CN108022220B - Ultrasonic image speckle noise removing method - Google Patents
Ultrasonic image speckle noise removing method Download PDFInfo
- Publication number
- CN108022220B CN108022220B CN201711275034.1A CN201711275034A CN108022220B CN 108022220 B CN108022220 B CN 108022220B CN 201711275034 A CN201711275034 A CN 201711275034A CN 108022220 B CN108022220 B CN 108022220B
- Authority
- CN
- China
- Prior art keywords
- image
- speckle
- noise
- block
- filtering
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 23
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 40
- 238000001914 filtration Methods 0.000 claims abstract description 36
- 230000000694 effects Effects 0.000 claims abstract description 17
- 238000002604 ultrasonography Methods 0.000 claims description 14
- 238000002474 experimental method Methods 0.000 claims description 13
- 230000007423 decrease Effects 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 8
- 230000008901 benefit Effects 0.000 abstract description 5
- 230000002411 adverse Effects 0.000 abstract description 4
- 238000003759 clinical diagnosis Methods 0.000 abstract description 3
- 238000003384 imaging method Methods 0.000 description 8
- 238000012360 testing method Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 3
- 230000001605 fetal effect Effects 0.000 description 3
- 238000011084 recovery Methods 0.000 description 3
- 230000002123 temporal effect Effects 0.000 description 3
- 239000000654 additive Substances 0.000 description 2
- 230000000996 additive effect Effects 0.000 description 2
- 210000003484 anatomy Anatomy 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 238000004321 preservation Methods 0.000 description 2
- 238000013441 quality evaluation Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 210000001015 abdomen Anatomy 0.000 description 1
- WYTGDNHDOZPMIW-RCBQFDQVSA-N alstonine Natural products C1=CC2=C3C=CC=CC3=NC2=C2N1C[C@H]1[C@H](C)OC=C(C(=O)OC)[C@H]1C2 WYTGDNHDOZPMIW-RCBQFDQVSA-N 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000013170 computed tomography imaging Methods 0.000 description 1
- 238000004195 computer-aided diagnosis Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 210000000232 gallbladder Anatomy 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 210000005075 mammary gland Anatomy 0.000 description 1
- 210000000496 pancreas Anatomy 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000001303 quality assessment method Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000012285 ultrasound imaging Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration by non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- 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
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Abstract
The invention discloses an ultrasonic image speckle noise removing method, wherein in the first iteration, a noise image is used as the input of a model, and a rough speckle removing image is obtained after the filtering of a Bayes non-local average filtering model; in the second iteration, the acquired speckle-removed image is used as the input of a filtering model to obtain a better speckle-removed image; the iteration process is repeated until the number of iterations reaches a preset value. Finally outputting a better speckle-removed image through a benign iterative filtering process. The invention has the advantages that: the adverse effect of speckle noise on clinical diagnosis or subsequent image processing is suppressed. And (3) calculating a conditional probability density function value in the Bayes non-local average filtering model by combining the statistical characteristics of the speckle noise, and outputting a better speckle-removed image through a benign iterative filtering process. In order to reduce the time complexity of the algorithm, three means of block filtering, pre-selecting blocks and controlling the iteration times are adopted, so that the method can be put into practical use.
Description
Technical Field
The invention relates to the technical field of ultrasonic image processing, in particular to an ultrasonic image speckle noise removing method.
Background
The ultrasonic imaging technology is a convenient and fast imaging technology, and is widely used in human body examination, particularly in the liver, gallbladder, pancreas, abdomen, mammary gland and other parts. Compared with the CT imaging technology, the ultrasonic imaging is safer, and the possible radiation damage caused by the exposure of a human body to X-rays is avoided; compared with MRI imaging technology, ultrasound imaging is much less expensive and is an important imaging modality for physical examination or first-round disease screening. However, images acquired by current ultrasound instruments are inevitably disturbed by speckle noise, which is caused by the coherent nature of the imaging by the ultrasound system. The existence of speckle noise reduces the resolution and contrast of the image, reduces the quality of the image, covers up certain detailed information, and causes adverse effects on clinical diagnosis and subsequent image processing analysis (such as feature extraction, lesion segmentation identification, image registration and the like), so that the inhibition of the speckle noise has very important significance on medical ultrasonic image analysis. Similar to other image denoising technical problems, the objective of ultrasonic image speckle noise removal is to effectively remove noise and preserve key detail texture features in an image as much as possible. However, the speckle noise is different from the gaussian noise considered in the conventional image denoising problem to a certain extent, and presents the effect of granular or serpentine speckle in form, and the noise generation model is different from additive white gaussian noise in noise statistical characteristics, so that the existing denoising method cannot be directly applied to the speckle noise elimination, but the characteristics of the speckle noise need to be fully considered in the denoising model, the existing model is improved, and the effect of the existing denoising technology on the speckle noise elimination is fully exerted.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an ultrasonic image speckle noise removing method which can effectively solve the problems in the prior art.
In order to realize the purpose, the technical scheme adopted by the invention is as follows:
a method for removing speckle noise of an ultrasonic image comprises the following specific steps:
step 1:
based on
In the above formula, z is the image containing speckle noise obtained by B-ultrasonic instrument,
is an estimated clean image, i is a pixel in the image space omegaA coordinate position; Δ (i) represents a neighborhood search region of size β × β centered on a coordinate point i, and a coordinate point j belongs to Δ (i); z (j) represents an image block with a size of α × α centered on a pixel point j in the image z; z (i) represents an image block with a pixel point i in the image z as the center and with the size of alpha multiplied by alpha; p (z (i) z (j)) is a conditional probability density function value of the image block z (i) under observation of the image block z (j),
for in an image block
The conditional probability density function value of the observed image block z (i);
each representing a speckle-free estimate of an image block centered at i; all estimated
Performing average fusion to obtain speckle-removed image
The meaning of the above formula is: in the first filtering step, a noise image z is directly used as input, and rough estimation of a speckle-free image can be obtained through weighted average and combination of image blocks; then, will again
As input for the next filtering, thereby modifying the filtering result;
p (z (i) z (j)) and p (z (i) z (j)) in a speckle noise environment are calculated according to the formula (1)
The speckle-removing image can be obtained because
Wherein n (i) is zero mean, standard deviation is sigma distributedGaussian noise, therefore z (i) u (i) N (u (i), u (i) σ2) Then, then
Step 2:
considering the assumption that the conditions of the pixel points in the image block are independent, the probability distribution condition of the image block can be obtained:
in the above equation, | R | represents the number of pixels in the image block, R represents the R-th pixel in the image block, and based on equation (2), p (z (i) | z (j)) in the first step of equation (1) can be calculated as:
and in the second step
Can be calculated as:
and step 3:
in order to further improve the speckle removing performance, the iteration times are expanded to K times, (2 ≦ K ≦ 4), and the complete iterative filtering speckle removing model is shown as the following formula:
the value of h is adjusted (5) to obtain the best despeckle effect, due to h and sigma2Correlation, let h be (C. sigma)2According to multiple experiments, the speckle removing effect is best when C is 1, and the effect is best when h is 8 under the condition that sigma cannot be known.
Preferably, the degree of overlap between image blocks is adjusted to reduce temporal complexity.
Preferably, a block preselection mechanism is added during filtering, when a block in the neighborhood is found
Mean and target image block of
If the mean values of (A) are too different, then the mean values of (B) are not considered
The contribution to the target block estimate, namely:
setting a preset block average value threshold value as mu, and setting the mu to be 0.7;
when in use
Exceed [ mu, 1/mu ]]Within range, then will
Block pair
The contribution of (a) is set to 0,
is composed of
Mean value of (i, v)
Is composed of
Is measured.
Preferably, the number of iterations K is controlled in the range of 2 to 4, and K gradually decreases to 2 as the parameter σ related to the speckle noise level increases, and a smaller number of iterations controls the time complexity of the algorithm to increase.
Compared with the prior art, the invention has the advantages that: the adverse effect of speckle noise on clinical diagnosis or subsequent image processing is greatly inhibited. And combining the statistical characteristics of speckle noise, and outputting a better speckle-removed image through a benign iterative filtering process. Meanwhile, in order to reduce the time complexity of the algorithm to the maximum extent, three means of block filtering, pre-selection of blocks and control of iteration times are adopted, so that the method can be put into practical use.
Drawings
FIG. 1 is a head noise phantom image (σ 10) and an intermediate de-speckle pattern generated by the method of the present invention in an iterative process;
fig. 1a is a head noise phantom image (σ ═ 10);
FIG. 1b is a resulting image of the proposed algorithm after the first iteration
FIG. 1c is a resulting image of the proposed algorithm after a second iteration
FIG. 1d is a resulting image of the proposed algorithm after the third iteration
FIG. 1e is a clean head phantom image;
fig. 2 is a speckle removing result diagram of different speckle removing algorithms for a head noise phantom image with σ of 10;
fig. 2a is a head noise phantom image (σ ═ 10);
FIG. 2b is a graph of the despeckle results of the SRAD;
FIG. 2c is a graph of the despeckle result of SRBF;
FIG. 2d is a plot of the despeckle results for TBNLMF;
FIG. 2e is a graph of the despeckle result of OBNLMF;
FIG. 2f is a graph of the despeckle result of our proposed algorithm;
FIG. 2g is a clean head phantom image;
FIG. 3 is a simulated ultrasound fetal image and a plaque removal image;
FIG. 3a is a simulated ultrasound fetal image generated by Field II Simulation Program;
FIG. 3b is the despeckle image of FIG. 3a for the proposed algorithm;
FIG. 4 is a real ultrasound image and a despeckle image;
FIG. 4a is a true ultrasound image;
FIG. 4b is a despeckle image produced by the proposed algorithm.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and examples.
Conventionally, speckle noise is considered to be a multiplicative signal model that obeys rayleigh distribution, however due to the complexity of the imaging mechanism, such a model is not suitable for modeling true speckle noise. In recent years, some scholars have proposed a more suitable speckle noise model, as follows:
z(i)＝u(i)+uγ(i)n(i). (1)
in the above equation, z is an image obtained by a B-ultrasonic apparatus and containing speckle noise, u is a clean image that we expect to recover, n is gaussian noise distributed with a standard deviation of σ and obeying a zero mean, i is a coordinate position of a pixel in an image space Ω, and γ is a parameter related to an ultrasonic apparatus, and is usually equal to 0.5. As can be seen from equation (1), the speckle noise is different from additive gaussian noise, and cannot be directly processed by using a gaussian noise removal algorithm, and the algorithm must be modified to adapt to the speckle noise environment in consideration of the statistical characteristics of the noise.
According to a Bayes non-local average filtering model, the application of the non-local average filtering in a non-Gaussian noise environment is possible. The model is as follows:
here, u (j) represents an image block having a size of α × α with a pixel point j in the image u as a center; z (i) represents an image block with a pixel point i in the image z as the center and with the size of alpha multiplied by alpha; Δ (i) represents a neighborhood search region of size β × β centered on a coordinate point i, and a coordinate point j belongs to Δ (i);
representing a speckle free estimation of an image block centered at i. All estimated
(i belongs to omega) to carry out average fusion to obtain the speckle-removed image
In fact, this is a way of noise filtering based on image blocks, formally consistent with the classical non-local average filtering model based on blocks.
It is obvious that the speckle-free image u is required as an input in the formula (2), which cannot be done, so to implement the model, a two-step iterative model is proposed, which skillfully solves the problem through a step-by-step refinement process:
the above formula represents the meanings: in the first filtering step, a noise image z is directly used as an input, and a rough estimation of a speckle-free image can be obtained; the estimated image is then used as input for the next filtering, thereby modifying the filtering result.
And then, removing speckle noise in the ultrasonic image by using the model corresponding to the formula (2), and deducing a probability density function required by the model according to the statistical characteristics of the speckle noise to obtain a good speckle removing effect. However, when the bayesian non-local average filtering model is implemented, the prior classical document omits the step of subsequent refinement, and the obtained speckle-removed image is rough. To solve the problem, the patent provides an ultrasound image speckle noise removing method, which specifically includes:
based on model (3), we only need to calculate p (z (i) z (j)) and p (z (i) z (j)) in a speckle noise environment
And obtaining a speckle-removing image. Due to the fact that
Therefore z (i) | u (i) to N (u (i), u (i) σ2) Then, then
Considering the assumption that the conditions of the pixel points in the image block are independent, the probability distribution condition of the image block can be obtained:
in the above equation, | R | represents the number of pixels in the image block, and R represents the R-th pixel in the image block. Based on (4), p (z (i) z (j)) in the first step of model (3) can be calculated as:
and in the second step
Can be calculated as:
in addition, in order to further improve the speckle removing performance, the iteration times are expanded to K times (2 ≦ K ≦ 4), and then the complete iterative filtering speckle removing model is as follows:
note that in the above formula we have replaced 2 σ with the variable h2In the experiments we adjusted the value of h to obtain the best depigmenting effect. Finally obtained
I.e. output
Due to h and sigma2Related, let h be (C σ)2According to multiple experiments, the speckle removing effect is best when C is 1. In the case where σ cannot be known, the value of h is 8, which is the best effect.
In order to reduce the time complexity of the iterative filtering algorithm, the method adopts the following three key processing modes:
(1) as can be seen from equation (7), the despeckle algorithm needs to iterate K times, so each filtering cannot take too much time, otherwise the algorithm is too time consuming. In fact, the filtering mode based on image blocks can effectively control the filtering time consumption by adjusting the overlapping degree between the image blocks (or the distance between the centers of the adjacent image blocks, which is set as d). According to analysis of the relevant literature [12]Therefore, the following steps are carried out: the temporal complexity of the non-local mean filtering algorithm based on blocks is O ((2 α +1)2(2β+1)2((N-d)/d)2) Where N is the side length of the image. It can be seen that if the distance d between image blocks is set to 3, the time for one filtering can be shortened by about 9 times compared with the time when d is 1, and the speckle removing performance will not be greatly different. And the iteration is performed for K times, so that 9K times of time can be saved.
(2) When the mean value of the blocks in the neighborhood is found to be too different from the target image block mean value, the contribution of the image block to the target block estimation value can be not considered, namely:
exceed [ mu, 1/mu ]]Range (μ is a preset block mean threshold, usuallyLet μ equal to 0.7), then
Block pair
The contribution of (1) is set to 0, so that the corresponding calculation time is shortened, and the adverse effect of the accumulated error on the final estimation is avoided.
(3) The number of iterations K is controlled to be in the range of 2 to 4, and as the speckle noise level σ increases, K gradually decreases to 2, with fewer iterations controlling the temporal complexity of the algorithm to grow.
Experiment and result analysis:
to test the performance of the proposed despeckle algorithm, we chose three test images, the first one that is a speckle noise image simulated according to the assumed noise model, the second one that is an ultrasound image simulated, and the third one that is a true ultrasound image obtained directly from the B-ultrasonic instrument. Meanwhile, a plurality of representative spot removing algorithms are selected for comparison with the algorithm, and the algorithms are respectively as follows: SRAD, SRBF, legacy blockwise NLMF (TBNLMF), OBNLMF. For the iterative filtering algorithm provided by the invention, in order to obtain a better speckle removing effect, the parameters are set as follows: the range Δ (i) of the non-local neighborhood is set to 19 × 19; the size of the image block is set to 5 × 5; when sigma is less than 10, the iteration number K is set to be 4, when the sigma is more than or equal to 10 and less than 20, the K is 3, and when the sigma is more than or equal to 20, the K is 2; h is set to sigma in the filter model2(ii) a In addition, when sigma is unknown, the value of h is adjusted according to the actual situation, and K is set to be 2 in consideration of the factors of convenience and rapidness. Below we verify the advantages of the proposed algorithm by three different sets of experiments.
Head noise phantom image experiment:
a plurality of head phantom images with different noise degrees are generated by using an assumed speckle noise model simulation, and for the test images, because corresponding speckle-free clean images are known, the quality of recovery images output by different algorithms can be quantitatively evaluated by adopting a full reference image quality evaluation index. The indexes used here are Peak Signal-to-Noise Ratio (PSNR) and Feature Similarity Index Measure (FSIM). The PSNR has the following calculation formula:
where MN is the size of the image and FSIM is an image quality evaluation index developed in recent years, where the phase consistency and gradient magnitude of the image are used to measure the difference of the image pair. When restoring the image
The closer to the clean image u, the closer to 1 the value of FSIM.
First, to test the correctness of the proposed iterative algorithm, we designed the following experiment. Under different noise levels, all restored images generated in the iterative process are recorded and respectively recorded as
And calculating corresponding PSNR and FSIM values, wherein the experimental results are shown in Table 1. The bold data in the table represents the best recovery result value for each row. It can be seen that, at any sigma,
positive ratio
And, when σ is small, as the number of iterations K increases,
it will be better and better, which fully proves the correctness of the proposed iterative filtering despeckle algorithm. Meanwhile, the algorithm is shown in fig. 1, which is generated in the process of processing head noise phantom images with sigma of 10
The correction process for the restored image for each iteration can be clearly seen.
TABLE 1 intermediate denoising results (PSNRs, FSIMs) of the iterative algorithm proposed in processing "head noise phantom images
Then, the head noise phantom images with different noise degrees are input into the selected speckle reduction algorithm to obtain different speckle reduction images, and the PSNR and FSIM values of the different speckle reduction images are calculated, and the result is shown in Table 2. Also, the bold data in the table represents the best recovery result value for each row. It can be seen that the proposed algorithm is the best one under the image quality assessment indicators of PSNR and FSIM. Meanwhile, we show a speckle removal result graph of each algorithm on a head noise phantom image with σ of 10 in fig. 2, and visually see that: the SRAD algorithm has strong capability of keeping details, but a plurality of spots still remain in the speckle removing image; the speckle removing image of the SRBF has a plurality of artifacts; the TBNLMF algorithm can remove most of spots, but loses many detail features; OBNLMF can achieve a certain balance between speckle removal and detail preservation, but noise traces still exist in the image background; the algorithm proposed by us not only removes most of the noise, but also performs well on the preservation of details and edges.
TABLE 2 speckle removal results (PSNRs, FSIMs) of different speckle removal algorithms on head noise phantom images of different noise levels
Ultrasound image experiments generated by Field II Simulation Program Simulation:
the Field II Simulation Program is an ultrasonic image Simulation generation Program developed according to the ultrasonic imaging principle, and has a high reference value. In this experiment, we selected the "fetal" image generated by the Field II Simulation Program Simulation as our test image, as shown in FIG. 3 a. This is then input into our proposed despeckle algorithm (note: h in the filter model is set to 12 since there is no information of σ) and the output despeckle image is shown in fig. 3 b. Comparing the two images, we can clearly see that: most of the speckle noise in the right image has been removed while the anatomy is still clearly intact, which fully explains the effectiveness of the proposed speckle removal algorithm.
And (3) real ultrasonic image experiment:
through the verification of the first two experiments, we confirm the effectiveness of the proposed algorithm, and in this experiment, a real ultrasound image is used as a test object, as shown in fig. 4 a. Then it is despecked using the proposed algorithm (note: h in the filter model is set to 8 in order to preserve as much detail as possible while despeckling), and the resulting image is shown in fig. 4 b. As can be seen from the figure: the de-speckled image not only removes speckle noise but also preserves the edges between adjacent anatomical structures well, which is certainly of great benefit to the diagnostic of the image diagnostician. Of course, what is illustrated here is: generally, for medical image processing, which is called as "computer-aided diagnosis", a doctor wants to combine the original image and the processed image to make a diagnosis result, rather than referring to the processed image only, because any signal processing method loses more or less detail characteristics of potential signals, and the diagnosis is more accurate compared with the reference of the image.
It will be appreciated by those of ordinary skill in the art that the examples described herein are intended to assist the reader in understanding the manner in which the invention is practiced, and it is to be understood that the scope of the invention is not limited to such specifically recited statements and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (4)
1. A method for removing speckle noise of an ultrasonic image is characterized by comprising the following specific steps:
step 1:
based on
In the above formula, z is the image containing speckle noise obtained by B-ultrasonic instrument,
is the estimated clean image, i is the coordinate position of the pixel in image space Ω; Δ (i) represents a neighborhood search region of size β × β centered on a coordinate point i, and a coordinate point j belongs to Δ (i); z (j) represents an image block with a size of α × α centered on a pixel point j in the image z; z (i) represents an image block with a pixel point i in the image z as the center and with the size of alpha multiplied by alpha; p (z (i) z (j)) is a conditional probability density function value of the image block z (i) under observation of the image block z (j),
for in an image block
The conditional probability density function value of the observed image block z (i);
each representing a speckle-free estimate of an image block centered at i; all estimated
Performing average fusion to obtain speckle-removed image
The meaning of the above formula is: in the first filtering step, a noise image z is directly used as input, and rough estimation of a speckle-free image can be obtained through weighted average and combination of image blocks; then, will again
As input for the next filtering, thereby modifying the filtering result;
p (z (i) z (j)) and p (z (i) z (j)) in a speckle noise environment are calculated according to the formula (1)
The speckle-removing image can be obtained because
Where N (i) is zero mean, Gaussian noise with standard deviation σ distribution, so z (i) | u (i) to N (u (i), u (i) σ2) Then, then
Step 2:
considering the assumption that the conditions of the pixel points in the image block are independent, the probability distribution condition of the image block can be obtained:
in the above equation, | R | represents the number of pixels in the image block, R represents the R-th pixel in the image block, and based on equation (2), p (z (i) | z (j)) in the first step of equation (1) can be calculated as:
and in the second step
Can be calculated as:
and step 3:
in order to further improve the speckle removing performance, the iteration times are expanded to K times, (2 ≦ K ≦ 4), and the complete iterative filtering speckle removing model is shown as the following formula:
the value of h is adjusted (5) to obtain the best despeckle effect, due to h and sigma2Correlation, let h be (C. sigma)2According to multiple experiments, the speckle removing effect is best when C is 1, and the effect is best when h is 8 under the condition that sigma cannot be known.
2. The method for removing speckle noise in an ultrasound image according to claim 1, wherein: and the overlapping degree between the image blocks is adjusted, so that the time complexity is reduced.
3. The method for removing speckle noise in an ultrasound image according to claim 1, wherein: adding a block preselection mechanism during filtering, when a block in the neighborhood is found
Mean and target image block of
If the mean values of (A) are too different, then the mean values of (B) are not considered
The contribution to the target block estimate, namely:
setting a preset block average value threshold value as mu, and setting the mu to be 0.7;
when in use
Exceed [ mu, 1/mu ]]Within range, then will
Block pair
The contribution of (a) is set to 0,
is composed of
The average value of (a) of (b),
is composed of
Is measured.
4. The method for removing speckle noise in an ultrasound image according to claim 1, wherein: the number of iterations K is controlled in the range of 2 to 4 and decreases gradually to 2 as the parameter σ related to the speckle noise level increases, with fewer iterations controlling the time complexity of the algorithm to grow.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711275034.1A CN108022220B (en) | 2017-12-06 | 2017-12-06 | Ultrasonic image speckle noise removing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711275034.1A CN108022220B (en) | 2017-12-06 | 2017-12-06 | Ultrasonic image speckle noise removing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108022220A CN108022220A (en) | 2018-05-11 |
| CN108022220B true CN108022220B (en) | 2021-04-20 |
Family
ID=62078607
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711275034.1A Active CN108022220B (en) | 2017-12-06 | 2017-12-06 | Ultrasonic image speckle noise removing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108022220B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109087266A (en) * | 2018-08-09 | 2018-12-25 | 苏州大学 | A kind of image speckle iteration reduction method of combination bilateral filtering |
| CN110390646B (en) * | 2019-06-12 | 2023-10-17 | 西南科技大学 | Detail-preserving image denoising method |
| CN114037616B (en) * | 2021-09-22 | 2024-04-05 | 南京莱斯电子设备有限公司 | SAR image noise suppression method and device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5594807A (en) * | 1994-12-22 | 1997-01-14 | Siemens Medical Systems, Inc. | System and method for adaptive filtering of images based on similarity between histograms |
| CN101661611A (en) * | 2009-09-25 | 2010-03-03 | 西安电子科技大学 | Realization method based on bayesian non-local mean filter |
| US7744534B2 (en) * | 2002-06-07 | 2010-06-29 | Verathon Inc. | 3D ultrasound-based instrument for non-invasive measurement of amniotic fluid volume |
| CN102609904A (en) * | 2012-01-11 | 2012-07-25 | 云南电力试验研究院（集团）有限公司电力研究院 | Bivariate nonlocal average filtering de-noising method for X-ray image |
| CN103886563A (en) * | 2014-04-18 | 2014-06-25 | 中南民族大学 | SAR image speckle noise inhibition method based on non-local mean and heterogeneity measurement |
| CN103927715A (en) * | 2014-03-14 | 2014-07-16 | 中瑞科技（常州）有限公司 | Ultrasound image speckle noise suppression method |
| US8928813B2 (en) * | 2010-10-28 | 2015-01-06 | Microsoft Corporation | Methods and apparatus for reducing structured noise in video |
| CN107220941A (en) * | 2016-12-15 | 2017-09-29 | 西南科技大学 | A kind of method and system for removing image impact noise |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR102025756B1 (en) * | 2013-01-04 | 2019-09-27 | 삼성전자주식회사 | Method, Apparatus and system for reducing speckles on image |
-
2017
- 2017-12-06 CN CN201711275034.1A patent/CN108022220B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5594807A (en) * | 1994-12-22 | 1997-01-14 | Siemens Medical Systems, Inc. | System and method for adaptive filtering of images based on similarity between histograms |
| US7744534B2 (en) * | 2002-06-07 | 2010-06-29 | Verathon Inc. | 3D ultrasound-based instrument for non-invasive measurement of amniotic fluid volume |
| CN101661611A (en) * | 2009-09-25 | 2010-03-03 | 西安电子科技大学 | Realization method based on bayesian non-local mean filter |
| US8928813B2 (en) * | 2010-10-28 | 2015-01-06 | Microsoft Corporation | Methods and apparatus for reducing structured noise in video |
| CN102609904A (en) * | 2012-01-11 | 2012-07-25 | 云南电力试验研究院（集团）有限公司电力研究院 | Bivariate nonlocal average filtering de-noising method for X-ray image |
| CN103927715A (en) * | 2014-03-14 | 2014-07-16 | 中瑞科技（常州）有限公司 | Ultrasound image speckle noise suppression method |
| CN103886563A (en) * | 2014-04-18 | 2014-06-25 | 中南民族大学 | SAR image speckle noise inhibition method based on non-local mean and heterogeneity measurement |
| CN107220941A (en) * | 2016-12-15 | 2017-09-29 | 西南科技大学 | A kind of method and system for removing image impact noise |
Non-Patent Citations (3)
| Title |
|---|
| Pierrick Coup＇等.BAYESIAN NON LOCAL MEANS-BASED SPECKLE FILTERING.《IEEE International Symposium on Biomedical Imaging: from Nano to Macro》.2008,第1291-1294页. * |
| Video Saliency Incorporating Spatiotemporal Cues and Uncertainty Weighting;Yuming Fang等;《IEEE TRANSACTIONS ON IMAGE PROCESSING》;20140716;第23卷(第9期);第3910-3921页 * |
| 超声图像去噪方法;韩寒等;《计算机工程与应用》;20111231;第47卷(第26期);第193-195页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108022220A (en) | 2018-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8842936B2 (en) | Method, apparatus, and program for aligning images | |
| JP3995854B2 (en) | Image processing method and apparatus, and recording medium | |
| Nobi et al. | A New Method to Remove Noise in Magnetic Resonance and Ultrasound Images. | |
| Coupé et al. | Bayesian non local means-based speckle filtering | |
| CN109961411B (en) | Non-downsampling shear wave transformation medical CT image denoising method | |
| CN108022220B (en) | Ultrasonic image speckle noise removing method | |
| de Araujo et al. | Smoothing of ultrasound images using a new selective average filter | |
| CN106530236B (en) | Medical image processing method and system | |
| EP2372646B1 (en) | Image processing device, method and program | |
| JP2005169077A (en) | Method for finding transformation relating data point of first data set to data point of second data set, and computer program for performing the method | |
| CN108038840B (en) | Image processing method and device, image processing equipment and storage medium | |
| Bodzioch et al. | New approach to gallbladder ultrasonic images analysis and lesions recognition | |
| CN109087266A (en) | A kind of image speckle iteration reduction method of combination bilateral filtering | |
| Li et al. | Unpaired low‐dose computed tomography image denoising using a progressive cyclical convolutional neural network | |
| CN111833383B (en) | Two-dimensional joint local displacement fitting region growing Bayesian motion tracking method | |
| JP2019515255A (en) | Medical imaging | |
| Roy et al. | Speckle de-noising of clinical ultrasound images based on fuzzy spel conformity in its adjacency | |
| Hai et al. | A fast denoising algorithm for X-Ray images with variance stabilizing transform | |
| Khodadadi et al. | Edge-preserving ultrasonic strain imaging with uniform precision | |
| Kumar et al. | Semiautomatic method for segmenting pedicles in vertebral radiographs | |
| EP2693397A1 (en) | Method and apparatus for noise reduction in an imaging system | |
| JP5537266B2 (en) | Energy subtraction processing apparatus and method, and program | |
| Martínez et al. | Estimating the size of polyps during actual endoscopy procedures using a spatio-temporal characterization | |
| Hai | Wavelet-based image fusion for enhancement of ROI in CT image | |
| Saoji et al. | Speckle and rician noise removal from medical images and Ultrasound images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |