CN108550121A - A kind of sediment sonar image processing method based on medium filtering and wavelet transformation - Google Patents
A kind of sediment sonar image processing method based on medium filtering and wavelet transformation Download PDFInfo
- Publication number
- CN108550121A CN108550121A CN201810291591.0A CN201810291591A CN108550121A CN 108550121 A CN108550121 A CN 108550121A CN 201810291591 A CN201810291591 A CN 201810291591A CN 108550121 A CN108550121 A CN 108550121A
- Authority
- CN
- China
- Prior art keywords
- image
- noise
- layer
- node
- high frequency
- 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
- 230000009466 transformation Effects 0.000 title claims abstract description 35
- 239000013049 sediment Substances 0.000 title claims abstract description 29
- 238000001914 filtration Methods 0.000 title claims abstract description 22
- 238000003672 processing method Methods 0.000 title claims abstract description 14
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 37
- 238000000034 method Methods 0.000 claims abstract description 36
- 239000000654 additive Substances 0.000 claims abstract description 24
- 230000000996 additive effect Effects 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 8
- 230000008569 process Effects 0.000 claims description 14
- 238000010586 diagram Methods 0.000 claims description 3
- 230000014509 gene expression Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 2
- 230000000750 progressive effect Effects 0.000 claims description 2
- 241000549548 Fraxinus uhdei Species 0.000 claims 1
- 239000000758 substrate Substances 0.000 abstract description 7
- 230000000694 effects Effects 0.000 abstract description 6
- 238000013507 mapping Methods 0.000 abstract description 3
- 230000000717 retained effect Effects 0.000 abstract description 2
- 239000002609 medium Substances 0.000 description 11
- 238000000605 extraction Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 3
- 238000003706 image smoothing Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 235000007926 Craterellus fallax Nutrition 0.000 description 1
- 240000007175 Datura inoxia Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 239000012736 aqueous medium Substances 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
- 239000013535 sea water Substances 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S15/00—Systems using the reflection or reradiation of acoustic waves, e.g. sonar systems
- G01S15/88—Sonar systems specially adapted for specific applications
- G01S15/89—Sonar systems specially adapted for specific applications for mapping or imaging
Abstract
The sediment sonar image processing method based on medium filtering and wavelet transformation that the invention discloses a kind of, belongs to seabed mapping field.Contain the sediment sonar original image of speckle noise and gaussian additive noise first against certain, carries out that multiplicative noise is become additivity after logarithmic transformation, obtain the image containing approximate Gaussian additive noise；Then the image containing approximate Gaussian additive noise is filtered using Optimal Space method, row index of going forward side by side transformation obtains smoothed image；The processing of Wavelet Denoising Method based on WAVELET PACKET DECOMPOSITION is carried out to smoothed image, obtains final denoising image.The present invention plays the role of edge sharpening to a certain extent, better ensures that the expressed intact of substrate image information, and deeper decomposition is carried out to the low-and high-frequency signal that decomposition obtains；The main component of original image can more be retained, obtain signal-to-noise ratio higher, the better denoising image of denoising effect.
Description
Technical field
The invention belongs to seabed mapping fields, are related to Sonar Image Smoothing and denoising, and specifically one kind is based on
The sediment sonar image processing method of medium filtering and wavelet transformation.
Background technology
Sediment sonar image can reflect the submarine geomorphy and substrate characteristic information compared with horn of plenty, be submarine geomorphy detection
With the important information source of sediment feature extraction, inverting of classifying.The sound wave that sonar is sent out is via the water surface, seawater, water-bed composition
Channel propagated, the aqueous medium of underwater acoustic channel and boundary have extremely complicated characteristic, while by ambient sea noise harmony
The limitation of work characteristics and performance, sediment sonar image is serious with noise jamming, image resolution ratio is low and edge
The features such as texture obscures.
When side scan sonar and multibeam sonar carry out seabed mapping, the echo of measurement is mainly bottom reverberation, for around
The random fluctuation that mean intensity occurs, shows obvious speckle noise, operation principle and the performance limit of the phenomenon and sonar
It is formed with pass.Similar noise phenomenon appears in medical ultrasonic image, SAR image.It is had shown that through research, this speckle noise is obeyed
Multiplicative noise model based on rayleigh distributed.In addition, marine environment stochastic variable can be considered gaussian additive noise, this part is made an uproar
Influence of the sound compared with speckle noise to image is smaller.For multiplicative noise, logarithmic transformation can be first carried out to image, become multiplicative noise into
Additive noise, to be removed with conventional additive noise suppression technology.
Bottom mounted sonar image is a kind of gray level image, and color recognition auxiliary, and mud, grit class bottom are lacked in feature extraction
For matter compared with roughness for rock type substrate and textural characteristics this unobvious, this makes most scholars carry out correlation to sediment and grind
Pay attention to substrate image characteristics extraction and the research of Seafloor Classification algorithm when studying carefully, ignores the image pre-processing phase before parameter extraction.
The noise and edge of sonar image are in Gray Level Jump part, and denoising can influence edge gray scale when changing noise spot gray value
Value, to weaken picture edge characteristic.
Wavelet transformation technique is the mathematic(al) manipulation method for carrying out time-frequency domain partial analysis to signal by wavelet function, can be with
The convolution algorithm for regarding original signal and wavelet function race and scaling function after scale is flexible as, by with different time-frequencies
The small echo of width matches the heterogeneity in original signal, realizes that the localization to signal is analyzed.Consider from filter angles,
Wavelet transformation is considered as the filtering operation of original signal and the small pass band filter of one group of different scale again, by signal decomposition
To in different frequency bands, there is preferable temporal resolution in the high frequency section of signal, there is preferable frequency in low frequency part
Rate resolution ratio, so as to extract effective information from signal (such as voice, image).Simultaneously as the letter that wavelet transformation includes
Number decomposition method can carry out deeper decomposition to the low-and high-frequency signal that decomposition obtains, so in threshold denoising utmostly
Retain the main component in original signal, obtains signal-to-noise ratio higher, the better denoising image of denoising effect.
Just because having the advantage that many conventional methods do not have in image processing method face, wavelet transformation is by more and more sections
Grind personnel applied to the image preprocessings such as remote sensing images, EO-1 hyperion dispersion image, microbiology cell image, morphological image and
Feature extraction step.As rectified within 2008, extra large roc et al. exists《A kind of image wavelet de-noising method based on medium filtering》It is carried in invention
Go out and carry out medium filtering pre-filtering using to digital remote sensing image, recycles the wavelet transformation decomposed based on wavelet basis to pre-filtering
Image carries out denoising, obtains denoising image.Face soldier in 2011 et al. exists《Image based on mean filter and wavelet transformation is gone
It makes an uproar technical research》The middle image de-noising method for proposing to be combined using mean filter and wavelet transformation, is first carried out noisy image
Wavelet decomposition is chosen threshold value appropriate in wavelet field and is handled wavelet coefficient, then carries out part weight to picture signal
Structure carries out mean filter to first layer, and using different templates to each details subgraph of first layer, finally by low-frequency approximation
Image synthesizes to obtain the image after denoising with high frequency detail image after mean filter.
But mean filter itself is a kind of linear filter, when including noise spot within the neighborhood of pixels of processing, is made an uproar
The presence of sound can influence the calculated for pixel values of the point to a certain extent.Wavelet Denoising Method based on wavelet decomposition only to low frequency part into
Row decomposes again, does not handle high-frequency information.
Invention content
Have and weaken to a certain degree to being all the marginal information of Gray Level Jump position for existing sonar image denoising process
The problem of, the present invention proposes a kind of sediment sonar image processing method based on medium filtering and wavelet transformation；It is one
Image processing method of the kind suitable for the first smooth rear denoising of sediment sonar image speckle noise feature.
It is as follows：
Step 1：The sediment sonar original image for containing speckle noise and gaussian additive noise for certain, carries out pair
Multiplicative noise is become into additivity after transformation of variables, obtains the image containing approximate Gaussian additive noise；
First, the speckle noise of sonar original image obeys the Multiplicative noise model based on rayleigh distributed, is expressed as：I=
RZ；
I is the signal containing spot observed from original image；R be it is expected to restore true noisy image, Z do not make an uproar for spot
Sound stochastic variable.
Then, multiplicative noise is become into approximate Gaussian additive noise form using logarithmic transformation：LnI=lnR+lnZ；
Step 2：The image containing approximate Gaussian additive noise is filtered using Optimal Space method；
Specially：With filter window to the image progressive traverse scanning containing approximate Gaussian additive noise, handle in window
Each pixel when, judge the pixel gray value whether be grey scale pixel value in current filter window very big or minimum；
If it is very big or minimum, then the pixel is noise spot, using the median filter process pixel, i.e. neighborhood territory pixel
Gray scale intermediate value as the grey scale pixel value；Otherwise, it is not handled, exports the gray value and remain unchanged.
Step 3：Exponential transform is carried out to filtered image, obtains smoothed image；
Exponential transform, that is, logarithmic transformation inverse operation；
Step 4：The processing of Wavelet Denoising Method based on WAVELET PACKET DECOMPOSITION is carried out to smoothed image, obtains final denoising image.
It is as follows：
Step 4.1 carries out image wavelet packet decomposition to smoothed image, obtains the high frequency node image and low frequency node of each layer
Image；
Detailed process is：First, suitable wavelet basis and Decomposition order are selected；Then it is decomposed since first layer and obtains two
A node becomes two tree nodes of the second layer, corresponds to high frequency tree node and low frequency tree node respectively；Then by the two of the second layer
A node decomposes simultaneously, obtains high frequency child node and low frequency child node again respectively, becomes four nodes of the third line；It is similarly right
Third layer is decomposed, and the 4th layer of eight nodes are obtained, and so on.
Corresponding two child nodes are respectively high frequency node and low frequency node under every layer of high frequency tree node；Every layer of low frequency tree
Corresponding child node is respectively high frequency node and low frequency node under node；
Step 4.2 carries out wavelet packet threshold denoising to the corresponding coefficient of every layer of each high frequency node image, and it is each to obtain the layer
High frequency coefficient after a denoising：
First, for jth layer, the corresponding each coefficient of each high frequency node image of the layer is counted, the estimation for calculating this layer is made an uproar
Sound standard variance；
Noise criteria variances sigma calculates as follows：
dj(w) the corresponding one group of coefficient of image of w-th of high frequency node in small echo jth layer is indicated, med is during Matlab is asked
The order of value.
Then, the wavelet packet noise-removed threshold value λ of jth layer is calculated using noise criteria variance；
N is the picture size of w-th of high frequency node of small echo jth layer.
Finally, each high frequency section in jth layer is judged as sediment image threshold denoising rule using soft-threshold rule
The absolute value of the corresponding coefficient sets of image of point further calculates the small echo after each application threshold value with the size of noise-removed threshold value λ
Coefficient wλValue.
Specific judgement is as follows：To dj(w) each coefficient carries out threshold process, if absolute coefficient is less than noise-removed threshold value λ
When, the corresponding w of w-th of high frequency node in jth layerλValue is 0；Otherwise, wλValue is to subtract noise-removed threshold value, i.e.,：
Sign (w) expressions take coefficient dj(w) symbol.
Step 4.3 is directed to every layer, and the low frequency coefficient of high frequency coefficient and this layer after this layer of each denoising is reconstructed,
Obtain complete denoising image：
For the bottom, by the high frequency coefficient coefficient corresponding with each low frequency node image of this layer after this layer of each denoising
It being reconstructed, reconstruct obtains the node of last layer to the node of the bottom upwards, and so on, until all straton figures have been reconstructed into
Whole denoising image.
The approximate part in low-frequency image is protected while eliminating high-frequency noise, completes sediment sonar image
Denoising process.
The advantage of the invention is that：
1), a kind of sediment sonar image processing method based on medium filtering and wavelet transformation, using based on neighborhood
The medium filtering image smoothing method of maximum carries out marginal information prior to identifying marginal point and noise spot before image denoising
Retain, noise information smoothly, play the role of edge sharpening to a certain extent, better ensure that substrate image information
Expressed intact.
2) a kind of, sediment sonar image processing method based on medium filtering and wavelet transformation utilizes wavelet packet point
Solution method decomposes the substrate sonar image in denoising stage, improve wavelet decomposition low frequency subgraph can only be continued frequency dividing and
The problem of high frequency subgraph cannot be decomposed, carries out deeper decomposition to the low-and high-frequency signal that decomposition obtains；The decomposition method is real
Existing image multi-level simulation tool, when extracting each layer threshold value and carrying out denoising, can more retain original image it is main at
Point, obtain signal-to-noise ratio higher, the better denoising image of denoising effect；Reach better denoising effect.
3) a kind of, sediment sonar image processing method based on medium filtering and wavelet transformation, the intermediate value filter of use
Wave device is substantially sort method filter, belongs to nonlinear filter, and noise spot is often directly neglected in medium filtering
It omits, in noise reduction, caused blurring effect is low.
Description of the drawings
Fig. 1 is a kind of sediment sonar image process flow based on medium filtering and wavelet transformation of the present invention
Figure.
Fig. 2 is WAVELET PACKET DECOMPOSITION structural schematic diagram of the present invention.
Specific implementation mode
The specific implementation method of the present invention is described in detail below in conjunction with the accompanying drawings.
The present invention propose it is a kind of suitable for multiplicative noise containing spottiness and gaussian additive noise it is first smooth after denoising
Sediment sonar image preprocess method, the effect of image smoothing be prior to identifying marginal point and noise spot before image denoising,
Marginal information is retained, noise information is carried out smoothly, to play the role of edge sharpening to a certain extent；To smoothed image
Denoising, multi-level exploded view image height are integrally carried out to image using the wavelet threshold denoising method based on WAVELET PACKET DECOMPOSITION again
Low frequency subgraph, adaptively selected frequency band threshold obtain resolution ratio higher, the clearer sediment denoising image of main information；
It is more preferable to retain the interference of raw information cancelling noise.
As shown in Figure 1, being as follows：
Step 1：The sediment sonar original image for containing speckle noise and gaussian additive noise for certain, carries out pair
Multiplicative noise is become into additivity after transformation of variables, obtains the image containing approximate Gaussian additive noise；
For sonar collected image containing spot I, it includes most of form of noise be speckle noise, spot makes an uproar
Sound is multiplicative noise, is expressed from the next to the Multiplicative noise model of Rayleigh distributed：
I=RZ
Wherein, R is the true not noisy image for it is expected to restore, and Z is speckle noise stochastic variable.
Logarithmic transformation is carried out to sonar acquisition image, change multiplicative noise is approximate Gaussian additive noise form.
LnI=lnR+lnZ
After above-mentioned transformation, the image containing speckle noise is changed to the image containing approximate Gaussian additive noise.
Step 2：The image of the additive noise containing approximate Gaussian is filtered using the median filtering method of neighborhood extreme value；
Specially：
First, traverse scanning is carried out to the image of the additive noise containing approximate Gaussian with filter window, handled every in window
When a pixel, judge current pixel gray value whether be grey scale pixel value in current filter window very big or minimum；
If matrix X [i, j] indicates the corresponding character matrix of sonar image gray value, wherein (i, j) indicates various point locations, x
(i, j) is the gray value.W [i, j] indicates the filter window that is arranged centered on point (i, j), according to sonar image size and
Window size is arranged in smoothness requirements, considers that the common size of substrate sonar image is about 120*120, sets filter window ruler herein
Very little is 5*5.
Then, judge whether currently processed grey scale pixel value is window extreme value, if it is, assert that the point is noise
Point, using median filter process, the pixel carries out smoothly；Otherwise, currently processed grey scale pixel value is not window extreme value, then not
It is handled, exports the gray value and remain unchanged.
Using median filter process, the pixel refers to：Using the gray scale intermediate value of neighborhood territory pixel as the grey scale pixel value.I.e.：
X (i, j)=med (W [i, j])
Med (W [i, j]) indicates the intermediate value of current window all pixels gray value.
Step 3：Exponential transform is carried out to the image that filtering is completed, i.e. the inverse transformation of logarithmic transformation obtains smoothed image；
Step 4：The processing of Wavelet Denoising Method based on WAVELET PACKET DECOMPOSITION is carried out to smoothed image, obtains final denoising image.
Denoising is carried out to image using wavelet transformation and generally follows following steps：(1) WAVELET PACKET DECOMPOSITION：Selection is suitable
Wavelet basis and Decomposition order to image carry out decomposition computation, calculate wavelet packet coefficient；(2) threshold denoising：That is wavelet transformed domain
The Nonlinear Processing of middle wavelet coefficient, each layer choosing select a suitable threshold value, and threshold value quantizing processing is carried out to high frequency coefficient.
(3) image reconstruction：By the low frequency coefficient of wavelet decomposition n-th layer and by threshold value quantizing, treated, and each layer high frequency coefficient carries out figure
As reconstruct.
WAVELET PACKET DECOMPOSITION is to carry out pretreated important link to image using wavelet transformation, be it is a kind of more than wavelet decomposition
Add fine signal analysis method, the high frequency section that wavelet decomposition is not segmented is decomposed again, and can be according to being divided
The signal characteristic of analysis, is adaptive selected frequency band, is allowed to match with signal spectrum, to promote time frequency resolution.
