CN111582198A - Automatic sea-land segmentation method for remote sensing image - Google Patents

Automatic sea-land segmentation method for remote sensing image Download PDF

Info

Publication number
CN111582198A
CN111582198A CN202010396644.2A CN202010396644A CN111582198A CN 111582198 A CN111582198 A CN 111582198A CN 202010396644 A CN202010396644 A CN 202010396644A CN 111582198 A CN111582198 A CN 111582198A
Authority
CN
China
Prior art keywords
sea
land
remote sensing
sensing image
segmentation
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
Application number
CN202010396644.2A
Other languages
Chinese (zh)
Other versions
CN111582198B (en
Inventor
吴诗婳
李亚钊
于子桓
李彭伟
李子
刘博�
马建辉
陆君之
赵祥智
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
CETC 28 Research Institute
Original Assignee
CETC 28 Research Institute
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by CETC 28 Research Institute filed Critical CETC 28 Research Institute
Priority to CN202010396644.2A priority Critical patent/CN111582198B/en
Publication of CN111582198A publication Critical patent/CN111582198A/en
Application granted granted Critical
Publication of CN111582198B publication Critical patent/CN111582198B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Astronomy & Astrophysics (AREA)
  • Remote Sensing (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a remote sensing image sea-land automatic segmentation method, which comprises the following steps: 1. carrying out linear enhancement processing on the remote sensing image to be segmented to improve the image contrast; 2. improving an FLICM clustering method by using a generalized radial Gaussian kernel function to obtain a land initial region, and extracting a coastline rough contour by combining a Canny operator; 3. and (5) taking the area and the contour obtained according to the rough segmentation result as initial conditions of the weighted CV model to finish the sea-land fine segmentation of the remote sensing image. 4. And correcting the sea and land fine segmentation result by adopting morphology and connected domain criteria to obtain a final sea and land segmentation result. The invention effectively realizes the automatic sea-land segmentation of the large-scene remote sensing image.

Description

Automatic sea-land segmentation method for remote sensing image
Technical Field
The invention relates to the field of remote sensing image sea-land segmentation, in particular to a remote sensing image sea-land automatic segmentation method.
Background
The sea and land separation can be realized through sea and land segmentation, the interference of a complex background of a land area is inhibited, and the difficulty of sea surface target detection is effectively reduced. Sea and land segmentation is used as the premise and the basis of sea surface target detection and identification, the quality of a segmentation result directly influences the accuracy of subsequent processing, and the method has important significance in the fields of civil use, military reconnaissance, ocean monitoring and the like. In addition, the coastline information obtained by sea-land segmentation plays a crucial role in a variety of application scenarios such as tidal observation, mapping, ocean resource management, and the like. Therefore, an effective sea and land division method is needed.
The traditional sea-land segmentation is generally based on methods such as a sea chart, a profile monitoring method, a topographic map and the like, consumes a large amount of time and manpower, and has poor dynamic tracking performance. With the attention of people on remote sensing images, a remote sensing image-based sea-land segmentation method gradually becomes a research hotspot. At present, the sea and land segmentation method based on remote sensing images is mainly divided into region information based and boundary contour based. The fuzzy C-means (FCM) clustering method is continuously improved as an effective image segmentation method based on regional information. In the fuzzy local information C-means (FLICM) clustering method proposed in recent years, the local space and gray level information of pixels are comprehensively considered in the pixel fuzzy membership degree judgment process, the stronger robustness is shown, but a kernel function mapping mode is not considered, the image is segmented only according to the gray level of an image area, the global information of the image is not considered, and the segmentation precision needs to be improved. On the other hand, the method based on the edge-free active contour model Chan-Vese (CV) utilizes image global information to establish an energy function so as to control curve motion, has high segmentation precision and low calculation efficiency, only utilizes the average value information of the region to determine the boundary, and is sensitive to initial conditions.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems in the field of the existing remote sensing image sea-land segmentation, the accuracy and the speed of the remote sensing image sea-land segmentation are further improved, and the invention discloses a remote sensing image sea-land automatic segmentation method based on improved FLICM and weighted CV. Aiming at the problems of unsatisfactory contrast and poor subjective visual effect of a remote sensing image, firstly, the remote sensing image is linearly stretched to highlight interesting information; then, carrying out sea-land coarse segmentation on the remote sensing image by adopting an improved FLICM clustering method, and introducing a generalized radial Gaussian kernel function to enable the pixel points to have better linear aggregability in a high-dimensional characteristic space; in order to avoid the defect that the CV model is sensitive to initial conditions, the land area and the coastline rough contour obtained by rough segmentation are respectively used as initial iteration conditions of the weighted CV model to realize sea and land fine segmentation; and finally, different characteristics of land and sea are comprehensively considered, and the morphological means and the communication domain criterion are utilized to further keep the integrity of the coastline and eliminate impurity scattering points at the same time, so that the final sea and land segmentation result is obtained.
The technical scheme is as follows: the invention discloses a remote sensing image sea-land automatic segmentation method based on improved FLICM and weighted CV, which comprises the following steps:
the method comprises the following steps:
step 1, performing linear transformation enhancement processing on a remote sensing image to be segmented;
step 2, introducing a generalized radial Gaussian kernel function to improve the FLICM clustering method, realizing sea-land rough segmentation of the remote sensing image, and extracting a coastline rough contour by combining a Canny algorithm;
step 3, calculating to obtain a sea-land fine segmentation result of the remote sensing image;
and 4, further correcting the sea and land fine segmentation result by using a morphological means and a connected domain criterion, and finally completing sea and land segmentation.
The step 1 comprises the following steps: stretching the remote sensing image to be segmented by using the following linear transformation formula:
Figure BDA0002487805630000021
in the formula, geG is the original image gray level of the remote sensing image to be segmented, g is the gray level of the enhanced imagemax、gminThe maximum value and the minimum value of the gray level of the original image are respectively.
The step 2 comprises the following steps:
step 2-1, initializing kernel fuzzy local information C mean value clustering parameters: setting the number n, the clustering number c, the membership index m and the maximum iteration number T of the pixels of the image to be clusteredMThe condition of stopping iteration is more than 0, the number of initialized iteration t is 0, and the original membership uikAnd cluster center { v1,v2,...,vc}，vcRepresenting the c-th cluster center;
step 2-2, introducing a generalized radial Gaussian kernel function, and calculating each sample x by using the following formulakOriginal degree of membership uik：
Figure BDA0002487805630000022
Wherein the value ranges of i and j are 1-c; kugrbAs a generalized radial gaussian kernel function:
Figure BDA0002487805630000031
in the formula, b, r and rho are adjustable parameters, b is greater than 0, b belongs to z, and z is an integer set; r is more than or equal to 1, P is a nuclear parameter, and D is the dimension of data; x and y respectively represent the abscissa and the ordinate of a pixel point on the image;
step 2-3, calculate and update each sample x using the following equationkNew fuzzy degree of membership of
Figure BDA0002487805630000032
Figure BDA0002487805630000033
In the formula, xkThe degree of membership of the i-th class region is uik；xkMembership degree of j-th class region is ujk；
p and q are index parameters for determining the correlation between the original membership function and the spatial function;
space function hikRepresenting x determined by local spatial pixelskA likelihood of belonging to a category i region;
space function hjkRepresenting x determined by local spatial pixelskThe likelihood of belonging to a class j region;
Figure BDA0002487805630000034
uitis a pixel xtDegree of membership belonging to class i local area, variable t belonging to Nk，NkIs given by xkA local area that is a center;
step 2-4, calculating and updating a clustering center:
calculating and updating the clustering center at the t +1 moment by using the following formula
Figure BDA0002487805630000036
：
Figure BDA0002487805630000035
v′iRepresents the cluster center at time t, u'ikRepresenting the updated membership;
step 2-5, if
Figure BDA0002487805630000037
Or T ═ TM，TMIf the threshold value is set, stopping iterative updating, otherwise, returning to the step 2-2, if t is t + 1;
step 2-6, according to the maximum blurMembership level decision criterion if uji＞ujkThen x isjDividing the remote sensing image into the ith type area, clustering and dividing the enhanced remote sensing image to obtain a remote sensing image sea-land rough division result ujiRepresents the membership degree of the i-th type area, k is 1, 2. i is not equal to k;
and 2-7, extracting a coastline rough contour by adopting a Canny algorithm according to the sea-land rough segmentation result of the remote sensing image obtained after rough segmentation in the step 2-6.
The step 3 comprises the following steps:
step 3-1, initializing a zero level set function according to the coastline rough contour obtained in the step 2-7
Figure BDA0002487805630000041
As shown in the following formula:
Figure BDA0002487805630000042
and 3-2, expressing the sea and land coarse segmentation result obtained in the step 2-6 as an improved kernel fuzzy local information clustering result g (x, y), taking the g (x, y) as the input of a weighted CV model, and calculating by improving a partial differential equation according to the following formula to minimize an energy function:
Figure BDA0002487805630000043
Figure BDA0002487805630000044
representing an ideal step function, which participates in the numerical operation using the following equation:
Figure BDA0002487805630000045
Figure BDA0002487805630000046
in the formula, a small positive number (generally 10) tending to 0 is represented-6)；
μ、v、λ1、λ2Is a constant number, C1、C2Respectively for introducing adaptive weight1And2the target and background fitting centers of (a), are defined as follows:
Figure BDA0002487805630000047
wherein the parameters
Figure BDA0002487805630000051
Parameter(s)
Figure BDA0002487805630000052
Self-adaptive adjustment is realized through iteration, and the coastline rough contour is subjected to CV model iteration segmentation to obtain an optimal contour;
and 3-3, for the obtained optimal contour, realizing sea-land fine segmentation through level set function updating and boundary evolution.
In step 3-3, the symbol distance function is selected as the level set function, and then the level set function becomes:
Figure BDA0002487805630000053
and 3, the contour is the optimal contour obtained in the step 3-2, and d is a symbolic distance function and represents the distance from a point (x, y) in the high-dimensional space to a zero level set.
Step 4 comprises the following steps: and (3) adopting a connected domain judgment criterion, taking the largest connected domain in the sea-land fine segmentation result as the sea, eliminating stray points and noise possibly existing in the sea by a morphological open operation method, and removing holes in the target by a morphological close operation method to finally finish the automatic sea-land segmentation.
Has the advantages that: compared with the prior art, the remote sensing image sea-land automatic segmentation method based on the improved FLICM and the weighted CV has the advantages that:
(1) the method organically combines two types of mainstream algorithms based on regional information and boundary contour in the sea and land segmentation of the existing remote sensing image, fully considers the locality and the globality of the image, makes up for the deficiencies, and improves the precision and the speed of the sea and land segmentation of the remote sensing image.
(2) The remote sensing image is preprocessed by the image enhancement method, the integral contrast of the image is improved, the use value of the image is improved, and good data conditions are provided for subsequent segmentation.
(3) And a generalized radial Gaussian kernel function is introduced to improve an FLICM clustering method, and a sample characteristic nonlinear mapping high-dimensional characteristic space of an image to be segmented is amplified, so that the method is beneficial to the resolution and extraction of characteristics. And the generalized radial Gaussian kernel function avoids the defect that a single Gaussian kernel is unique in adjustable parameter, and the learning and generalization capability is improved through multiple parameters, so that the clustering performance of the FLICM method is improved, and the sea-land rough segmentation of the remote sensing image is realized.
(4) The remote sensing image kernel fuzzy local information C mean value clustering result is used as an initial region and an outline of the weighted CV model, so that the problems of unknown initial conditions and long convergence time of the traditional CV model are solved, and the speed and the intelligent degree of the method are improved. Meanwhile, a self-adaptive weight weighted average strategy is constructed to replace the traditional arithmetic average calculation pixel point contribution value to the fitting center, the difference is fully considered, the segmentation result is more accurate, and the sea and land fine segmentation of the remote sensing image is realized.
(5) And correcting the fine segmentation result by adopting morphology and a connected domain criterion according to different region characteristics, and further ensuring the performance of the segmentation method.
Drawings
The foregoing and/or other advantages of the invention will become further apparent from the following detailed description of the invention when taken in conjunction with the accompanying drawings.
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is a remote sensing image to be segmented of the present invention;
FIG. 3 is a result of enhancing a remote sensing image to be segmented according to the present invention;
FIG. 4 is a sea-land rough segmentation result of the remote sensing image of the present invention;
FIG. 5 is a rough contour of the coastline of the remote sensing image of the present invention;
FIG. 6 is a sea-land fine segmentation result of the remote sensing image according to the present invention;
FIG. 7 is the final result of sea-land segmentation of the remote sensing image according to the present invention;
FIG. 8 shows the superposition effect of the remote sensing image sea-land segmentation result and the original image.
Detailed Description
The invention provides a remote sensing image sea-land automatic segmentation method based on an improved FLICM and a weighted CV. The method comprises the steps of firstly, carrying out linear stretching on a remote sensing image to improve the readability of the image and highlight interesting information; then introducing a generalized radial Gaussian kernel function to improve an FLICM clustering method to carry out sea-land coarse segmentation on the remote sensing image; respectively taking the land area and the coastline rough contour obtained by rough segmentation as initial iteration conditions of the weighted CV model to realize sea and land fine segmentation; and finally, comprehensively considering different characteristics of land and sea, and eliminating impurity scattering points by using morphological means and a connected domain criterion to obtain a final sea and land segmentation result.
The invention is further elucidated with reference to the drawings and the detailed description.
The flow diagram for implementing the invention is shown in fig. 1, and the method comprises the following specific implementation steps:
step 1: and preprocessing the remote sensing image to be segmented. In view of the fact that the gray level of the original image is often concentrated in a certain gray level interval, the image contrast is poor, and subsequent segmentation is not facilitated. In order to solve the problem, the following linear transformation formula is used to stretch the image to be segmented (as shown in fig. 2), the gray scale range of the image is readjusted, the visual effect of the image is improved, and the enhanced image is shown in fig. 3.
Figure BDA0002487805630000061
In the formula, geFor the enhanced image gray level, g is the original image gray level, gmax、gminThe maximum value and the minimum value of the gray level of the original image are respectively.
The step 2 comprises the following steps:
and 2-1, initializing a kernel fuzzy local information C mean value clustering parameter. Setting the number n, the clustering number c, the membership index m and the maximum iteration number T of the pixels of the image to be clusteredMThe condition of stopping iteration is more than 0, the number of initialized iteration t is 0, and the original membership uikAnd cluster center { v1,v2,...,vc}。
And 2-2, calculating an original membership value. Introducing a generalized radial Gaussian kernel function, and calculating each sample x by using the following formulakOriginal degree of membership uik。
Figure BDA0002487805630000071
Wherein, KugrbAs a generalized radial gaussian kernel function:
Figure BDA0002487805630000072
in the formula, b is larger than 0, b belongs to z, r is larger than or equal to 1, P is a nuclear parameter, and D is the dimensionality of data.
In this case, b is 2, r is 2, D is 3, P is 1, and D is 1.
And 2-3, calculating and updating the new membership value. Each sample x is calculated and updated taking into account the spatially local characteristics of the pixels using the following formulakNew fuzzy degree of membership of
Figure BDA0002487805630000073
Improve the accuracy of clustering to a certain extent, restrain the interference:
Figure BDA0002487805630000074
in the formula (I), the compound is shown in the specification,
Figure BDA0002487805630000075
uitis a pixel xtDegree of membership, N, belonging to class i local areakIs given by xkA central local area. p and q determine hikAnd uikThe association of (a). When the p value is unchanged, the q value is increased to a certain extent, the more the spatial local characteristics of the pixels are considered, the better the clustering effect is. In this case, p is 2 and q is 6.
And 2-4, calculating and updating the clustering center. Calculating and updating cluster centers using the following equation
Figure BDA0002487805630000085
Figure BDA0002487805630000081
Step 2-5, if
Figure BDA0002487805630000086
Or T ═ TMIn this case, set to 0.1, TmAnd stopping the iterative updating at 150, otherwise, returning to the step 2-2 at t + 1.
Step 2-6, according to the maximum fuzzy membership degree judgment criterion, if uji＞ujkThen x isjDividing the remote sensing image into an i-th type area, and performing clustering segmentation on the enhanced remote sensing image to obtain a sea-land rough segmentation result, wherein k is 1, 2. i ≠ k (as shown in fig. 4).
And 2-7, extracting a coastline rough contour by adopting a Canny algorithm according to the land area obtained after the rough segmentation in the step 2-6, as shown in figure 5.
The step 3 comprises the following steps:
step 3-1, initializing a zero level set function according to the coastline rough contour obtained in step 2-7
Figure BDA0002487805630000084
As shown in the following formula:
Figure BDA0002487805630000082
step 3-2, considering that the traditional CV model adopts arithmetic mean to calculate the fitting center and does not consider the difference of contribution values of any pixel point to the fitting center to cause inaccurate segmentation result, the method adoptsWeighted CV model, introducing adaptive weight1And2and calculating a fitting center to improve the segmentation effect. And (3) taking the improved kernel fuzzy local information clustering result g (x, y) obtained in the step (2-6) as the input of the weighted CV model, calculating according to the improved partial differential equation of the following formula to minimize the energy function, and obtaining the optimal contour through level set updating and boundary evolution, thereby realizing sea and land fine segmentation, as shown in FIG. 6.
Figure BDA0002487805630000083
In the formula, mu, v, lambda1、λ2As a constant, in this case set λ1＝λ2＝1、μ＝1、v＝0，C1、C2Respectively for introducing adaptive weight1And2the target and background fitting centers of (a), are defined as follows:
Figure BDA0002487805630000091
wherein
Figure BDA0002487805630000092
Adaptive adjustment is achieved by iteration.
Step 4 comprises the following steps: considering the fragmentation, connectivity and continuity of the sea, the maximum connected domain in the sea-land fine segmentation result is taken as the sea by adopting a connected domain judgment criterion, so that the interference of a land complex scene is avoided. In addition, stray points and noise possibly existing in the ocean are eliminated through a morphological opening operation method, then holes in the target are removed through a morphological closing operation method, the integrity of the target is kept, and finally the sea and land automatic segmentation is completed (as shown in fig. 7), and the superposition effect of the sea and land segmentation result and the original image is shown in fig. 8.
The invention provides a remote sensing image sea and land automatic segmentation method, and a plurality of methods and ways for implementing the technical scheme are provided, the above description is only a preferred embodiment of the invention, it should be noted that, for a person skilled in the art, a plurality of improvements and decorations can be made without departing from the principle of the invention, and these improvements and decorations should also be regarded as the protection scope of the invention. All the components not specified in the present embodiment can be realized by the prior art.

Claims (6)

1. A remote sensing image sea-land automatic segmentation method is characterized by comprising the following steps:
step 1, performing linear transformation enhancement processing on a remote sensing image to be segmented;
step 2, introducing a generalized radial Gaussian kernel function to improve the FLICM clustering method, realizing sea-land rough segmentation of the remote sensing image, and extracting a coastline rough contour by combining a Canny algorithm;
step 3, calculating to obtain a sea-land fine segmentation result of the remote sensing image;
and 4, further correcting the sea and land fine segmentation result by using a morphological means and a connected domain criterion, and finally completing sea and land segmentation.
2. The method of claim 1, wherein step 1 comprises: stretching the remote sensing image to be segmented by using the following linear transformation formula:
Figure FDA0002487805620000011
in the formula, geG is the original image gray level of the remote sensing image to be segmented, g is the gray level of the enhanced imagemax、gminThe maximum value and the minimum value of the gray level of the original image are respectively.
3. The method of claim 2, wherein step 2 comprises:
step 2-1, initializing kernel fuzzy local information C mean value clustering parameters: setting the number n, the clustering number c, the membership index m and the maximum iteration number T of the pixels of the image to be clusteredMThe condition of stopping iteration is more than 0, the number of initialized iteration t is 0, and the original membership uikAnd cluster center { v1,v2,...,vc}，vcRepresenting the c-th cluster center;
step 2-2, introducing a generalized radial Gaussian kernel function, and calculating each sample x by using the following formulakOriginal degree of membership uik：
Figure FDA0002487805620000012
Wherein the value ranges of i and j are 1-c; kugrbAs a generalized radial gaussian kernel function:
Figure FDA0002487805620000013
in the formula, b, r and rho are adjustable parameters, b is greater than 0, b belongs to z, and z is an integer set; r is more than or equal to 1, P is a nuclear parameter, and D is the dimension of data; x and y respectively represent the abscissa and the ordinate of a pixel point on the image;
step 2-3, calculate and update each sample x using the following equationkNew fuzzy degree of membership of
Figure FDA0002487805620000021
Figure FDA0002487805620000022
In the formula, xkThe degree of membership of the i-th class region is uik；xkMembership degree of j-th class region is ujk；
p and q are index parameters for determining the correlation between the original membership function and the spatial function;
space function hikRepresenting x determined by local spatial pixelskA likelihood of belonging to a category i region;
space function hjkRepresenting x determined by local spatial pixelskThe likelihood of belonging to a class j region;
Figure FDA0002487805620000023
uitis a pixel xtDegree of membership belonging to class i local area, variable t belonging to Nk，NkIs given by xkA local area that is a center;
step 2-4, calculating and updating a clustering center:
calculating and updating the clustering center at the t +1 moment by using the following formula
Figure FDA0002487805620000024
Figure FDA0002487805620000025
v′iRepresents the cluster center at time t, u'ikRepresenting the updated membership;
step 2-5, if
Figure FDA0002487805620000026
Or T ═ TM，TMIf the threshold value is set, stopping iterative updating, otherwise, returning to the step 2-2, if t is t + 1;
step 2-6, according to the maximum fuzzy membership degree judgment criterion, if uji＞ujkThen x isjDividing the remote sensing image into the ith type area, clustering and dividing the enhanced remote sensing image to obtain a remote sensing image sea-land rough division result ujiRepresents the membership degree of the i-th type area, k is 1, 2. i is not equal to k;
and 2-7, extracting a coastline rough contour by adopting a Canny algorithm according to the sea-land rough segmentation result of the remote sensing image obtained after rough segmentation in the step 2-6.
4. The method of claim 3, wherein step 3 comprises:
step 3-1, initializing a zero level set function according to the coastline rough contour obtained in the step 2-7
Figure FDA0002487805620000039
As shown in the following formula:
Figure FDA0002487805620000031
and 3-2, expressing the sea and land coarse segmentation result obtained in the step 2-6 as an improved kernel fuzzy local information clustering result g (x, y), taking the g (x, y) as the input of a weighted CV model, and calculating by improving a partial differential equation according to the following formula to minimize an energy function:
Figure FDA0002487805620000032
Figure FDA0002487805620000033
representing an ideal step function, which participates in the numerical operation using the following equation:
Figure FDA0002487805620000034
Figure FDA0002487805620000035
in the formula, a small positive number tending to 0 is represented;
μ、v、λ1、λ2is a constant number, C1、C2Respectively for introducing adaptive weight1And2the target and background fitting centers of (a), are defined as follows:
Figure FDA0002487805620000036
wherein the parameters
Figure FDA0002487805620000037
Parameter(s)
Figure FDA0002487805620000038
Self-adaptive adjustment is realized through iteration, and the coastline rough contour is subjected to CV model iteration segmentation to obtain an optimal contour;
and 3-3, for the obtained optimal contour, realizing sea-land fine segmentation through level set function updating and boundary evolution.
5. The method according to claim 4, wherein in step 3-3, the symbol distance function is selected as the level set function, and the level set function becomes:
Figure FDA0002487805620000041
and 3, the contour is the optimal contour obtained in the step 3-2, and d is a symbolic distance function and represents the distance from a point (x, y) in the high-dimensional space to a zero level set.
6. The method of claim 5, wherein step 4 comprises: and (3) adopting a connected domain judgment criterion, taking the largest connected domain in the sea-land fine segmentation result as the sea, eliminating stray points and noise possibly existing in the sea by a morphological open operation method, and removing holes in the target by a morphological close operation method to finally finish the automatic sea-land segmentation.
CN202010396644.2A 2020-05-12 2020-05-12 Automatic sea-land segmentation method for remote sensing image Active CN111582198B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010396644.2A CN111582198B (en) 2020-05-12 2020-05-12 Automatic sea-land segmentation method for remote sensing image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010396644.2A CN111582198B (en) 2020-05-12 2020-05-12 Automatic sea-land segmentation method for remote sensing image

Publications (2)

Publication Number Publication Date
CN111582198A true CN111582198A (en) 2020-08-25
CN111582198B CN111582198B (en) 2022-10-04

Family

ID=72113401

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010396644.2A Active CN111582198B (en) 2020-05-12 2020-05-12 Automatic sea-land segmentation method for remote sensing image

Country Status (1)

Country Link
CN (1) CN111582198B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112434642A (en) * 2020-12-07 2021-03-02 北京航空航天大学 Sea-land segmentation method suitable for processing large-scene optical remote sensing data
CN114742854A (en) * 2022-04-02 2022-07-12 西安电子科技大学 SAR image sea-land segmentation method based on scene prior knowledge and region combination
CN116318406A (en) * 2023-05-19 2023-06-23 深圳市飞思卓科技有限公司 Signal compensation method and system for optical fiber communication system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103208116A (en) * 2013-03-20 2013-07-17 西安电子科技大学 Neighborhood-information-based method for segmentation of images with non-uniform gray levels for blurred active contour model
WO2016101279A1 (en) * 2014-12-26 2016-06-30 中国海洋大学 Quick detecting method for synthetic aperture radar image of ship target
WO2017071160A1 (en) * 2015-10-28 2017-05-04 深圳大学 Sea-land segmentation method and system for large-size remote-sensing image
CN109658416A (en) * 2018-12-27 2019-04-19 北京航空航天大学 SAR image sea land dividing method, device, computer equipment and readable storage medium storing program for executing

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103208116A (en) * 2013-03-20 2013-07-17 西安电子科技大学 Neighborhood-information-based method for segmentation of images with non-uniform gray levels for blurred active contour model
WO2016101279A1 (en) * 2014-12-26 2016-06-30 中国海洋大学 Quick detecting method for synthetic aperture radar image of ship target
WO2017071160A1 (en) * 2015-10-28 2017-05-04 深圳大学 Sea-land segmentation method and system for large-size remote-sensing image
CN109658416A (en) * 2018-12-27 2019-04-19 北京航空航天大学 SAR image sea land dividing method, device, computer equipment and readable storage medium storing program for executing

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112434642A (en) * 2020-12-07 2021-03-02 北京航空航天大学 Sea-land segmentation method suitable for processing large-scene optical remote sensing data
CN112434642B (en) * 2020-12-07 2022-11-11 北京航空航天大学 Sea-land segmentation method suitable for processing large-scene optical remote sensing data
CN114742854A (en) * 2022-04-02 2022-07-12 西安电子科技大学 SAR image sea-land segmentation method based on scene prior knowledge and region combination
CN114742854B (en) * 2022-04-02 2023-06-02 西安电子科技大学 SAR image sea-land segmentation method based on scene priori knowledge and region combination
CN116318406A (en) * 2023-05-19 2023-06-23 深圳市飞思卓科技有限公司 Signal compensation method and system for optical fiber communication system
CN116318406B (en) * 2023-05-19 2023-08-08 深圳市飞思卓科技有限公司 Signal compensation method and system for optical fiber communication system

Also Published As

Publication number Publication date
CN111582198B (en) 2022-10-04

Similar Documents

Publication Publication Date Title
CN110163110B (en) Pedestrian re-recognition method based on transfer learning and depth feature fusion
CN111582198B (en) Automatic sea-land segmentation method for remote sensing image
Li et al. SAR image change detection using PCANet guided by saliency detection
CN111340824B (en) Image feature segmentation method based on data mining
Pang et al. Automatic segmentation of crop leaf spot disease images by integrating local threshold and seeded region growing
CN105894490A (en) Fuzzy integration multiple classifier integration-based uterine neck cell image identification method and device
CN109978848A (en) Method based on hard exudate in multiple light courcess color constancy model inspection eye fundus image
CN107516083B (en) Recognition-oriented remote face image enhancement method
CN111091129B (en) Image salient region extraction method based on manifold ordering of multiple color features
CN104036526A (en) Gray target tracking method based on self-adaptive window
CN109712160A (en) Improved lion group algorithm is combined to realize carrying out image threshold segmentation method based on generalized entropy
Wang et al. Segmentation of corn leaf disease based on fully convolution neural network
Xiao et al. Traffic sign detection based on histograms of oriented gradients and boolean convolutional neural networks
CN111199245A (en) Rape pest identification method
Zhao et al. A novel Neutrosophic image segmentation based on improved fuzzy C-means algorithm (NIS-IFCM)
CN103971362B (en) SAR image change-detection based on rectangular histogram and elite genetic algorithm for clustering
CN110135435B (en) Saliency detection method and device based on breadth learning system
CN107423771A (en) A kind of two phase method for detecting change of remote sensing image
Lou et al. Research on edge detection method based on improved HED network
CN110853030B (en) Bioreactor virus infected cell quality evaluation method
CN110211106B (en) Mean shift SAR image coastline detection method based on segmented Sigmoid bandwidth
CN111539985A (en) Self-adaptive moving target tracking method fusing multiple features
CN108205814B (en) Method for generating black and white contour of color image
CN112200831B (en) Dynamic template-based dense connection twin neural network target tracking method
Kong et al. Adaptive image edge detection model using improved Canny algorithm

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
CB02 Change of applicant information

Address after: No.1 Lingshan South Road, Qixia District, Nanjing, Jiangsu Province, 210000

Applicant after: THE 28TH RESEARCH INSTITUTE OF CHINA ELECTRONICS TECHNOLOGY Group Corp.

Address before: 210007 No. 1 East Street, alfalfa garden, Jiangsu, Nanjing

Applicant before: THE 28TH RESEARCH INSTITUTE OF CHINA ELECTRONICS TECHNOLOGY Group Corp.

CB02 Change of applicant information
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant