CN103425992B - A kind of seafloor sediment classification method and system based on synthetic aperture sonar picture - Google Patents
A kind of seafloor sediment classification method and system based on synthetic aperture sonar picture Download PDFInfo
- Publication number
- CN103425992B CN103425992B CN201210151963.2A CN201210151963A CN103425992B CN 103425992 B CN103425992 B CN 103425992B CN 201210151963 A CN201210151963 A CN 201210151963A CN 103425992 B CN103425992 B CN 103425992B
- Authority
- CN
- China
- Prior art keywords
- synthetic aperture
- image
- sediment
- gray level
- aperture sonar
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- 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
- G01S15/8902—Side-looking sonar
- G01S15/8904—Side-looking sonar using synthetic aperture techniques
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
Abstract
The present invention relates to a kind of seafloor sediment classification method based on synthetic aperture sonar picture, the method is used for the sediment of synthetic aperture sonar image is classified, and comprises：Step 101）Read in the to be sorted synthetic aperture sonar picture comprising sediment；Step 102）Calculate the gray level co-occurrence matrixes of synthetic aperture sonar picture；Step 103）Obtain the characteristic parameter that can react submarine bottom characteristic based on gray level co-occurrence matrixes, all parameter compositions can react the characteristic vector of textural characteristics；Step 104）Characteristic vector is compared with statistical information, and then completes seafloor sediment classification；Described statistical information is：According to the feature of synthetic aperture image difference sediment types, using the textural characteristics parameter in all types of sediment gray level co-occurrence matrixes, structural features vector；And all characteristic vectors are trained respectively, obtain the corresponding representative value in all sediment regions, and all representative values carried out storage constituting statistical information.
Description
Technical field
The present invention relates to seafloor sediment classification field is and in particular to a kind of sediment based on synthetic aperture sonar picture
Sorting technique and system.
Background technology
Sediment（Bottom sediment）Sort research be geophysical exploration, marine charting, ocean engineering etc. application
The basis in field, is all of great significance in terms of civilian and military.Ocean engineering, marine petroleum development and military neck
Submarine base in domain selects, sweeps the sediment types that mine-laying operation etc. all have to be understood that seabed.
Acoustic method remote measurement sediment type has an efficient work because of it, the features such as the data that obtains is continuous, abundant, becomes
A kind of rapid and reliable seafloor sediment classification method.Synthetic aperture sonar（SAS, Synthetic Aperture Sonar）It is
A kind of high-resolution Underwater Imaging sonar, it is possible to obtain high-quality subsea image data.Synthetic aperture is that one kind does not need length to connect
Receive the technology that battle array just can significantly improve azimuth resolution, synthetic aperture sonar is obtained by this technology and complicated imaging algorithm
To all very high image of azimuth resolution and range resolution, thus research field has very high value under water.With
Common sonar is compared, and synthetic aperture sonar improves array aperture by the linear movement of basic matrix, from principle for, synthesize hole
The resolution of footpath sonar image is all unrelated with operating frequency and operating distance, therefore can be with less sonar transducer array and relatively low
Operating frequency meets closely simultaneously and remote detects needs.Synthetic aperture sonar can reflect the ground in seabed well
Shape, landforms and textural characteristics, are therefore analyzed to sonar image receiving significant attention to realize seafloor sediment classification.
Sonar echo intensity is a complicated physical quantity, and the many factors such as its same tranmitting frequency, sediment types, glancing angle have
Close, different sediment types are likely to be of identical echo strength, and the gray value only by means of prior art carries out Seafloor Classification
It is inaccurate.But research finds, the texture image that different substrates is presented in sonar image is different, and texture is seabed
The direct reaction of coarse surface structure degree, therefore Seafloor Classification can be carried out using it.For texture difference such as rock, sand and mud
Obvious sonar image, with the naked eye just can significantly distinguish；Less for the texture difference such as mud and clay
Sonar image then directly cannot be differentiated by naked eyes.
Content of the invention
It is an object of the invention to, for overcoming prior art to only rely on synthetic aperture sonar when carrying out seafloor sediment classification
The not high technical problem of the nicety of grading that the half-tone information of image leads to, thus provide a kind of based on synthetic aperture sonar as
Seafloor sediment classification method and system.
For realizing above-mentioned technical purpose, the invention provides a kind of seafloor sediment classification based on synthetic aperture sonar picture
Method, the method is used for the sediment of synthetic aperture sonar image is classified, and methods described comprises：
Step 101）Read in the to be sorted synthetic aperture sonar picture comprising sediment；
Step 102）Calculate the gray level co-occurrence matrixes of synthetic aperture sonar picture；
Step 103）Obtain the characteristic parameter that can react submarine bottom characteristic, all parameter groups based on gray level co-occurrence matrixes
One-tenth can react the characteristic vector of textural characteristics；
Step 104）Characteristic vector is compared with statistical information, and then completes seafloor sediment classification；
Wherein, described statistical information is：According to the feature of synthetic aperture image difference sediment types, make full use of all
Textural characteristics parameter in the sediment gray level co-occurrence matrixes of type, structural features vector；And to all characteristic vectors respectively
It is trained, obtains the corresponding representative value in all sediment regions, and all representative values are carried out storage constituting statistical information,
And described representative value is obtained by the characteristic vector of representative region known to analysis, described representative region including but not limited to：Husky,
Gravel and sludge region.
In technique scheme, described gray level co-occurrence matrixes are adopted and are obtained with the following method：
Take synthetic aperture sonar picture（N×N）Middle any point（x,y）And deviate its another point（x+a,y+b）If, should
Point to gray value be（g1,g2）；
Order（x,y）Entire image moves, then can obtain various（g1,g2）Value, if the series of gray scale is k, then（g1,
g2）Compound mode have k2Kind；
For entire image, count each（g1,g2）The number of times that value occurs, is then arranged in a square formation, then
With（g1,g2）They are normalized to the probability P (g of appearance by the total degree occurring1,g2), thus obtain the gray scale symbiosis of image
Matrix；
Wherein, work as a=1, during b=0, pixel to being level, that is, correspond to SAS image distance to textural characteristics；Work as a=0, b
When=1, pixel to being vertical, that is, correspond to SAS image orientation to textural characteristics；Work as a=1, during b=1, pixel is to being along right right
Angle, that is, correspond to the right diagonally opposed textural characteristics of SAS image；Work as a=1, during b=-1, pixel to being left diagonal, that is, corresponds to
The left diagonally opposed textural characteristics of SAS image；
Wherein, above (x, y) represents the coordinate of the pixel in image, and a and b denotation coordination interval；g1,g2Represent picture
The gray value of vegetarian refreshments pair.
In technique scheme, described textural characteristics parameter comprises：Angular second moment/energy, contrast, dependency and entropy.
In technique scheme, described step 102）With step 103）Between also comprise：Calculate the two of gray level co-occurrence matrixes
The step of secondary statistic.
Based on said method present invention also offers a kind of seafloor sediment classification system based on synthetic aperture sonar picture,
This system is used for the sediment of synthetic aperture sonar image is classified, and described system comprises：
Input module, for reading in the to be sorted synthetic aperture sonar picture comprising sediment；
First processing module, for calculating the gray level co-occurrence matrixes of synthetic aperture sonar picture；
Second processing module, for obtaining, based on gray level co-occurrence matrixes, the characteristic parameter that submarine bottom characteristic can be reacted,
All parameters composition can react textural characteristics characteristic vector；With
Multilevel iudge module, for comparing characteristic vector with statistical data library module, and then completes seafloor sediment classification；
Wherein, described staqtistical data base is used for stored statistical information, and this statistical information is：According to synthetic aperture image not
With the feature of sediment types, make full use of the textural characteristics parameter in all types of sediment gray level co-occurrence matrixes, construction
Characteristic vector；And all characteristic vectors are trained respectively, obtain the corresponding representative value in all sediment regions, and by institute
There is representative value to carry out storage and constitute statistical information.
In technique scheme, described first processing module comprises further：Input module and processing module；
Described processing module further comprises：
Starting point arranges submodule, is used for taking image（N×N）Middle any point（x,y）And deviate its another point（x+a,
y+b）If, this to gray value be（g1,g2）；
Mobile setting submodule, for making（x,y）Entire image moves, then can obtain various（g1,g2）Value, if
The series of gray scale is k, then（g1,g2）Compound mode have k2Kind；
Gray level co-occurrence matrixes output sub-module, for for entire image, counting each（g1,g2）It is secondary that value occurs
Number, is then arranged in a square formation, then uses（g1,g2）They are normalized to the probability P (g of appearance by the total degree occurring1,g2),
Thus obtain the gray level co-occurrence matrixes of image；
Wherein, work as a=1, during b=0, pixel to being level, that is, correspond to SAS image distance to textural characteristics；Work as a=0, b
When=1, pixel to being vertical, that is, correspond to SAS image orientation to textural characteristics；Work as a=1, during b=1, pixel is to being along right right
Angle, that is, correspond to the right diagonally opposed textural characteristics of SAS image；Work as a=1, during b=-1, pixel to being left diagonal, that is, corresponds to
The left diagonally opposed textural characteristics of SAS image.
In technique scheme, described textural characteristics parameter comprises：Angular second moment/energy, contrast, dependency and entropy.
In technique scheme, also comprise second degree statistics between described first processing module and Second processing module and calculate
Module, this module calculates the second degree statistics of gray level co-occurrence matrixes, and inputs to Second processing module.
Compared with prior art, the present invention's it is a technical advantage that：
Technical scheme can be good at the sediment of SAS image is classified, thus solving existing
Technology is difficult to the technical problem that the sediment difficult to SAS image is classified.
Brief description
The flow chart of the seafloor sediment classification method based on synthetic aperture sonar picture that Fig. 1 present invention provides；
What Fig. 2 embodiment of the present invention adopted waits for the original SAS figure comprising silt mixing matter of seafloor sediment classification
Picture；
The original SAS image comprising mud substrate waiting for seafloor sediment classification that Fig. 3 embodiment of the present invention adopts；
The original SAS image comprising husky substrate waiting for seafloor sediment classification that Fig. 4 embodiment of the present invention adopts；
Fig. 5-a be the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate energy feature parameter with distance to skew
The situation cartogram of change；
Fig. 5-b be the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate correlative character parameter with distance to skew
And the situation cartogram changing；
Fig. 5-c be the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate contrast metric parameter with distance to skew
And the situation cartogram changing；
Fig. 5-d is that the entropy characteristic parameter of the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate becomes to skew with distance
The situation cartogram changed；
Fig. 6-a is the energy feature parameter of the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate with orientation skew
The situation cartogram of change；
Fig. 6-b is that the correlative character parameter of the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate offsets with orientation
And the situation cartogram changing；
Fig. 6-c is that the contrast metric parameter of the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate offsets with orientation
And the situation cartogram changing；
Fig. 6-d is that the entropy characteristic parameter of the gray level co-occurrence matrixes of mud substrate of the present invention and husky substrate offsets with orientation and becomes
The situation cartogram changed.
Specific embodiment
The present invention will be described in detail with specific embodiment below in conjunction with the accompanying drawings.
In order to solve the above problems, it is an object of the invention to provide a kind of synthetic aperture sonar that is based on is as sediment
The new method of classification.We utilize the eigenvalue of gray level co-occurrence matrixes to solve the problems, such as Seafloor Classification, described synthetic aperture sonar
Seafloor Classification method concrete steps include：
Step 1：Read in original synthetic aperture sonar picture
Step 2：Calculate the gray level co-occurrence matrixes of original image
Step 3：Calculate the eigenvalue of gray level co-occurrence matrixes
Step 4：The eigenvalue of analysis gray level co-occurrence matrixes
Step 5：Eigenvalue using gray level co-occurrence matrixes carries out target detection
The feature of SAS image texture aspect can be described using gray level co-occurrence matrixes, existed by calculating gray level co-occurrence matrixes
Orientation and distance to energy, dependency, contrast and entropy, and structural features vector, can be to bottoms different in SAS image
Classified in matter region.From experimental result as can be seen that SAS image has abundant texture information, texture information can be based on
Realize the differentiation of husky substrate and mud substrate.
1st, analyzing image texture
The expression of textural characteristics and analysis generally have statistic law, Structure Method and modelling, and are based on gray level co-occurrence matrixes
（GLCM）The method of texture feature extraction is a kind of typical statistical analysis technique, is closed using gray level co-occurrence matrixes herein
Become expression and the analysis of aperture sonar image.GLCM texture blending method has stronger adaptability and robustness, in recent years
It has been increasingly being used for detection and the sort research of image.
Haralick in 1973 first propose gray level co-occurrence matrixes, surface its be better than ray level run-length matrices method and spectrum side
Method, is a kind of wide variety of texture statistics method and texture e measurement technology.P.P.Ohanian is given to several textures within 1992
The comparative result of e measurement technology, and proved according to experiment：It is used in the feature realizing Texture classification at four kinds, based on gray scale altogether
The statistical nature of raw matrix is better than FRACTAL DIMENSION, Markov model and Gabor filter model.
1.1st, the generation of gray level co-occurrence matrixes
Grey level histogram is that have, to single pixel in image, the result that certain gray scale is counted, and gray level co-occurrence matrixes
It is then that the situation that two pixels pushing away and keeping certain distance in image are respectively provided with certain gray scale carries out counting obtaining.Take image（N×
N）Middle any point（x,y）And deviate its another point（x+a,y+b）If, this to gray value be（g1,g2）.Order（x,y）?
Move in entire image, then can obtain various（g1,g2）Value, if the series of gray scale is k, then（g1,g2）Compound mode have
k2Kind.For entire image, count each（g1,g2）The number of times that value occurs, is then arranged in a square formation, then uses
（g1,g2）They are normalized to the probability P (g of appearance by the total degree occurring1,g2), thus obtain the gray scale symbiosis square of image
Battle array.Take different combinations of values apart from difference value (a, b), the gray level co-occurrence matrixes of different situations can be obtained.
Work as a=1, during b=0, pixel to being level, that is, correspond to SAS image distance to textural characteristics；Work as a=0, during b=1,
Pixel to being vertical, that is, correspond to SAS image orientation to textural characteristics；Work as a=1, during b=1, pixel to being along right diagonal,
Correspond to the right diagonally opposed textural characteristics of SAS image；Work as a=1, during b=-1, pixel to being left diagonal, that is, corresponds to SAS image
Left diagonally opposed textural characteristics；
1.2nd, the feature of gray level co-occurrence matrixes
Gray level co-occurrence matrixes reaction be gradation of image with regard to direction, adjacent spaces, amplitude of variation integrated information.Pass through
Gray level co-occurrence matrixes can be with the local pattern of analysis of the image and queueing discipline etc., in order to more intuitively use gray scale symbiosis square
Battle array description texture information, typically not directly using gray level co-occurrence matrixes, but calculates second degree statistics on its basis.
Haralick et al. defines 14 gray level co-occurrence matrixes characteristic parameters for texture analysiss, and Ulaby et al. research finds,
In 14 textural characteristics based on GLCM, only 4 characteristic quantities are incoherent, and this four characteristic quantities are both easy to calculate, and energy
Provide higher nicety of grading, typically adopt following four characteristic parameter to extract the textural characteristics of image.
First, angular second moment/energy（ASM）
Angular second moment is the quadratic sum of each element of gray level co-occurrence matrixes, also known as energy, and reflection gradation of image is evenly distributed
Degree and texture fineness.If all values of gray level co-occurrence matrixes are all equal, ASM value is less；If some of them value
Larger, other values are less, then ASM value is larger.When ASM value is larger, texture is thick, and energy is big；Conversely, working as ASM value hour, stricture of vagina
Reason is thin, and energy is little.
2nd, contrast（CON）
Contrast is the moment of inertia near gray level co-occurrence matrixes leading diagonal, the distribution situation of metric matrix intermediate value and image
Localized variation situation, the reflection readability of the image and depth degree of texture rill.If the rill of texture is deeper,
CON is larger, and effect is clear；Whereas if texture rill is shallower, then CON is less, and effect obscures.Gray scale difference is that contrast is big
To more, then CON is bigger, and that is, bigger away from cornerwise element value in gray level co-occurrence matrixes, CON is bigger for pixel.
3rd, dependency（COR）
Wherein,
Relativity measurement be gray level correlative matrix element be expert at or column direction on similarity degree, its size can be anti-
Reflect the local correlations of image.When gray level co-occurrence matrixes element value is uniformly equal, COR is larger；On the contrary, gray level co-occurrence matrixes are worked as
When pixel value difference is larger, COR is less.If image has the texture in certain direction, gray level co-occurrence matrixes in this direction
COR value larger.
4th, entropy（ENT）
Entropy can be used as the eigenvalue of tolerance amount of image information, and texture information falls within image information, be a kind of with
The tolerance of machine.When in gray level co-occurrence matrixes, all elements have the randomness of maximum, in gray level co-occurrence matrixes, all values are almost
When equal, in gray level co-occurrence matrixes during element dispersed and distributed, entropy is larger.Entropy represent in image the non-uniform degree of texture or
Person's complexity, if there is no any texture in image, the almost nil matrix of gray level co-occurrence matrixes；If texture is multiple in image
Miscellaneous, then entropy is larger.
2nd, SAS image sediment analysis
SAS image contains abundant textural characteristics, and gray level co-occurrence matrixes have abundant characteristic parameter, can be from difference
Angle careful portraying is carried out to texture.Below with the characteristic quantity of gray level co-occurrence matrixes, respectively to fluctuating substrate region and all
The texture features in even substrate region are compared analysis.
Fig. 2 is the synthetic aperture sonar picture that the examination of certain lake obtains, and water-bed region is made up of the farmland flooded and river course.As
In figure mark is described, and this width SAS image top is the terraced fields flooding, and sediment types are mud, and bottom is river course, and sediment types are
Husky.Now intercept terraced fields respectively and the representative region in river course to be analyzed studying as mud substrate and husky substrate, white edge in such as Fig. 2
Shown.Fig. 3 and Fig. 4 is mud substrate and the SAS image of husky substrate respectively, and the texture analysiss being carried out herein are based on both classes
The SAS image of type.
SAS image is imaged to using pulse compression principle in distance, and then adopts synthetic aperture principle to be imaged in orientation,
Therefore, the gray level co-occurrence matrixes of SAS image have different characteristics in distance to orientation, and that is, the texture of SAS image is in side
Position to distance to taking on a different character.The size in the mud substrate region intercepting herein and husky substrate region is 512 × 512
Pixel, the skew span of gray level co-occurrence matrixes is [180], calculates the energy under various drift condition respectively（ASM）, phase
Guan Xing（COR）, contrast（CON）And entropy（ENT）.
Fig. 5-a, 5-b, 5-c and 5-d are the situation that the characteristic quantity of gray level co-occurrence matrixes changes to skew with distance.Mud
The energy in substrate region will be noticeably greater than the energy in husky substrate region, and the energy in two kinds of substrate regions not with distance to inclined
Move increase and change.The texture in this explanation mud substrate region is thicker, and energy is big；And the texture in sand substrate region is thinner, energy
Little.
In terms of dependency, mud substrate is consistent with the variation tendency of husky substrate, and the dependency in mud substrate region is less than
Husky substrate region.Mud substrate and husky substrate be all distance to offset less when, dependency is higher, then as distance to skew
Increase and quickly fall near zero.Relativity measurement be gray level co-occurrence matrixes element row or column similarity degree, permissible
The local similarity of reflection image.The texture structure unit of this explanation mud substrate and husky substrate is all less, with distance to skew
Increase, dependency reduces rapidly, and texture features also just rapidly disappear.
The contrast metric of mud substrate and husky substrate is just contrary with energy feature, and the contrast in husky substrate region is significantly big
In the contrast in mud substrate region, and the contrast in two kinds of substrate regions does not change to the increase of skew with distance.Right
It is the moment of inertia near gray level co-occurrence matrixes leading diagonal than degree, the distribution situation of metric matrix intermediate value and the localized variation of image
Situation, the reflection readability of image and the depth degree of texture rill.That is the texture rill in husky substrate region is deeper,
Effect is clear；And the texture rill in mud substrate region is shallower, effect obscures.
When entropy, the entropy in mud substrate region is relatively stable, becomes to the increase of skew little with distance
Change, and remain near 0.9；And the entropy in sand substrate region then changes relatively acutely, and it is maintained near 0.5.Entropy
Value can represent the non-uniform degree of texture or complexity in image, therefore the randomness in mud substrate region is than husky substrate
Greatly, show significant Statistic Texture.
As to the comparison offseting, we analyze each characteristic quantity that orientation offsets gray level co-occurrence matrixes, figure with distance
6-a, 6-b, 6-c and 6-d are the situation that the characteristic quantity of gray level co-occurrence matrixes offsets with orientation and changes.For gray scale symbiosis
The energy of matrix, dependency, contrast and entropy, mud substrate region and husky substrate region are in distance to skew and orientation skew
Almost there is no difference, this also illustrates the distance of SAS image to as broad as long with the gray level co-occurrence matrixes of orientation in the case of two kinds,
In the characteristic quantity of selective discrimination zones of different, only take a direction, or the meansigma methodss taking both direction.
Analyzed according to above, for the feature of description SAS image difference sediment types, make full use of in gray level co-occurrence matrixes
The features such as energy, dependency, contrast and entropy, structural features vector v={ ASM, COR, CON, ENT }.Wherein ASM, COR,
CON, ENT are respectively energy, dependency, contrast and the entropy of gray level co-occurrence matrixes.By being trained to characteristic vector, obtain
To the representative value of husky substrate and mud substrate region, then can be used for the classification in substrate region.
In a word, the present invention proposes a kind of seafloor sediment classification method being applied to synthetic aperture sonar picture.Different
The texture image that substrate is presented in sonar image is different, and texture is the directly anti-of seabed coarse surface structure degree
Should, therefore Seafloor Classification can be carried out using it.Synthetic aperture sonar can reflect that landform, landforms and the texture in seabed is special well
Levy, therefore sonar image is analyzed receiving significant attention to realize seafloor sediment classification.This method adopts gray scale symbiosis square
The feature of battle array description SAS image texture aspect, by calculate gray level co-occurrence matrixes orientation and distance to energy, correlation
Property, contrast and entropy, and structural features vector, are classified in substrate regions different in SAS image.Permissible from experimental result
Find out, SAS image has abundant texture information, the differentiation of husky substrate and mud substrate can be realized based on texture information.
It should be noted that the embodiment of present invention described above and not limit.Those skilled in the art should
Work as understanding, any modification to technical solution of the present invention or equivalent substitute are without departure from the spirit of technical solution of the present invention and model
Enclose, it all should be covered in scope of the presently claimed invention.
Claims (8)
1. a kind of seafloor sediment classification method based on synthetic aperture sonar picture, the method is used for synthetic aperture sonar image
Sediment classified, methods described comprises：
Step 101) read in the to be sorted synthetic aperture sonar picture comprising sediment；
Step 102) calculate synthetic aperture sonar picture gray level co-occurrence matrixes；
Step 103) obtained based on gray level co-occurrence matrixes and can react the characteristic parameter of submarine bottom characteristic, all parameters form energy
Enough react the characteristic vector of textural characteristics；
Step 104) characteristic vector is compared with statistical information, and then complete seafloor sediment classification；
Wherein, described statistical information is：According to the feature of synthetic aperture image difference sediment types, make full use of all types
Sediment gray level co-occurrence matrixes in textural characteristics parameter, structural features vector；And all characteristic vectors are carried out respectively
Training, obtains the corresponding representative value in all sediment regions, and all representative values are carried out storage composition statistical information, and institute
State representative value by analyze representative region characteristic vector obtain, described representative region including but not limited to：Sand, gravel and mud
Region.
2. the seafloor sediment classification method based on synthetic aperture sonar picture according to claim 1 is it is characterised in that institute
State gray level co-occurrence matrixes and adopt and obtain with the following method：
Take any point (x, y) and another point (x+a, y+b) deviateing it in synthetic aperture sonar picture (N × N), if this point is right
Gray value be (g1,g2)；
Make (x, y) to move in entire image, then can obtain various (g1,g2) value, if the series of gray scale is k, then (g1,g2)
Compound mode has k2Kind；
For entire image, count each (g1,g2) value occur number of times, be then arranged in a square formation, then use
(g1,g2) they are normalized to the probability P (g that occurs by the total degree that occurs1,g2), thus obtain the gray scale symbiosis square of image
Battle array；
Wherein, work as a=1, during b=0, pixel to being level, that is, correspond to SAS image distance to textural characteristics；Work as a=0, b
When=1, pixel to being vertical, that is, correspond to SAS image orientation to textural characteristics；Work as a=1, during b=1, pixel is to being along right
Diagonal, that is, correspond to the right diagonally opposed textural characteristics of SAS image；Work as a=1, during b=-1, pixel is to being left diagonal, that is, right
Answer the left diagonally opposed textural characteristics of SAS image；
Wherein, above (x, y) represents the coordinate of the pixel in image, and a and b denotation coordination interval；g1,g2Represent pixel
To gray value.
3. the seafloor sediment classification method based on synthetic aperture sonar picture according to claim 2 is it is characterised in that institute
State textural characteristics parameter to comprise：Angular second moment/energy, contrast, dependency and entropy.
4. the seafloor sediment classification method based on synthetic aperture sonar picture according to claim 2 is it is characterised in that institute
State step 102) and step 103) between also comprise：The step calculating the second degree statistics of gray level co-occurrence matrixes.
5. a kind of seafloor sediment classification system based on synthetic aperture sonar picture, this system is used for synthetic aperture sonar image
Sediment classified, described system comprises：
Input module, for reading in the to be sorted synthetic aperture sonar picture comprising sediment；
First processing module, for calculating the gray level co-occurrence matrixes of synthetic aperture sonar picture；
Second processing module, for obtaining, based on gray level co-occurrence matrixes, the characteristic parameter that can react submarine bottom characteristic, owns
Parameter composition can react textural characteristics characteristic vector；With
Multilevel iudge module, for comparing characteristic vector with statistical data library module, and then completes seafloor sediment classification；
Wherein, described staqtistical data base is used for stored statistical information, and this statistical information is：According to synthetic aperture image difference bottom
The feature of matter type, makes full use of the textural characteristics parameter in all types of sediment gray level co-occurrence matrixes, structural features
Vector；And all characteristic vectors are trained respectively, obtain the corresponding representative value in all sediment regions, and by all allusion quotations
Offset carries out storage and constitutes statistical information.
6. the seafloor sediment classification system based on synthetic aperture sonar picture according to claim 5 is it is characterised in that institute
State first processing module to comprise further：Input module and processing module；
Described processing module further comprises：
Starting point arranges submodule, is used for taking any point (x, y) and another point (x+a, the y+ that deviate it in image (N × N)
B), if this to gray value be (g1,g2)；
Mobile setting submodule, for making (x, y) to move in entire image, then can obtain various (g1,g2) value, if gray scale
Series be k, then (g1,g2) compound mode have k2Kind；
Gray level co-occurrence matrixes output sub-module, for for entire image, counting each (g1,g2) value occur number of times, so
After be arranged in a square formation, then with (g1,g2) they are normalized to the probability P (g that occurs by the total degree that occurs1,g2), so
Just obtain the gray level co-occurrence matrixes of image；
Wherein, work as a=1, during b=0, pixel to being level, that is, correspond to SAS image distance to textural characteristics；Work as a=0, b
When=1, pixel to being vertical, that is, correspond to SAS image orientation to textural characteristics；Work as a=1, during b=1, pixel is to being along right
Diagonal, that is, correspond to the right diagonally opposed textural characteristics of SAS image；Work as a=1, during b=-1, pixel is to being left diagonal, that is, right
Answer the left diagonally opposed textural characteristics of SAS image.
7. the seafloor sediment classification system based on synthetic aperture sonar picture according to claim 6 is it is characterised in that institute
State textural characteristics parameter to comprise：Angular second moment/energy, contrast, dependency and entropy.
8. the seafloor sediment classification system based on synthetic aperture sonar picture according to claim 5 is it is characterised in that institute
State and also comprise second degree statistics computing module between first processing module and Second processing module, this module calculates gray scale symbiosis square
The second degree statistics of battle array, and input to Second processing module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210151963.2A CN103425992B (en) | 2012-05-16 | 2012-05-16 | A kind of seafloor sediment classification method and system based on synthetic aperture sonar picture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210151963.2A CN103425992B (en) | 2012-05-16 | 2012-05-16 | A kind of seafloor sediment classification method and system based on synthetic aperture sonar picture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103425992A CN103425992A (en) | 2013-12-04 |
| CN103425992B true CN103425992B (en) | 2017-03-08 |
Family
ID=49650702
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210151963.2A Expired - Fee Related CN103425992B (en) | 2012-05-16 | 2012-05-16 | A kind of seafloor sediment classification method and system based on synthetic aperture sonar picture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103425992B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104732190B (en) * | 2013-12-20 | 2017-12-12 | 中国科学院声学研究所 | A kind of synthetic aperture sonar object detection method based on orthogonal texture correlation analysis |
| CN106897687A (en) * | 2017-02-21 | 2017-06-27 | 中海石油环保服务（天津）有限公司 | A kind of oil-containing deposit automatic identifying method and system based on sonar technique |
| CN107526069A (en) * | 2017-07-27 | 2017-12-29 | 天津大学 | It is a kind of to utilize the inverting of forward scattering intensity and the method for classification seabed surface deposits |
| CN108363076B (en) * | 2018-01-16 | 2021-09-10 | 北京理工大学 | System and method for estimating instantaneous frequency of man-made interference based on short-time Renyi entropy |
| CN108388899B (en) * | 2018-01-29 | 2022-03-01 | 哈尔滨工程大学 | Underwater sound image feature extraction method based on fusion of texture features and shape features |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034109A (en) * | 2009-12-08 | 2011-04-27 | 中国科学院声学研究所 | Statistical property-based method for synthetic aperture sonar target detection |
| CN103426156A (en) * | 2012-05-15 | 2013-12-04 | 中国科学院声学研究所 | SAS image segmentation method and system based on SVM classifier |
-
2012
- 2012-05-16 CN CN201210151963.2A patent/CN103425992B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034109A (en) * | 2009-12-08 | 2011-04-27 | 中国科学院声学研究所 | Statistical property-based method for synthetic aperture sonar target detection |
| CN103426156A (en) * | 2012-05-15 | 2013-12-04 | 中国科学院声学研究所 | SAS image segmentation method and system based on SVM classifier |
Non-Patent Citations (1)
| Title |
|---|
| 第二讲 合成孔径声纳成像及其研究进展;张春华等;《物理》;20060531;第35卷(第5期);第408-413页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103425992A (en) | 2013-12-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108038445B (en) | SAR automatic target identification method based on multi-view deep learning framework | |
| CN103425992B (en) | A kind of seafloor sediment classification method and system based on synthetic aperture sonar picture | |
| Ji et al. | Seabed sediment classification using multibeam backscatter data based on the selecting optimal random forest model | |
| CN102609701B (en) | Remote sensing detection method based on optimal scale for high-resolution SAR (synthetic aperture radar) | |
| CN103425990B (en) | A kind of synthetic aperture sonar object detection method based on textural characteristics | |
| CN104463193B (en) | Classification of Polarimetric SAR Image method based on the sparse ICA of depth | |
| CN103426156A (en) | SAS image segmentation method and system based on SVM classifier | |
| CN107944370A (en) | Classification of Polarimetric SAR Image method based on DCCGAN models | |
| CN103366371B (en) | Based on K distribution and the SAR image segmentation method of textural characteristics | |
| CN109359533B (en) | Coastline extraction method based on multiband remote sensing image | |
| CN111709487B (en) | Underwater multi-source acoustic image substrate classification method and system based on decision-level fusion | |
| CN106650744A (en) | Image object co-segmentation method guided by local shape migration | |
| CN116563726A (en) | Remote sensing image ship target detection method based on convolutional neural network | |
| Yuan et al. | Capturing small objects and edges information for cross-sensor and cross-region land cover semantic segmentation in arid areas | |
| CN102938148A (en) | High-spectrum image texture analysis method based on V-GLCM (Gray Level Co-occurrence Matrix) | |
| Kekre et al. | SAR image segmentation using vector quantization technique on entropy images | |
| CN104732190B (en) | A kind of synthetic aperture sonar object detection method based on orthogonal texture correlation analysis | |
| CN108388899A (en) | A kind of Underwater Image feature extracting method blended based on textural characteristics and shape feature | |
| CN102737232B (en) | Cleavage cell recognition method | |
| Purkis et al. | Visualizing lateral anisotropy in modern carbonates | |
| Manashti et al. | PSDNet: Determination of Particle Size Distributions Using Synthetic Soil Images and Convolutional Neural Networks | |
| Wang et al. | Underwater Terrain Image Stitching Based on Spatial Gradient Feature Block. | |
| Guo et al. | Object-Level Hybrid Spatiotemporal Fusion: Reaching a Better Trade-Off Among Spectral Accuracy, Spatial Accuracy and Efficiency | |
| Pharsook et al. | The texture classification using the fusion of decisions from different texture classifiers | |
| CN116721264A (en) | SAR-to-optical image translation method, system, device and medium based on improved CGAN |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170308 Termination date: 20210516 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |