CN104657736A - Active contour-based sonar image mine target recognition method - Google Patents
Active contour-based sonar image mine target recognition method Download PDFInfo
- Publication number
- CN104657736A CN104657736A CN201310587459.1A CN201310587459A CN104657736A CN 104657736 A CN104657736 A CN 104657736A CN 201310587459 A CN201310587459 A CN 201310587459A CN 104657736 A CN104657736 A CN 104657736A
- Authority
- CN
- China
- Prior art keywords
- target
- active contour
- sonar image
- level set
- hyperellipse
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000004069 differentiation Effects 0.000 claims description 7
- 238000011478 gradient descent method Methods 0.000 claims description 7
- 238000009795 derivation Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 description 31
- 230000008569 process Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Landscapes
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
Abstract
The invention relates to the field of automatic target recognition of images, in particular to a sonar image mine target recognition method. According to the active contour model-based sonar image mine target recognition method, a hyperelliptic restriction active contour model and a multiphase level set shape restriction active contour model are provided. Compared with the common target recognition method, the sonar image mine target recognition method has the advantages of satisfying the requirements of mine targets in large-noise sonar images and obtaining the correct contours of the mine targets under the condition that the mine target images are fuzzy.
Description
Technical Field
The invention relates to the field of automatic target identification of images, in particular to a method for identifying a submarine mine target of a sonar image.
Background
Sonar is underwater remote sensing equipment for surveying and mapping submarine geomorphology, and is a powerful tool for surveying and mapping the submarine. The working principle is as follows: the sound pulse is emitted from the transducer and propagates to a far distance in the form of spherical wave, the backscattering wave generated at the sea bottom is successively transmitted back to the transducer, and the echo signals are displayed by the graphic equipment to form an image reflecting the geological characteristics of the sea bottom.
As shown in fig. 1, the sonar image is divided into 3 regions by the gray scale distribution: an echo zone (target zone), an acoustic shadow zone, and a background zone. The echo of the echo area is stronger and is a brighter area in the image, the echo of the background area is weaker, the echo area is darker in the image, and the non-echo area forms an acoustic shadow area which is the darkest part in the image. The sonar image has the following characteristics: 1. the echo intensity and structure are strongly influenced by many factors. 2. Sonar images generally have high noise, poor edges, relatively serious shape distortion and uneven image intensity. 3. Generally, the characteristics of the sound shadow area are stable.
The object automatic identification technology of sonar images applies digital image processing and pattern recognition technology to the field of underwater sound, and the research field is paid attention and developed in recent years. Due to the complexity of the underwater imaging environment and the difference of parameters such as resolution of different imaging sonar systems, the image formed by the target is greatly influenced by the characteristics of the target. Therefore, the automatic target recognition of the underwater acoustic image is also focused on theoretical research and has a certain distance from practical application. The recognition algorithm applied to sonar images at present mainly comprises the following steps of integrating relevant documents at home and abroad: a matched filter algorithm, a Moore neighborhood tracking algorithm, a fractal-based method, a level set segmentation algorithm, a Markov random field model and a sonar image segmentation algorithm based on spectral clustering. The research aiming at sonar image mine target identification is published less, and no identification algorithm can well identify the mine so far.
Because the sonar image has the characteristics of serious noise pollution, uneven intensity and fuzzy target boundary, a good extraction effect is difficult to achieve by adopting a general method. The active contour method is a global-based algorithm, which is less affected by noise, and therefore, an active contour model-based method is studied herein to identify a target. Active contour models are mainly divided into two categories: edge-based and region-based. Because the edges of the targets in the sonar images are fuzzy, if an edge-based contour evolution model is adopted, the edge leakage phenomenon is likely to occur in the evolution process. Therefore, a region-based contour evolution model is adopted: the Chan-Vese model is called C-V model for short. On the basis of the C-V model, the shape feature of the mine target and the gray feature of the sonar image are added to obtain the shape-preserving active contour model identification method based on the multiphase level set.
Disclosure of Invention
In order to solve the influence of characteristics such as large noise of a sonar image, fuzzy target edge and the like on the recognition of the mine, the invention provides a multi-phase level set active contour model based on the super-elliptical constraint by adopting the evolution thought of a Chan-Vese model according to the shape characteristics and the gray characteristics of a mine target in the sonar image.
The technical scheme adopted by the invention for realizing the purpose is as follows: a sonar image mine target identification method based on active contour comprises the following steps:
introducing a super-elliptic shape constrained level set function into the active contour model according to the shape characteristics of the mine target;
according to the characteristics of a target sound shadow area in a sonar image, a multiphase level set function is introduced into an active contour model, so that the sonar image is divided into three areas: a background area, a target area and a sound-shadow area;
taking the minimum difference between the gray mean values of the target area and the sound-shadow area as an evolution target, namely the minimum energy function of the sound-shadow area of the mine target;
and solving the super-ellipse-constrained active contour model of the multiphase level set by using a gradient descent method to obtain a parameter evolution equation of the active contour model.
The level set function of the hyperelliptic shape constraint is:
wherein (x)0,y0) The coordinate of the center of the hyperellipse is shown as a, b are respectively the major axis and the minor axis of the hyperellipse, and theta is the rotation angle of the hyperellipse.
Another level set function introduced by the multi-phase level set function is:
wherein (x)0,y0) The coordinate of the center of the hyperellipse is shown as a, b are respectively the major axis and the minor axis of the hyperellipse, and theta is the rotation angle of the hyperellipse.
The three regions are specifically divided into: when in useWhen it represents the background area of the sonar image, whenAnd isWhen representing the target area of the sonar image, whenAnd isTime represents the sonographic area of the sonar image.
The energy function of the sound shadow area of the mine target is as follows:
wherein, c11Is the mean value of the gray levels of the target region, c12Mean value of gray levels of the sound and shadow region, c2Is the mean value of the gray levels of the background region,the region of interest is expressed in terms of,the sound-shadow area is expressed,a background region is expressed in the form of,the inner region of the entire hyperellipse is expressed.
The method for solving the super-ellipse-constrained active contour model of the multiphase level set by using a gradient descent method to obtain a parameter evolution equation of the active contour model comprises the following steps of:
the first step is as follows: handleAndas a constant, the two sides of equation (5) are respectively aligned with c11、c12And c2The derivation is carried out, the fourth term is not considered as a constant term, and obtaining:
the second step is that: handle c11、c12And c2As a constant, the two sides of equation (5) are respectively pairedAndmake a derivation and order Obtaining:
wherein,andare all variable x0、y0A, b, theta;
the third step: the formula (5) is paired with x0、y0Partial differentiation is solved by five parameters of a, b and theta, then time t is introduced, and a parameter evolution equation is obtained by using a gradient descent method:
when the a < b is greater than the b,
when a > b is greater than the first threshold value,
in the formulae (14) and (15),
wherein,
a partial differential equation obtained by partial differentiating the five parameters with the level set function of the formula (3), wherein,
A=(x-x0)cosθ+(y-y0)sinθ
B=-(x-x0)sinθ+(y-y0)cosθ
when the a < b is greater than the b,
when a > b is greater than the first threshold value,
and (4) partial differential equation obtained by partial differentiation of the five parameters by the level set function of the formula (4).
The invention has the following advantages and beneficial effects:
1. under the condition that a large amount of submarine noise exists in the sonar image, the underwater mine target to be identified can be accurately found, and the anti-noise performance is good.
2. Since the shadow of the target is also recognized as a part of the target, the obtained contour curve includes the torpedo target and its shadow area.
3. Due to the addition of the shape constraint to the level set function, the resulting profile is a smooth hyperellipse with deformation resistance.
Drawings
FIG. 1 is a schematic view of a sonar image including a mine;
fig. 2 is a schematic diagram of a hyperelliptic curve with s = 2;
FIG. 3 is a diagram illustrating the evolution of the original C-V model;
FIG. 4 is a diagram illustrating evolution results of a C-V model with ellipse constraints;
FIG. 5 is a diagram illustrating evolution results of the method.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
1. Level set function of hyperelliptic constraint active contour model
The conventional level set function is defined as a symbol distance function, and the profile curve represented by the zero level set of the level set changes into an arbitrary shape along with the evolution process. When the target boundary has large distortion, the target contour curve of the method has large deformation in the evolution process, and the finally obtained contour curve is irregular and unsatisfactory. In order to solve the problem, a level set function introducing an elliptical shape constraint is proposed, and an ideal target contour is obtained. Because the shape of the mine target is more regular, the shape of the mine target is constrained by adding the shape of the mine target to the level set function by adopting a shape-preserving active contour model. The mines studied here are close to rectangles, but rectangles have instability in the evolution process, so hyperelliptic curves are used to approximate rectangles.
Hyperellipses are a family of curves that expand on the basis of ellipses by expanding the range of exponential values. The expression for a standard hyperellipse is:
in the formula, a, b and s are real numbers larger than zero. From equation (7), it can be seen that a hyperellipse is a series of closed curves obtained by allowing the index s of the x and y terms to vary in the corresponding ellipse equation.
When s → ∞ the hyperellipse becomes rectangular, but as can be seen from fig. 2, when s =2 the hyperellipse has already substantially approached the rectangle, with only slight curve transitions at the corners, which do not affect the recognition of the torpedo target and can therefore be used as a shape constraint for the torpedo target recognition.
Comprehensively considering the specific shape of a thunder target in a sonar image and the complexity of an algorithm during realization, adding translation, scaling and rotation parameters to the formula (1) by adopting an s =2 hyperelliptic curve to obtain an expression of s =2 hyperelliptic:
in the formula (x)0,y0) The coordinate of the center of the hyperellipse is shown as a, b are respectively the major axis and the minor axis of the hyperellipse, and theta is the rotation angle of the hyperellipse.
According to the definition of the level set function, the level set function of the hyperellipse is obtained as follows:
according to the definition of the level set function, a single level set function can only divide the image into a background area and a homogeneous target area. According to the gray distribution of the sonar image, the sonar image is divided into a background area, a target area and a sound shadow area, so that a multi-phase level set is required to be utilized, and another level set function is introduced. Assuming that the target zone and the insonification zone are both inside the hyperellipse and bounded by the major axis of the hyperellipse, the target zone and the insonification zone can be distinguished by only taking the major axis of the hyperellipse as the zero level set of the other level set function, thus obtaining the expression of the other level set function:
introduction ofAndafter two level set functions, the sonar image is divided into three regions. When in useWhen it represents the background area of the sonar image, whenAnd isWhen representing the target area of the sonar image, whenAnd isTime represents the sonographic area of the sonar image.
Energy function of 2-hyperelliptic constrained multi-phase level set active contour model
The energy function of the C-V model is divided into two terms which respectively represent the gray variance inside and outside the contour curve, and the final goal of evolution is to simultaneously minimize the gray variance inside and outside the curve. On the basis of the C-V model, an energy function is improved according to the gray level distribution characteristics of a sonar image.
In the sonar image, gray values of a target area, a sound shadow area and a background area are obviously different, and the gray value distribution in each area is uniform, so that the gray variance of the three areas reaches the minimum at the same time and is used as an evolution target of an energy function. In the whole image, the brightness and darkness contrast of the target area and the sound and shadow area is the maximum, so the difference between the gray level mean values of the target area and the sound and shadow area is added as an evolution target. Finally, an expression of the energy function is obtained:
wherein, c11Is the mean value of the gray levels of the target region, c12Mean value of gray levels of the sound and shadow region, c2Is the mean value of the gray levels of the background region,the region of interest is expressed in terms of,the sound-shadow area is expressed,a background region is expressed in the form of,the inner region of the overall hyperellipse is represented, including the target zone and the sonographic zone.
3 solving of shape-preserving active contour model
The following minimization formula (5) is carried out in two steps:
the first step is as follows: handleAndas a constant, the two sides of equation (5) are respectively aligned with c11、c12And c2The derivation is carried out, the fourth term can be regarded as a constant term and is not considered, and obtaining:
the second step is that: handle c11、c12And c2As a constant, the two sides of equation (5) are respectively pairedAndmake a derivation and order Obtaining:
as is clear from the formulae (3) and (4),andare all variable x0、y0A, b, theta, and so the evolution of the level set function can be translated into changes to these five parameters. And (3) performing partial differentiation on the five parameters by using the level set function of the formula (3) to obtain a partial differential equation:
wherein, A = (x-x)0)cosθ+(y-y0)sinθ,B=-(x-x0)sinθ+(y-y0)cosθ。
The partial differentiation of equation (4) needs to be divided into two cases:
when the a < b is greater than the b,
when a > b is greater than the first threshold value,
the formula (5) is paired with x0、y0Partial differentiation is solved by five parameters of a, b and theta, then time t is introduced, and a parameter evolution equation is obtained by using a gradient descent method:
when the a < b is greater than the b,
when a > b is greater than the first threshold value,
in the formulae (14) and (15),
4 results of the experiment
The method is experimentally verified by utilizing matlab language programming, and for comparison, an evolution result of an original C-V model and a C-V model test result added with ellipse constraint are given.
FIG. 3 is the result of an iteration of a sonar image with the original C-V model. As can be seen from the results, since the sonar image is affected by relatively serious noise, the region with relatively large noise is also recognized as a target and wrapped by a contour line in fig. 3; the edges are damaged due to the serious noise pollution of the target, so that the contour line of the wrapped target is not smooth.
FIG. 4 is the result of evolution using a C-V model based on an elliptical shape constraint. The contour line of the mine target obtained by the method is smooth, and a good experimental result can be obtained when the target edge is polluted. However, the method has the disadvantage that oscillation occurs near the true value during the evolution process. The reason for this is because the torpedo target is cylindrical and the two ends of the ellipse are relatively sharp, so that the truth value is not unique. Another disadvantage of this method is that the shadowgraph area of the mine target is not used as a feature for identifying the mine, and the result is mistaken.
Fig. 5 is a result obtained iteratively using the shape preserving active contour model based torpedo target identification method proposed herein. The result shows that under the condition that the submarine noise exists in the sonar image, the method can accurately find the underwater mine target to be identified, and has good anti-noise performance. Since we recognize the shadow of the target as part of the target, the obtained contour curve wraps the mine target and its shadow area. Due to the addition of the shape constraint to the level set function, the resulting profile is a smooth hyperellipse with deformation resistance.
Claims (6)
1. A sonar image mine target identification method based on an active contour is characterized by comprising the following steps:
introducing a super-elliptic shape constrained level set function into the active contour model according to the shape characteristics of the mine target;
according to the characteristics of a target sound shadow area in a sonar image, a multiphase level set function is introduced into an active contour model, so that the sonar image is divided into three areas: a background area, a target area and a sound-shadow area;
taking the minimum difference between the gray mean values of the target area and the sound-shadow area as an evolution target, namely the minimum energy function of the sound-shadow area of the mine target;
and solving the super-ellipse-constrained active contour model of the multiphase level set by using a gradient descent method to obtain a parameter evolution equation of the active contour model.
2. The active contour-based sonar image mine target identification method according to claim 1, wherein the level set function of the super-elliptic shape constraint is:
wherein (x)0,y0) The coordinate of the center of the hyperellipse is shown as a, b are respectively the major axis and the minor axis of the hyperellipse, and theta is the rotation angle of the hyperellipse.
3. The active contour-based sonar image mine target identification method according to claim 1, wherein the other level set function introduced by the multi-phase level set function is:
wherein (x)0,y0) The coordinate of the center of the hyperellipse is shown as a, b are respectively the major axis and the minor axis of the hyperellipse, and theta is the rotation angle of the hyperellipse.
4. The active contour-based sonar image mine target identification method according to claim 2 or 3, wherein the three areas are divided into: when in useWhen it represents the background area of the sonar image, whenAnd isWhen representing the target area of the sonar image, whenAnd isTime represents the sonographic area of the sonar image.
5. The active contour-based sonar image mine target identification method according to claim 1, wherein the energy function of the mine target shadow area is:
wherein, c11Is the mean value of the gray levels of the target region, c12Mean value of gray levels of the sound and shadow region, c2Is the mean value of the gray levels of the background region,the region of interest is expressed in terms of,the sound-shadow area is expressed,a background region is expressed in the form of,the inner region of the entire hyperellipse is expressed.
6. The sonar image mine target identification method based on active contour according to claim 5, wherein the active contour model constrained by the hyperellipse of the multi-phase level set is solved by using a gradient descent method to obtain a parameter evolution equation of the active contour model, comprising the following steps:
the first step is as follows: handleAndas a constant, the two sides of equation (5) are respectively aligned with c11、c12And c2The derivation is carried out, the fourth term is not considered as a constant term, and obtaining:
the second step is that: handle c11、c12And c2As a constant, the two sides of equation (5) are respectively pairedAndmake a derivation and order Obtaining:
wherein,andare all variable x0、y0A, b, theta;
the third step: the formula (5) is paired with x0、y0Partial differentiation is solved by five parameters of a, b and theta, then time t is introduced, and a parameter evolution equation is obtained by using a gradient descent method:
when the a < b is greater than the b,
when a > b is greater than the first threshold value,
in the formulae (14) and (15),
wherein,
a partial differential equation obtained by partial differentiating the five parameters with the level set function of the formula (3), wherein,
A=(x-x0)cosθ+(y-y0)sinθ
B=-(x-x0)sinθ+(y-y0)cosθ
when the a < b is greater than the b,
when a > b is greater than the first threshold value,
and (4) partial differential equation obtained by partial differentiation of the five parameters by the level set function of the formula (4).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310587459.1A CN104657736B (en) | 2013-11-19 | 2013-11-19 | A kind of sonar image mine recognition method based on active profile |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310587459.1A CN104657736B (en) | 2013-11-19 | 2013-11-19 | A kind of sonar image mine recognition method based on active profile |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104657736A true CN104657736A (en) | 2015-05-27 |
CN104657736B CN104657736B (en) | 2017-10-27 |
Family
ID=53248836
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310587459.1A Active CN104657736B (en) | 2013-11-19 | 2013-11-19 | A kind of sonar image mine recognition method based on active profile |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104657736B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106250848A (en) * | 2016-07-29 | 2016-12-21 | 西北工业大学 | A kind of single class Acoustic Object recognition methods merged based on multi-model |
CN108141201A (en) * | 2015-10-09 | 2018-06-08 | Med-El电气医疗器械有限公司 | Estimated using movable contour model for the harmonic frequency of hearing implant acoustic coding |
CN109166132A (en) * | 2018-07-16 | 2019-01-08 | 哈尔滨工程大学 | A kind of sidescan-sonar image target identification method of variable initial distance sign function |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102289812A (en) * | 2011-08-26 | 2011-12-21 | 上海交通大学 | Object segmentation method based on priori shape and CV (Computer Vision) model |
CN102496155A (en) * | 2011-10-28 | 2012-06-13 | 河海大学 | Underwater optical image processing method for optimizing C-V (chan-vese) model |
-
2013
- 2013-11-19 CN CN201310587459.1A patent/CN104657736B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102289812A (en) * | 2011-08-26 | 2011-12-21 | 上海交通大学 | Object segmentation method based on priori shape and CV (Computer Vision) model |
CN102496155A (en) * | 2011-10-28 | 2012-06-13 | 河海大学 | Underwater optical image processing method for optimizing C-V (chan-vese) model |
Non-Patent Citations (3)
Title |
---|
TANG YANDONG 等: "Automatic Segmentation of the Papilla in a Fundus Image Based on the C-V Model and a Shape Restraint", 《PROC OF THE 18TH INTERNATIONAL CONFERENCE ON PATERN RECOGNITION》 * |
VESE L A 等: "A Multiphase Level Set Framework for Image Segmentation Using the Mumford and Shah Model", 《INTERNATIONAL JOURNAL OF COMPUTER VISION》 * |
高山 等: "声呐图像水雷目标自动识别", 《水雷战与舰船防护》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108141201A (en) * | 2015-10-09 | 2018-06-08 | Med-El电气医疗器械有限公司 | Estimated using movable contour model for the harmonic frequency of hearing implant acoustic coding |
CN108141201B (en) * | 2015-10-09 | 2021-03-26 | Med-El电气医疗器械有限公司 | Harmonic frequency estimation for hearing implant sound coding using active contour model |
CN106250848A (en) * | 2016-07-29 | 2016-12-21 | 西北工业大学 | A kind of single class Acoustic Object recognition methods merged based on multi-model |
CN106250848B (en) * | 2016-07-29 | 2019-08-09 | 西北工业大学 | A kind of single class Acoustic Object recognition methods based on multi-model fusion |
CN109166132A (en) * | 2018-07-16 | 2019-01-08 | 哈尔滨工程大学 | A kind of sidescan-sonar image target identification method of variable initial distance sign function |
CN109166132B (en) * | 2018-07-16 | 2022-01-07 | 哈尔滨工程大学 | Side-scan sonar image target identification method with variable initial distance symbolic function |
Also Published As
Publication number | Publication date |
---|---|
CN104657736B (en) | 2017-10-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107145874B (en) | Ship target detection and identification method in complex background SAR image | |
CN104504686B (en) | A kind of hyperspectral image abnormal detection method of use local auto-adaptive Threshold segmentation | |
Feng et al. | Multiphase SAR image segmentation with $ G^{0} $-statistical-model-based active contours | |
CN105005042B (en) | A kind of GPR buried target localization method | |
CN103985125B (en) | Complicated background SAR image naval ship tail track detection method | |
CN102324021A (en) | Infrared dim-small target detection method based on shear wave conversion | |
CN112087774B (en) | Communication radiation source individual identification method based on residual error neural network | |
Debes et al. | Target discrimination and classification in through-the-wall radar imaging | |
CN102542277A (en) | Method for detecting ship trail of ocean synthetic aperture radar image | |
CN105303526A (en) | Ship target detection method based on coastline data and spectral analysis | |
CN101901476A (en) | SAR image de-noising method based on NSCT domain edge detection and Bishrink model | |
CN104657736B (en) | A kind of sonar image mine recognition method based on active profile | |
CN109829858B (en) | Ship-borne radar image oil spill monitoring method based on local adaptive threshold | |
CN105574529A (en) | Target detection method of side scan sonar | |
CN102542540B (en) | Method for inhibiting infrared image background based on PDE (Partial Differential Equation) | |
CN105809649A (en) | Variation multi-scale decomposing based SAR image and visible light image integration method | |
CN109584256B (en) | Pulsar dispersion value estimation method based on Hough line detection | |
CN102750675A (en) | Non-local means filtering method for speckle noise pollution image | |
CN103377465B (en) | Based on the SAR image method for reducing speckle that sketch map and core are selected | |
CN107369163A (en) | A kind of quick SAR image object detection method based on best entropy Double Thresholding Segmentation | |
CN109461171A (en) | The small IR targets detection algorithm of DoG filtering is improved based on multichannel | |
Sung et al. | Sonar image translation using generative adversarial network for underwater object recognition | |
CN110097562B (en) | Sea surface oil spill area image detection method | |
Kim et al. | GPR image enhancement based on frequency shifting and histogram dissimilarity | |
Ji-yang et al. | On-board ship targets detection method based on multi-scale salience enhancement for remote sensing image |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |