample to be tested in excessively complete dictionary
Sparse decomposition coefficients are obtained, according to sparse decomposition coefficients and excessively complete dictionary reconstructs to obtain ultrasonic echo time-domain signal；
By it is each reconstruct obtained ultrasonic echo time-domain signal and combine according to spatial position to form the reconstruct of sample to be tested and surpass
Acoustic image.
Its further technical solution is to construct the construction dictionary based on Gabor function, packet using Control echo signal
It includes:
Control echo signal is fitted based on Gabor function, obtains the dictionary atom in construction dictionary, Gabor letter
Number are as follows:
Wherein, g indicates that the dictionary atom in construction dictionary, sr are zooming parameters, and fr is frequency domain parameter, and ur is translation ginseng
Number, A indicate that dictionary atom normalizes the amplitude used, and t indicates the time, and zooming parameter, frequency domain parameter and translation parameters are based on ginseng
Echo-signal is examined to determine.
Its further technical solution is that construction dictionary was extended to complete dictionary, comprising:
The zooming parameter of the dictionary atom constructed in dictionary is extended to downwards 0.5sr: 0.05sr: sr, by frequency domain parameter to
Under be extended to 0.5fr: 0.05fr: fr, it is 1:1:M that translation parameters, which is extended to entire time domain space, and M is ultrasonic echo time domain letter
Number length, complete dictionary is obtained.
Its further technical solution is that the ultrasound of m dictionary atom pair sample to be tested in the excessively complete dictionary of utilization is returned
Wave time-domain signal carries out sparse decomposition and obtains sparse decomposition coefficients, comprising:
Initialize sparse decomposition coefficients x0=0, residual error r0=y-D*x0The supported collection of=y, solutionY is ultrasonic echo
Time-domain signal, D were complete dictionaries；
For kth time iteration, m dictionary atom in excessively complete dictionary is added in the supported collection of -1 iteration of kth and is obtained
To temporary support collection It indicated that the dictionary atom in complete dictionary, i were parameter and 1≤i≤m, utilized
Temporary support collection calculates r=| | y-D (Stemp)*pinv(D(Stemp))*y2The corresponding residual error of each dictionary atom is obtained, k is ginseng
The initial value of number and k are 1；
Temporary support is concentrated the smallest dictionary atom of residual error for making to be calculated be added in the supported collection of -1 iteration of kth
The supported collection of kth time iteration is obtained, calculates x using the supported collection of kth time iterationk=pinv (D (Sk)) * y update sparse decomposition system
Number；
Judge whether to reach stopping criterion for iteration, if reaching stopping criterion for iteration, iteration ends simultaneously obtain sparse decomposition
Coefficient；If not up to stopping criterion for iteration, k=k+1 is enabled, and is re-executed for kth time iteration, it will be in excessively complete dictionary
M dictionary atom is added in the supported collection of -1 iteration of kth and obtains temporary support collectionThe step of.
Its further technical solution is to combine each obtained ultrasonic echo time-domain signal that reconstructs according to spatial position
Form the reconstruct ultrasound image of sample to be tested, comprising:
Ultrasonic echo time-domain signal after sparse representation is drawn together according to spatial position, forms class B-scan figure
Two-dimensional reconstructed image；
The two-dimensional reconstructed image of class B-scan figure is combined to the three-dimensional reconstruct ultrasound image to form sample to be tested.
The method have the benefit that:
The application discloses one kind and is based on for the deficiency of the image processing method of existing high frequency ultrasound detection tiny flaw
The ultrasound image reconstructing method of sparse reconstruct, the application is different from natural image processing, to corresponding deblurring model and calculation
Method parameter is adjusted, and the application constructs the excessively complete dictionary of Gabor, which improves the sparsity of sparse representation, is reduced
Computation rate is carried out sparse heavy using ultrasound detection of the dictionary to sample to be tested ultrasonic echo time-domain signal obtained
Structure, and then reconstruct is realized to B-san image and 3-D image, the transverse and longitudinal resolution ratio of image can be improved, can quickly have
The position for observing microdefect, size and the distribution situation of effect.