It is as follows：
Step 4.1 carries out WAVELET PACKET DECOMPOSITION to smoothed image, obtains the high frequency node image and low frequency node diagram of each layer
Picture；
The information of image is divided into low frequency channel and high frequency is believed by the substantially image scaling down processing that image wavelet decomposes
Then road carries out denoising to noise-containing high frequency section, the small of two dimensional image can be realized using wpdec2 in MATLAB
Wave packet operation splitting.
As shown in Fig. 2, enabling WAVELET PACKET DECOMPOSITION number of plies j=3, image wavelet packet decomposable process is as follows：One layer of decomposition obtains two
A node becomes two tree nodes of the second row, corresponding high frequency section and low frequency part；Two layers are decomposed to low frequency node and high frequency
Node decomposes simultaneously, respectively obtains high frequency section and low frequency part, becomes four nodes of the third line；Three layers are decomposed similarly, are obtained
To eight nodes of fourth line.
Corresponding two child nodes are respectively high frequency node and low frequency node under every layer of high frequency node；Every layer of low frequency node
Under corresponding child node be respectively high frequency node and low frequency node；
Step 4.2 carries out wavelet packet threshold denoising to the corresponding coefficient of each high frequency node image of decomposition obtain every layer, obtains
To the high frequency coefficient of this layer of each denoising：
First, for jth layer, this layer of each corresponding each coefficient of node image for including high-frequency information is counted, by right
This group of coefficient seeks intermediate value, calculates the estimation noise criteria variance of this layer；
Noise criteria variances sigma calculates as follows：
J is the wavelet decomposition number of plies, dj(w) the corresponding system of image of w-th of high frequency node in small echo jth layer is indicated
Number, med is the order that Matlab seeks intermediate value.
Then, the wavelet packet noise-removed threshold value λ of jth layer is calculated using noise criteria variance；
N is the picture size of w-th of high frequency node of small echo jth layer.
Finally, each high frequency section in jth layer is judged as sediment image threshold denoising rule using soft-threshold rule
The absolute value of the corresponding coefficient sets of image of point further calculates the small echo after each application threshold value with the size of noise-removed threshold value λ
Coefficient wλValue.
Specific judgement is as follows：To dj(w) each coefficient carries out threshold process, if absolute coefficient is less than noise-removed threshold value λ
When, the corresponding w of w-th of high frequency node in jth layerλValue is 0；Otherwise, wλValue is to subtract noise-removed threshold value, i.e.,：
Sign (w) expressions take coefficient dj(w) symbol, wλIt is the wavelet coefficient size for applying threshold value.
Step 4.3 is directed to every layer, and the low frequency coefficient of high frequency coefficient and this layer after this layer of each denoising is reconstructed,
Obtain complete denoising image：
By after WAVELET PACKET DECOMPOSITION in step 4.1 low frequency coefficient and step 4.2 in by thresholding treated each high frequency
Coefficient is reconstructed, and the reconstructed operation of 2-d wavelet packet decomposition coefficient can be realized using wprec2 orders in MATLAB.Three
Reconstruct obtains two node layers to node layer upwards, and so on, until all straton figures are reconstructed into complete denoising image.
The approximate part in low-frequency image is protected while eliminating high-frequency noise, completes sediment sonar image
Denoising process.
Claims (4)
1. a kind of sediment sonar image processing method based on medium filtering and wavelet transformation, which is characterized in that specific step
It is rapid as follows：
Step 1：Contain the sediment sonar original image of speckle noise and gaussian additive noise for certain, carries out logarithm change
Multiplicative noise is become into additivity after changing, obtains the image containing approximate Gaussian additive noise；
Step 2：The image containing approximate Gaussian additive noise is filtered using Optimal Space method；
Specially：With filter window to the image progressive traverse scanning containing approximate Gaussian additive noise, handle every in window
When a pixel, judge the pixel gray value whether be grey scale pixel value in current filter window very big or minimum；
If it is very big or minimum, then the pixel is noise spot, using the median filter process pixel, the i.e. ash of neighborhood territory pixel
Intermediate value is spent as the grey scale pixel value；Otherwise, it is not handled, exports the gray value and remain unchanged；
Step 3：Exponential transform is carried out to filtered image, obtains smoothed image；
Step 4：The processing of Wavelet Denoising Method based on WAVELET PACKET DECOMPOSITION is carried out to smoothed image, obtains final denoising image；
It is as follows：
Step 4.1 carries out image wavelet packet decomposition to smoothed image, obtains the high frequency node image and low frequency node diagram of each layer
Picture；
Step 4.2 carries out wavelet packet threshold denoising to the corresponding coefficient of every layer of each high frequency node image, obtains this layer and each goes
High frequency coefficient after making an uproar：
First, for jth layer, the corresponding each coefficient of each high frequency node image of the layer is counted, the estimation noise mark of this layer is calculated
Quasi- variance；
Noise criteria variances sigma calculates as follows：
dj(w) the corresponding one group of coefficient of image of w-th of high frequency node in small echo jth layer is indicated, med is that Matlab seeks intermediate value
Order；
Then, the wavelet packet noise-removed threshold value λ of jth layer is calculated using noise criteria variance；
N is the picture size of w-th of high frequency node of small echo jth layer；
Finally, each high frequency node in jth layer is judged as sediment image threshold denoising rule using soft-threshold rule
The absolute value of the corresponding coefficient sets of image further calculates the wavelet coefficient after each application threshold value with the size of noise-removed threshold value λ
wλValue；
Specific judgement is as follows：To dj(w) each coefficient carries out threshold process, if absolute coefficient is less than noise-removed threshold value λ, jth
The corresponding w of w-th of high frequency node in layerλValue is 0；Otherwise, wλValue is to subtract noise-removed threshold value, i.e.,：
Sign (w) expressions take coefficient dj(w) symbol；
Step 4.3 is directed to every layer, and the low frequency coefficient of high frequency coefficient and this layer after this layer of each denoising is reconstructed, has been obtained
Whole denoising image：
For the bottom, the high frequency coefficient coefficient corresponding with each low frequency node image of this layer after this layer of each denoising is carried out
Reconstruct, reconstruct obtains the node of last layer to the node of the bottom upwards, and so on, until all straton figures are reconstructed into and completely go
It makes an uproar image.
2. a kind of sediment sonar image processing method based on medium filtering and wavelet transformation as described in claim 1,
It is characterized in that, the step one, obtaining the image containing approximate Gaussian additive noise, the specific method is as follows：
First, the speckle noise of sonar original image obeys the Multiplicative noise model based on rayleigh distributed, is expressed as：I=RZ；
I is the signal containing spot observed from original image；R is the true not noisy image for it is expected to restore, Z be speckle noise with
Machine variable；
Then, multiplicative noise is become into approximate Gaussian additive noise form using logarithmic transformation：LnI=lnR+lnZ.
3. a kind of sediment sonar image processing method based on medium filtering and wavelet transformation as described in claim 1,
It is characterized in that, the step three, exponential transform is logarithmic transformation inverse operation.
4. a kind of sediment sonar image processing method based on medium filtering and wavelet transformation as described in claim 1,
It is characterized in that, in the step 4.1, the detailed process that image wavelet packet decomposition is carried out to smoothed image is：
First, suitable wavelet basis and Decomposition order are selected；Then it is decomposed since first layer and obtains two nodes, become second
Two tree nodes of layer, correspond to high frequency tree node and low frequency tree node respectively；Then two nodes of the second layer are decomposed simultaneously,
It obtains high frequency child node and low frequency child node again respectively, becomes four nodes of the third line；Similarly third layer is decomposed,
The 4th layer of eight nodes are obtained, and so on；
Corresponding two child nodes are respectively high frequency node and low frequency node under every layer of high frequency tree node；Every layer of low frequency tree node
Under corresponding child node be respectively high frequency node and low frequency node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810291591.0A CN108550121A (en) | 2018-03-30 | 2018-03-30 | A kind of sediment sonar image processing method based on medium filtering and wavelet transformation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810291591.0A CN108550121A (en) | 2018-03-30 | 2018-03-30 | A kind of sediment sonar image processing method based on medium filtering and wavelet transformation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108550121A true CN108550121A (en) | 2018-09-18 |
Family
ID=63514007
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810291591.0A Pending CN108550121A (en) | 2018-03-30 | 2018-03-30 | A kind of sediment sonar image processing method based on medium filtering and wavelet transformation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108550121A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109345475A (en) * | 2018-09-19 | 2019-02-15 | 长安大学 | A kind of unmanned aerial vehicle remote sensing mountain highway Image Fusion Filtering method |
| CN109685728A (en) * | 2018-11-30 | 2019-04-26 | 中南大学 | Digital image processing method based on local time-frequency domain conversation |
| CN110174281A (en) * | 2019-06-05 | 2019-08-27 | 北京博识创智科技发展有限公司 | A kind of electromechanical equipment fault diagnosis method and system |
| CN110348459A (en) * | 2019-06-28 | 2019-10-18 | 西安理工大学 | Based on multiple dimensioned quick covering blanket method sonar image fractal characteristic extracting method |
| CN110852959A (en) * | 2019-10-14 | 2020-02-28 | 江苏帝一集团有限公司 | Sonar image filtering method based on novel median filtering algorithm |
| CN110907811A (en) * | 2019-11-18 | 2020-03-24 | 广东欧文特电气有限公司 | Medium-voltage switch cabinet contact stroke measuring method |
| CN111476809A (en) * | 2020-04-08 | 2020-07-31 | 北京石油化工学院 | Side-scan sonar image target identification method |
| CN111833275A (en) * | 2020-07-20 | 2020-10-27 | 山东师范大学 | Image denoising method based on low-rank analysis |
| CN112157368A (en) * | 2020-09-24 | 2021-01-01 | 长春理工大学 | Laser non-penetration welding seam penetration nondestructive testing method |
| CN116912233A (en) * | 2023-09-04 | 2023-10-20 | 深圳市明亚顺科技有限公司 | Defect detection method, device, equipment and storage medium based on liquid crystal display screen |
| CN117058143A (en) * | 2023-10-12 | 2023-11-14 | 深圳市合成快捷电子科技有限公司 | Intelligent detection method and system for pins of circuit board |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101094312A (en) * | 2006-06-20 | 2007-12-26 | 西北工业大学 | Self-adapting method for filtering image with edge being retained |
| CN103426145A (en) * | 2012-05-23 | 2013-12-04 | 中国科学院声学研究所 | Synthetic aperture sonar speckle noise suppression method based on multiresolution analysis |
| US20140198992A1 (en) * | 2013-01-15 | 2014-07-17 | Apple Inc. | Linear Transform-Based Image Processing Techniques |
| CN105654434A (en) * | 2015-12-25 | 2016-06-08 | 浙江工业大学 | Medical ultrasonic image denoising method based on statistical model |
| CN106651788A (en) * | 2016-11-11 | 2017-05-10 | 深圳天珑无线科技有限公司 | Image denoising method |
-
2018
- 2018-03-30 CN CN201810291591.0A patent/CN108550121A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101094312A (en) * | 2006-06-20 | 2007-12-26 | 西北工业大学 | Self-adapting method for filtering image with edge being retained |
| CN103426145A (en) * | 2012-05-23 | 2013-12-04 | 中国科学院声学研究所 | Synthetic aperture sonar speckle noise suppression method based on multiresolution analysis |
| US20140198992A1 (en) * | 2013-01-15 | 2014-07-17 | Apple Inc. | Linear Transform-Based Image Processing Techniques |
| CN105654434A (en) * | 2015-12-25 | 2016-06-08 | 浙江工业大学 | Medical ultrasonic image denoising method based on statistical model |
| CN106651788A (en) * | 2016-11-11 | 2017-05-10 | 深圳天珑无线科技有限公司 | Image denoising method |
Non-Patent Citations (1)
| Title |
|---|
| 雷飞等: "改进小波软硬折衷法在水下图像去噪中的应用", 《计算机技术与发展》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109345475A (en) * | 2018-09-19 | 2019-02-15 | 长安大学 | A kind of unmanned aerial vehicle remote sensing mountain highway Image Fusion Filtering method |
| CN109345475B (en) * | 2018-09-19 | 2021-07-23 | 长安大学 | Unmanned aerial vehicle remote sensing mountain road image fusion filtering method |
| CN109685728B (en) * | 2018-11-30 | 2023-01-31 | 中南大学 | Digital image processing method based on local time-frequency domain transformation |
| CN109685728A (en) * | 2018-11-30 | 2019-04-26 | 中南大学 | Digital image processing method based on local time-frequency domain conversation |
| CN110174281A (en) * | 2019-06-05 | 2019-08-27 | 北京博识创智科技发展有限公司 | A kind of electromechanical equipment fault diagnosis method and system |
| CN110348459A (en) * | 2019-06-28 | 2019-10-18 | 西安理工大学 | Based on multiple dimensioned quick covering blanket method sonar image fractal characteristic extracting method |
| CN110852959A (en) * | 2019-10-14 | 2020-02-28 | 江苏帝一集团有限公司 | Sonar image filtering method based on novel median filtering algorithm |
| CN110907811A (en) * | 2019-11-18 | 2020-03-24 | 广东欧文特电气有限公司 | Medium-voltage switch cabinet contact stroke measuring method |
| CN111476809A (en) * | 2020-04-08 | 2020-07-31 | 北京石油化工学院 | Side-scan sonar image target identification method |
| CN111833275A (en) * | 2020-07-20 | 2020-10-27 | 山东师范大学 | Image denoising method based on low-rank analysis |
| CN112157368A (en) * | 2020-09-24 | 2021-01-01 | 长春理工大学 | Laser non-penetration welding seam penetration nondestructive testing method |
| CN112157368B (en) * | 2020-09-24 | 2021-11-23 | 长春理工大学 | Laser non-penetration welding seam penetration nondestructive testing method |
| CN116912233A (en) * | 2023-09-04 | 2023-10-20 | 深圳市明亚顺科技有限公司 | Defect detection method, device, equipment and storage medium based on liquid crystal display screen |
| CN116912233B (en) * | 2023-09-04 | 2024-02-13 | 深圳市明亚顺科技有限公司 | Defect detection method, device, equipment and storage medium based on liquid crystal display screen |
| CN117058143A (en) * | 2023-10-12 | 2023-11-14 | 深圳市合成快捷电子科技有限公司 | Intelligent detection method and system for pins of circuit board |
| CN117058143B (en) * | 2023-10-12 | 2024-01-26 | 深圳市合成快捷电子科技有限公司 | Intelligent detection method and system for pins of circuit board |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108550121A (en) | A kind of sediment sonar image processing method based on medium filtering and wavelet transformation | |
| CN102930512B (en) | Based on the underwater picture Enhancement Method of HSV color space in conjunction with Retinex | |
| CN101950414B (en) | Non-local mean de-noising method for natural image | |
| CN104657948B (en) | A kind of denoising of Laser Underwater image and Enhancement Method for marine exploration | |
| CN101944230B (en) | Multi-scale-based natural image non-local mean noise reduction method | |
| CN103854264A (en) | Improved threshold function-based wavelet transformation image denoising method | |
| CN101482617A (en) | Synthetic aperture radar image denoising method based on non-down sampling profile wave | |
| CN104504664B (en) | The automatic strengthening system of NSCT domains underwater picture based on human-eye visual characteristic and its method | |
| Zhang et al. | Image denoising using a neural network based non-linear filter in wavelet domain | |
| CN103295204A (en) | Image adaptive enhancement method based on non-subsampled contourlet transform | |
| Dhanushree et al. | Acoustic image denoising using various spatial filtering techniques | |
| CN113238190A (en) | Ground penetrating radar echo signal denoising method based on EMD combined wavelet threshold | |
| CN104732498B (en) | A kind of thresholded image denoising method based on non-downsampling Contourlet conversion | |
| CN108428221A (en) | A kind of neighborhood bivariate shrinkage function denoising method based on shearlet transformation | |
| CN102314675B (en) | Wavelet high-frequency-based Bayesian denoising method | |
| Zhang et al. | A reverberation noise suppression method of sonar image based on shearlet transform | |
| CN103426145A (en) | Synthetic aperture sonar speckle noise suppression method based on multiresolution analysis | |
| CN111652810A (en) | Image denoising method based on wavelet domain singular value differential model | |
| Singh et al. | Noise reduction in ultrasound images using wavelet and spatial filtering techniques | |
| CN113204051B (en) | Low-rank tensor seismic data denoising method based on variational modal decomposition | |
| CN103854258A (en) | Image denoising method based on Contourlet transformation self-adaptation direction threshold value | |
| Ni et al. | Speckle suppression for sar images based on adaptive shrinkage in contourlet domain | |
| Chen et al. | Research on sonar image denoising method based on fixed water area noise model | |
| Guo et al. | An Image Denoising Algorithm based on Kuwahara Filter | |
| Naik et al. | Underwater Acoustic Image Processing for Detection of Marine Debris |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180918 |
|
| RJ01 | Rejection of invention patent application after publication |