Detailed description of the invention
Fig. 1 is the method flow diagram of the ultrasound image reconstructing method disclosed in the present application based on sparse reconstruct.
Fig. 2 (a) is the ultrasonic echo time-domain signal in experimental example 1, after original ultrasonic echo time-domain signal and reconstruct
Comparison.
Fig. 2 (b) is the supported collection schematic atomic diagram in the sparse representation result in experimental example 1.
Fig. 2 (c) is another supported collection schematic atomic diagram in the sparse representation result in experimental example 1.
Fig. 2 (d) is the schematic diagram of the sparse decomposition coefficients in experimental example 1.
Fig. 3 (a) is the C-scan image schematic diagram at the microdefect interface in experimental example 2.
Fig. 3 (b) is the signal for sweeping two dimensional image in experimental example 2 according to the class B that original ultrasonic echo time-domain signal is constituted
Figure.
Fig. 3 (c) is the result of the sparse decomposition coefficients absolute value in experimental example 2.
Fig. 3 (d) is the schematic diagram of the class B-scan two dimensional image reconstructed in experimental example 2.
Fig. 4 (a) is the schematic diagram of the sparse decomposition coefficients in experimental example 3.
Fig. 4 (b) is the three-dimensional reconstruct ultrasound image that reconstruct obtains in experimental example 3.
Fig. 4 (c) is the schematic diagram of the 3-D image scanned in experimental example 3.
Specific embodiment
The following further describes the specific embodiments of the present invention with reference to the drawings.
This application discloses a kind of ultrasound image reconstructing methods based on sparse reconstruct, referring to FIG. 1, this method includes such as
Lower step:
Step 1 obtains the ultrasonic echo time-domain signal of sample to be tested using ultrasonic probe.Specifically, configuration high-purity
Sample to be tested is totally submerged in deionized water by deionized water as couplant, the setting of the focal plane of ultrasonic probe to
The bottom surface of sample is simultaneously scanned, and saves the ultrasonic echo time-domain signal of whole positions in scanning process.
Step 2 obtains Control echo signal, constructs the construction dictionary based on Gabor function using Control echo signal,
Wherein, Control echo signal is used to characterize the surface echo signal of ideal infinity plane, is a kind of echo ideally
Time-domain signal, specific:
(1), it chooses much larger than the plane of ultrasonic probe as ideal infinity plane.
(2), the ideal infinity plane is detected using ultrasonic probe (usually 230MHz), it is ideal unlimited to obtain
The surface echo signal of big plane is as Control echo signal.
(3), Control echo signal is fitted based on Gabor function, obtains one group of parameter (sr,fr,ur), thus
To the dictionary atom in construction dictionary, Gabor function are as follows:
Wherein, g indicates the single dictionary atom in construction dictionary, srIt is zooming parameter, frIt is frequency domain parameter, urIt is translation
Parameter, A indicate that dictionary atom normalizes the amplitude used, and t indicates time, zooming parameter sr, frequency domain parameter frWith translation parameters ur
It is to be determined based on Control echo signal.
Construction dictionary was extended to complete dictionary, and included m dictionary atom in excessively complete dictionary, m is positive whole by step 3
Number, the length of each dictionary atom are n, and n is integer and n < m.Specifically, the zooming parameter s that will be fitted in step 2r
It is extended to 0.5s downwardsr: 0.05sr: sr, by frequency domain parameter frIt is extended to 0.5f downwardsr: 0.05fr: fr, by translation parameters urIt expands
It is 1:1:M to entire time domain space, M is the length of ultrasonic echo time-domain signal, and obtained Gabor dictionary was complete word
Allusion quotation D.
Step 4 is carried out using the ultrasonic echo time-domain signal of m dictionary atom pair sample to be tested in excessively complete dictionary D
Sparse decomposition obtains sparse decomposition coefficients, and according to sparse decomposition coefficients and excessively complete dictionary reconstructs to obtain ultrasonic echo time domain letter
Number.It is specific:
1, ultrasonic echo time-domain signal and excessively complete dictionary D are inputted, sets stopping criterion for iteration, stopping criterion for iteration according to
Actual needs setting.
2, iterative parameter, including initialization sparse decomposition coefficients x are initialized0=0, residual error r0=y-D*x0The branch of=y, solution
Support collectionWherein, y is the ultrasonic echo time-domain signal actually got, and D was complete dictionary,Indicate empty set.
3, it iteratively solves, specific as follows:
(1), it scans: for kth time iteration, -1 iteration of kth is added in m dictionary atom in excessively complete dictionary D
Supported collection Sk-1In obtain temporary support collection Indicated that the dictionary atom in complete dictionary D, i were parameter
And it is 1 that 1≤i≤m, k, which are the initial value of parameter and k,.Utilize temporary support collection StempCalculate r=| | y-D (Stemp)*pinv(D
(Stemp))*y||2Obtain the corresponding residual error of each dictionary atom, wherein pinv () indicates to seek the pseudo inverse matrix of matrix.
(2), supported collection is updated: by temporary support collection StempIn the smallest dictionary atom of the residual error being calculated is added
The supported collection S of -1 iteration of kthk-1In, it updates and obtains the supported collection S of kth time iterationk。
(3), more new explanation: in the supported collection S of kth time iterationkUnder, calculate xk=pinv (D (Sk)) * y, to update sparse
Decomposition coefficient.
(4), judge whether to reach preconfigured stopping criterion for iteration.If reaching stopping criterion for iteration, iteration ends
And obtain final sparse decomposition coefficients；If not up to stopping criterion for iteration, enable k=k+1, and re-execute above-mentioned (1) after
It is continuous to be iterated solution.
4, output is as a result, after obtaining final sparse decomposition coefficients, utilizes the final sparse decomposition coefficients and excessively complete
Standby dictionary carries out the reconstruct of one-dimensional ultrasonic echo time-domain signal, and the ultrasonic echo time-domain signal reconstructed is yK=D*xK,
In, yKIndicate that the ultrasonic echo time-domain signal that reconstruct obtains, xK indicate the final sparse decomposition coefficients that iteration obtains.
Each obtained ultrasonic echo time-domain signal that reconstructs is combined to form sample to be tested by step 5 according to spatial position
Reconstruct ultrasound image.
The application has good quality reconstruction to two dimensional image and 3-D image, above-mentioned utilizing for two dimensional image
Each step reconstructs after obtaining the corresponding ultrasonic echo time-domain signal in each position, by each ultrasonic echo time domain for reconstructing and obtaining
Signal is drawn together according to spatial position, forms the two-dimensional reconstruct ultrasound image of class B-scan figure.For 3-D image,
After equally forming the two-dimensional reconstruct ultrasound image of class B-scan figure with two dimensional image, then by the reconstruct image of each class B-scan figure
The three-dimensional reconstruct ultrasound image of sample to be tested is formed as being combined according to spatial position.
In the actual application of the application, microdefect is generally included inside sample to be tested, then is obtained by the above method
After taking the reconstruct ultrasound image of sample to be tested, the distribution situation of the microdefect inside sample to be tested can be intuitively observed, and
Compared to original ultrasonic echo time-domain signal, the ultrasonic echo time-domain signal reconstructed eliminates the dry of noise well
It disturbs, the prominent reflection echo signal for reflecting each interface, to effectively improve the microdefect position identification under profile scanning
Accuracy, and improve recognition efficiency.Applicant also demonstrates the feasibility of the above method and accurate by following actual experiment
Property, specifically, applicant is first prepared for a series of laboratory sample that inside have microdefect, preparation process is as follows:
(1), standby piece: use the tungsten disk that diameter is 4 cun as material is prepared, two-sided mechanical polishing, place are carried out to leaf
It manages into a thickness of 500 ± 100 μm, sample of the Ra (surface roughness) lower than 0.1 μm.
(2), deposition mask: depositing a layer thickness on tungsten surface using magnetron sputtering method is 3 μm of aluminium films as covering
Film.
(3), mask rule: using the figure for being dry-etched in cover up rule on exposure mask.
(4), ICP is etched: using SF6(sulfur hexafluoride) performs etching tungsten, and etching depth is 50 μm.
(5), exposure mask is removed, tungsten disk is cut into the laboratory sample of 10mm*10mm using cutting machine.
A series of laboratory sample can be prepared using above-mentioned steps (1)-(5), it is micro- inside these laboratory samples
The size of defect can be differed at 10 μm -100 μm.After laboratory sample is prepared, scanning electron microscope can be passed through
(SEM) to laboratory sample be scanned obtain defect sample pictures, to have intuitive understanding to internal microdefect, and facilitate it is subsequent with
Reconstruct ultrasound image is compared.
Meanwhile laboratory sample is reconstructed using the above method, the effect of the application can further pass through Comsol
Multiphysics carries out analog simulation and description of test, please refers to following three experimental example:
Experimental example 1: the experimental example for verify the application have to one-dimensional ultrasonic echo time-domain signal it is good reconstitution
Energy.This test is emulated under Comsol Multiphysics simulation software, emulates used Gaussian pulse simulation ultrasound letter
Number, the frequency of signal is 230MHz, and the laboratory sample of emulation is that the TSV of upside-down mounting welding core is encapsulated.It is returned to resulting ultrasound is emulated
Wave signal carries out sparse representation, and the residual values of sparse representation are set as 0.5.Original ultrasonic echo time domain is shown in Fig. 2 (a)
The comparison for the ultrasonic echo time-domain signal that signal and reconstruct obtain, wherein dotted portion is original ultrasonic echo time-domain signal,
Bold portion is the ultrasonic echo time-domain signal that reconstruct obtains, and can intuitively find out that the Approximation effect of reconstruction signal is very good,
It not only clearly indicates 2 echoes in original signal, has also filtered noise well.What Fig. 2 (b) and Fig. 2 (c) was respectively indicated
It is two supported collection atoms in sparse representation result, each atom pair answers a signal echo.And its corresponding sparse decomposition
Shown in coefficient such as Fig. 2 (d), the absolute value of sparse decomposition coefficients shows that more greatly the reflection signal of the point is stronger, and sparse decomposition system
Several is positive and negative, indicates that corresponding echo is opposite with atom.
Experimental example 2: the experimental example has good quality reconstruction to B-scan cross-section image for verifying the present invention.This reality
It tests and the laboratory sample of 50 μm of width is scanned using SAM30E equipment, the frequency probe of scanning is 110MHz, scanning figure
Picture resolution ratio is 500 × 476 pixels, and scanning step is 2 μm.The C-scan image at obtained microdefect interface such as Fig. 3 (a) institute
Show, the ultrasonic echo time-domain signal under 500 ultrasound probe positions of the 300th column in figure is drawn together, and extracts wherein
Ultrasonic echo time-domain signal of the echo time between 180ns-300ns is constituted using original ultrasonic echo time-domain signal
The class B of microdefect is swept shown in two dimensional image such as Fig. 3 (b), can clearly find out in Fig. 3 (b) each interface position and
Time situation.Sparse representation is carried out using ultrasonic echo time-domain signal of the method provided by the present application to each point, wherein each
Shown in sparse decomposition coefficients absolute value result such as Fig. 3 (c) under probe positions, it reflects the ejected wave reflection of corresponding position
Situation, resolving system numerical value are 0, then it represents that do not have reflection echo, that is, acoustic impedance interface is not present, can be considered as no defect.
Fig. 3 (d) be reconstructed according to sparse decomposition coefficients and the dictionary used come class B-scan figure two-dimensional reconstruct ultrasound figure
Picture, compared to Fig. 3 (b), reconstruction signal eliminates the interference of noise, the prominent reflection echo letter for reflecting each interface well
Number.Sparse reconstruct is carried out to B-scan image it can effectively improve microdefect position under profile scanning to identify accuracy, with
And improve recognition efficiency.
Experimental example 3: the experimental example has preferable reconstruction property to 3-D image for verifying the present invention.This experiment uses
110MHz, scanning resolution are 500 × 476 pixels, and scanning step is 2 μm, and each scanning element acquisition 2000ns duration time domain is returned
500 pixels of each column have been corresponded to a vertical section in experimental example 2, have been carried out sparse representation by wave, and finally combination is all
476 sections sparse representation result.Fig. 4 (a) is the three-dimensional figure of sparse decomposition coefficients, therefrom can intuitively be observed interior
The three-dimensional position distribution situation of portion's microdefect.Further controlled in such a way that control supported collection size is with setting threshold residual value
Iterative process in sparse representation carries out sparse representation to the ultrasonic echo time-domain signal in whole scanning ranges, in Fig. 4 (a)
Corresponding three-dimensional coordinate z-axis is echo time-domain position, is calculated with spread speed 4620m/s of the sound in tungsten, by time domain
Position is converted into corresponding depth location, and as shown in Fig. 4 (b), the section width of 5 square grooves in figure is respectively 40 μm, 46 μ
M, 44 μm and 46 μm.The original scanned by sparse representation microdefect depth information obtained and with laser confocal microscopy
The sample bottom surface defect map 4 (c) of beginning is compared, and the end face width of 5 square grooves is respectively 46.1 μm, 45.5 μm, 46.8 μ in figure
M, 44.9 μm and 46.1 μm.By comparison it can be seen that the result of the slot in left side has slightly deviation, other 4 groove profiles are micro- to be lacked
Sunken result all extremely meets.
Above-described is only the preferred embodiment of the application, and present invention is not limited to the above embodiments.It is appreciated that this
The other improvements and change that field technical staff directly exports or associates without departing from the spirit and concept in the present invention
Change, is considered as being included within protection scope of the present invention.
Claims (5)
1. a kind of ultrasound image reconstructing method based on sparse reconstruct, which is characterized in that the described method includes:
The ultrasonic echo time-domain signal of sample to be tested is obtained using ultrasonic probe；
Control echo signal is obtained, constructs the construction dictionary based on Gabor function, the reference using the Control echo signal
Echo-signal is used to characterize the surface echo signal of ideal infinity plane；
The construction dictionary was extended to complete dictionary, includes m dictionary atom in the excessively complete dictionary, m is positive integer；
It is carried out using the ultrasonic echo time-domain signal of sample to be tested described in m dictionary atom pair in the excessively complete dictionary sparse
Decomposition obtains sparse decomposition coefficients, reconstructs to obtain ultrasonic echo time domain according to the sparse decomposition coefficients and the excessively complete dictionary
Signal；
By it is each reconstruct obtained ultrasonic echo time-domain signal and combine according to spatial position to form the reconstruct of the sample to be tested and surpass
Acoustic image.
2. the method according to claim 1, wherein described be based on using Control echo signal building
The construction dictionary of Gabor function, comprising:
The Control echo signal is fitted based on Gabor function, obtains the dictionary atom in the construction dictionary,
Gabor function are as follows:
Wherein, g indicates the dictionary atom in the construction dictionary, srIt is zooming parameter, frIt is frequency domain parameter, urIt is translation parameters,
A indicates that the dictionary atom normalizes the amplitude used, and t indicates time, the zooming parameter, frequency domain parameter and translation parameters base
It is determined in the Control echo signal.
3. according to the method described in claim 2, it is characterized in that, described be extended to complete dictionary for the construction dictionary,
Include:
The zooming parameter of dictionary atom in the construction dictionary is extended to downwards 0.5sr: 0.05sr: sr, the frequency domain is joined
Number is extended to downwards 0.5fr: 0.05fr: fr, it is 1:1:M that the translation parameters, which is extended to entire time domain space, and M is described super
The length of sound echo time-domain signal obtains the excessively complete dictionary.
4. the method according to claim 1, wherein described former using m dictionary in the excessively complete dictionary
Son carries out sparse decomposition to the ultrasonic echo time-domain signal of the sample to be tested and obtains sparse decomposition coefficients, comprising:
Initialize sparse decomposition coefficients x0=0, residual error r0=y-D*x0The supported collection of=y, solutionWhen y is the ultrasonic echo
Domain signal, D are the excessively complete dictionaries；
For kth time iteration, m dictionary atom in the excessively complete dictionary is added in the supported collection of -1 iteration of kth and is obtained
To temporary support collection Indicating the dictionary atom in the excessively complete dictionary, i is parameter and 1≤i≤m,
R=is calculated using the temporary support collection | | y-D (Stemp)*pinv(D(Stemp))*y||2It is corresponding to obtain each dictionary atom
Residual error, k are that the initial value of parameter and k are 1；
The temporary support is concentrated the smallest dictionary atom of residual error for making to be calculated be added in the supported collection of -1 iteration of kth
The supported collection of kth time iteration is obtained, calculates x using the supported collection of the kth time iterationk=pinv (D (Sk)) * y updates sparse point
Solve coefficient；
Judge whether to reach stopping criterion for iteration, if reaching the stopping criterion for iteration, iteration ends simultaneously obtain sparse decomposition
Coefficient；If the not up to described stopping criterion for iteration, enables k=k+1, and re-executes the iteration secondary for kth, by the mistake
M dictionary atom in complete dictionary is added in the supported collection of -1 iteration of kth and obtains temporary support collection's
Step.
5. the method according to claim 1, wherein described by each ultrasonic echo time-domain signal for reconstructing and obtaining
The reconstruct ultrasound image to form the sample to be tested is combined according to spatial position, comprising:
Ultrasonic echo time-domain signal after sparse representation is drawn together according to spatial position, forms the two dimension of class B-scan figure
Reconstructed image；
The two-dimensional reconstructed image of the class B-scan figure is combined to the three-dimensional reconstruct ultrasound to form the sample to be tested
Image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810927140.1A CN109087246B (en) | 2018-08-15 | 2018-08-15 | Ultrasonic image reconstruction method based on sparse reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810927140.1A CN109087246B (en) | 2018-08-15 | 2018-08-15 | Ultrasonic image reconstruction method based on sparse reconstruction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109087246A true CN109087246A (en) | 2018-12-25 |
| CN109087246B CN109087246B (en) | 2023-03-14 |
Family
ID=64793453
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810927140.1A Active CN109087246B (en) | 2018-08-15 | 2018-08-15 | Ultrasonic image reconstruction method based on sparse reconstruction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109087246B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112183297A (en) * | 2020-09-23 | 2021-01-05 | 中国民航大学 | Ultrasonic phased array signal sparse feature extraction method |
| CN113254214A (en) * | 2021-06-08 | 2021-08-13 | 西安科技大学 | OpenMP-based flip chip acoustic-time-frequency-domain and time-domain imaging method |
| CN113419227A (en) * | 2021-05-07 | 2021-09-21 | 北京林业大学 | Dielectric characteristic analysis system and method for radial layered structure of tree branches |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103473451A (en) * | 2013-09-05 | 2013-12-25 | 江苏大学 | Ultrasonic professional dictionary construction and use method |
| CN103679762A (en) * | 2013-12-13 | 2014-03-26 | 江苏大学 | Ultrasonic signal reestablishing method based on sparse data |
| WO2018099321A1 (en) * | 2016-11-30 | 2018-06-07 | 华南理工大学 | Generalized tree sparse-based weighted nuclear norm magnetic resonance imaging reconstruction method |
| CN108230280A (en) * | 2018-04-11 | 2018-06-29 | 哈尔滨工业大学 | Image speckle noise minimizing technology based on tensor model and compressive sensing theory |
-
2018
- 2018-08-15 CN CN201810927140.1A patent/CN109087246B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103473451A (en) * | 2013-09-05 | 2013-12-25 | 江苏大学 | Ultrasonic professional dictionary construction and use method |
| CN103679762A (en) * | 2013-12-13 | 2014-03-26 | 江苏大学 | Ultrasonic signal reestablishing method based on sparse data |
| WO2018099321A1 (en) * | 2016-11-30 | 2018-06-07 | 华南理工大学 | Generalized tree sparse-based weighted nuclear norm magnetic resonance imaging reconstruction method |
| CN108230280A (en) * | 2018-04-11 | 2018-06-29 | 哈尔滨工业大学 | Image speckle noise minimizing technology based on tensor model and compressive sensing theory |
Non-Patent Citations (3)
| Title |
|---|
| 吴彪: "超声无损检测缺陷定位与稀疏信号重构方法", 《中国博士学位论文全文数据库工程科技Ⅱ辑》 * |
| 曹宇明等: "图像的多成分混合字典压缩感知表示及重构", 《计算机工程与应用》 * |
| 齐爱玲等: "遗传优化的稀疏分解算法在超声缺陷提取中的应用", 《无损检测》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112183297A (en) * | 2020-09-23 | 2021-01-05 | 中国民航大学 | Ultrasonic phased array signal sparse feature extraction method |
| CN112183297B (en) * | 2020-09-23 | 2022-08-16 | 中国民航大学 | Ultrasonic phased array signal sparse feature extraction method |
| CN113419227A (en) * | 2021-05-07 | 2021-09-21 | 北京林业大学 | Dielectric characteristic analysis system and method for radial layered structure of tree branches |
| CN113254214A (en) * | 2021-06-08 | 2021-08-13 | 西安科技大学 | OpenMP-based flip chip acoustic-time-frequency-domain and time-domain imaging method |
| CN113254214B (en) * | 2021-06-08 | 2023-03-10 | 西安科技大学 | OpenMP-based flip chip acoustic-time-frequency-domain and time-domain imaging method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109087246B (en) | 2023-03-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109087246A (en) | A kind of ultrasound image reconstructing method based on sparse reconstruct | |
| Lorintiu et al. | Compressed sensing reconstruction of 3D ultrasound data using dictionary learning and line-wise subsampling | |
| CN106485764B (en) | The quick exact reconstruction methods of MRI image | |
| Gan | Acoustical imaging: techniques and applications for engineers | |
| CN106841402A (en) | A kind of phased array supersonic signal reconstruction optimization method based on greedy algorithm | |
| Mei et al. | Visual geometry group-UNet: deep learning ultrasonic image reconstruction for curved parts | |
| Tuysuzoglu et al. | Sparsity driven ultrasound imaging | |
| CN106814141A (en) | A kind of phased array supersonic compression method based on orthogonal matching pursuit | |
| CN113449683B (en) | High-frequency ultrasonic sparse denoising method and system based on K-SVD training local dictionary | |
| Jin et al. | Sparse deconvolution method for ultrasound images based on automatic estimation of reference signals | |
| Scharrer et al. | Ultrasonic imaging of complex specimens by processing multiple incident angles in full-angle synthetic aperture focusing technique | |
| Chean et al. | Mutual conversion between B-mode image and acoustic impedance image | |
| KR20200040301A (en) | Model-based image reconstruction method | |
| CN109102552B (en) | Non-uniform shape constrained pixel value domain filtering ultrasonic imaging reconstruction method | |
| Jin et al. | A blind deconvolution method for attenuative materials based on asymmetrical Gaussian model | |
| CN106404923B (en) | A method of based on sparse tiny flaw high frequency ultrasound micro-imaging super-resolution | |
| Cui et al. | A robust indicator based on singular value decomposition for flaw feature detection from noisy ultrasonic signals | |
| CN115236206A (en) | High-frequency ultrasonic denoising method and system based on multipath matching pursuit algorithm improvement | |
| KR100530861B1 (en) | Method for Ultrasonic Nondestructive Testing by 3-D Processing of Pulse-Echo Signal | |
| Mohammadi et al. | Resolution improvement of scanning acoustic microscopy using sparse signal representation | |
| Naify et al. | Introduction to the special issue on Additive Manufacturing and Acoustics | |
| WO2018209791A1 (en) | Method and system for reconstructing shape of 3d object | |
| Chira et al. | Blind deconvolution for ultrasound sequences using a noninverse greedy algorithm | |
| Wu et al. | Enhancing the image resolution of acoustic microscopy with FISTA | |
| Lian et al. | Advances in Post-processing Technology for Laser Ultrasound Detection Signals: A Review |
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 |