CN107784285B - Method for automatically judging civil and military attributes of optical remote sensing image ship target - Google Patents

Method for automatically judging civil and military attributes of optical remote sensing image ship target Download PDF

Info

Publication number
CN107784285B
CN107784285B CN201711012734.1A CN201711012734A CN107784285B CN 107784285 B CN107784285 B CN 107784285B CN 201711012734 A CN201711012734 A CN 201711012734A CN 107784285 B CN107784285 B CN 107784285B
Authority
CN
China
Prior art keywords
ship target
military
civil
attributes
ship
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201711012734.1A
Other languages
Chinese (zh)
Other versions
CN107784285A (en
Inventor
帅通
楚博策
师本慧
陈金勇
刘翔
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 54 Research Institute
Original Assignee
CETC 54 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 54 Research Institute filed Critical CETC 54 Research Institute
Priority to CN201711012734.1A priority Critical patent/CN107784285B/en
Publication of CN107784285A publication Critical patent/CN107784285A/en
Application granted granted Critical
Publication of CN107784285B publication Critical patent/CN107784285B/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
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures

Landscapes

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

Abstract

The invention discloses an automatic discrimination method for the military and civil attributes of an optical remote sensing image ship target, and relates to the field of automatic processing of optical remote sensing images. The method comprises the steps of obtaining a ship target training sample, carrying out a series of operations such as military and civil attribute manual discrimination, normalization resampling, generation of a collection, average image calculation, difference image calculation, orthogonal vector calculation and the like on the training sample to construct a military and civil attribute discrimination feature space of the ship target, projecting a test image to the space, and automatically discriminating the military and civil attribute of the ship target in the test image through distance measurement. According to the method, only one-time construction of a warship target civil and military attribute discrimination feature space is needed, and civil and military attributes of a warship target found in a test image can be automatically discriminated by using the feature space in future work. By the method, the civil and military attributes of the ship target in the newly found optical remote sensing image can be automatically distinguished, the brain of professional interpreters is liberated, and the working efficiency of ship target identification is greatly improved.

Description

Method for automatically judging civil and military attributes of optical remote sensing image ship target
Technical Field
The invention relates to the field of automatic processing of optical remote sensing images, in particular to an automatic civil and military attribute discrimination method for an optical remote sensing image ship target.
Background
The satellite remote sensing technology is an effective means for detecting marine ship targets, and when the spatial resolution of a remote sensing image is low, only the ship targets can be found; when the spatial resolution of the remote sensing image is high, the ship target can be found, and the attribute and the type of the ship target can be distinguished and identified.
According to the characteristics of remote sensing images, researchers provide detection algorithms of various ship targets, and generally speaking, detection is carried out on the basis of the gray level difference between the target gray level and the background. Such as a target extraction method based on minimum error threshold segmentation, a morphological contrast method, a method for realizing separation of a target and a background based on a fuzzy analysis theory, a ship detection method based on an Otsu segmentation method, a detection method based on a visual attention mechanism, CFAR detection based on a statistical model, and the like. Based on the technology and the method, the ship target in the remote sensing image can be found.
There are two main algorithms for identifying the type of the ship target: the method is characterized in that a method capable of distinguishing the target from the false alarm is found by utilizing priori knowledge according to the difference of the real target and the false alarm in the characteristics of shape, gray level, texture and the like, so as to identify the target; the other method is to apply the idea of machine learning to target recognition, and the basic idea is to take the result of manual interpretation as training data of machine learning, generate a discriminator capable of classifying the target and the false alarm, and recognize the type of the target by using the classifier.
The marine vessels are various in types and quantity, and generally can be divided into military vessels and civil vessels. Compared with military ships, the civil ships are more complex in type and wider in application, meanwhile, the ship surface covers are various in type and random in arrangement, and even if ships of the same type have different carrying objects, the texture and the gray scale of the ships are greatly different; and the military ships have relatively high consistency in layout, material properties and the like. Aiming at different application purposes, the military and civil attributes of the ship target are automatically judged, so that not only can the basic judgment be rapidly carried out on the attributes of the ship target, but also the later-stage fine identification efficiency of the ship target can be effectively improved. According to different characteristics of military and civil ships, an automatic discrimination method for the military and civil attributes of an optical remote sensing image ship target is provided.
Disclosure of Invention
In view of the above, the main object of the present invention is to provide an automatic algorithm for determining military and civil attributes of a ship target, which is used to automatically classify the ship target into military ships and civil ships, so as to meet the target observation requirements of different military and civil users and narrow the range for further fine identification of the ship target, aiming at the problem that it is extremely difficult to find out an interested target by manual screening in optical remote sensing.
In order to achieve the aim, the invention provides a method for automatically judging the military and civil attributes of an optical remote sensing image ship target, which comprises the following steps:
step 1, automatically rotating a ship target according to a ship target detection result, and obtaining a horizontal sample target slice of the ship target by using a minimum external rectangle;
step 2, M ship target slices are selected as training samples, and the ship target slices in the training samples are subjected to civil and military attribute manual judgment, wherein M is a positive integer and is not lower than the number of ship target categories;
step 3, carrying out normalization resampling on the ship target slices in the training sample to a fixed size and a gray level;
step 4, acquiring a set S containing M ship target slices, wherein each line of numerical values in the S is a one-dimensional vector after the ship target slices are unfolded;
step 5, adding the M ship target slices in the S set in corresponding dimensions, then averaging, and calculating an average image psi;
step 6, calculating a difference image of each ship target slice and the average image psi in the S set to form a difference image phi;
step 7, calculating M orthogonal unit vectors according to the difference image phi to describe the distribution of phi and form a characteristic space for judging the military and civil attributes of the ship target;
step 8, carrying out normalization resampling on the test sample to a fixed size and a gray level;
and 9, respectively carrying out characteristic space projection on the test sample and the training sample, calculating the space distance between the test sample and the training sample, and determining the military and civil attributes of the ship target according to the space distance.
Wherein, the step 1 specifically comprises the following steps:
according to a ship target detection result, a rectangular area containing a ship target is intercepted, the ship target is rotated by- α degrees based on a ship target orientation angle α obtained through ship target detection, a horizontal ship target sample is obtained, the horizontal ship target is enveloped by the minimum rectangle, and the minimum rectangle is intercepted, so that a horizontal sample target slice of the ship target is obtained.
Wherein, the step 3 specifically comprises the following steps: each ship target slice in the training sample is resampled to a 256-level gray map of fixed pixel size.
Wherein, the step 4 specifically comprises the following steps: each ship target slice in the training sample is saved as a vector Γ () of dimension 1 × N, then the S set is M × N:
S＝{Γ1,Γ2,Γ3,,ΓMand f, wherein N is the size of the fixed pixel, and M is the number of ship target slices in the training sample.
Wherein, in the step 7, M orthogonal unit vectors ulExpressed as:
Figure GDA0001494154760000031
wherein v islM eigenvectors of the L matrix as follows:
Figure GDA0001494154760000032
where Φ is the difference image.
Wherein, the step 9 specifically comprises the following steps:
(a) projecting the training samples to the feature space of step 7;
that is, the image in each training sample is represented by a feature vector, wherein the kth weight is:
Figure GDA0001494154760000033
wherein k is 1,2.. M;
the M weights form a vector, and represent the positions of the images in the training sample in the feature space:
ΩT＝[ω1,ω2,,ωM]
(b) projecting the test sample to the feature space in the step 7, and performing the same calculation method as the step (a);
(c) obtaining a training sample closest to the test sample in the characteristic space through distance calculation, and automatically judging the military and civil attributes of the test sample according to the military and civil attributes of the training sample, wherein the distance measurement is carried out by adopting the Euclidean distance:
εk＝||Ω-Ωk||2
where Ω denotes the ship to be distinguished, ΩkRepresents the kth ship target in the training set when the distance is epsilonkAnd when the number of the warships is less than the threshold value, the warship to be judged is closest to the kth warship in the training set, and the warship and the kth warship have consistent military and civil attributes.
Compared with the prior art, the invention has the following beneficial effects:
according to the method for automatically distinguishing the warship target civil and military attributes through the optical remote sensing image, only the characteristic space for distinguishing the civil and military attributes of the warship target is constructed once, and therefore the civil and military attributes of the warship target found in the test image can be automatically distinguished through the characteristic space in future work. On one hand, the warship target military and civil attributes are automatically judged through the computer, so that a great deal of energy of professional interpreters is released, and the labor cost for identifying the warship target is reduced; on the other hand, the automatic discrimination efficiency of the computer is far higher than that of manual discrimination, and the work efficiency of ship target recognition can be greatly improved by the aid of parallel calculation.
Drawings
FIG. 1 is a process flow diagram of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings 1 in conjunction with the specific embodiments.
The invention provides an automatic discrimination method for the civil and military attributes of an optical remote sensing image naval target, which is continuously carried out on the premise that the naval target is found, namely the naval target detection is finished, and comprises the following steps:
the method comprises the following steps of 1, obtaining a ship target horizontal sample, intercepting a rectangular area containing a ship target according to a ship target detection result, wherein the ship target is different in orientation, rotating the target by- α degrees based on a target orientation angle α obtained by ship target detection to obtain a horizontal ship target sample, enveloping the target by using a minimum rectangle, intercepting the minimum rectangle to obtain a horizontal sample image of the ship target, wherein the orientation of all the ship targets is consistent, more than 90% of the rectangular slice is generally a ship area, and only a small number of background pixels are arranged on two sides of a bow and a stern.
And 2, selecting M ship target slices as training samples, and manually judging the training samples. And acquiring a certain number (M) of samples from the acquired ship target total samples as training samples, wherein the training samples are required to contain ships of different types as much as possible, and manually distinguishing and labeling military and civil attributes of the training samples.
And 3, resampling the target slice in the training sample. To improve the computational efficiency and ensure comparability, each ship target slice in the training sample is resampled to a 256-level gray scale map of fixed pixel size (e.g., 20 × 100 pixels).
And 4, acquiring a set S containing M ship images. M is the ship target slice number of the training sample, where each image is stored as a 1 × N-dimensional vector Γ (N is the size of the fixed pixel in step 3, e.g., N is 20 × 100), then the S-set is M × N-dimensional.
S＝{Γ1,Γ2,Γ3,,ΓM}
And 5, calculating an average image psi. And adding the M ship images in the S set in corresponding dimensions, and averaging to obtain an average image psi of all the ship images, wherein psi is a 1 xN-dimensional vector, and if the average image psi is restored back to the image, an average ship image can be obtained.
Figure GDA0001494154760000051
And 6, calculating a difference image phi. Calculating a difference image phi of each ship image and the average image psi in the S setiI.e., subtracting the average ship from each ship pixel sequence in the S set.
Φi＝Γi-Ψ(i＝1,2,,M)
And 7, calculating M orthogonal unit vectors u for describing phi distribution and forming a characteristic space for judging the military and civil attributes of the ship target.
u-th (k-1, 2,3.. M) vectors ukCalculated by the following formula:
Figure GDA0001494154760000052
when lambda iskWhen the (eigenvalue) takes the minimum value, u is determinedkThe value is obtained.
Calculate the upper ukIn fact, the feature vector of the following covariance matrix is calculated:
Figure GDA0001494154760000061
wherein
A＝{Φ1,Φ2,Φ3,,Φn}(n＝1,2,,M)
If the number of training images is less than the dimensionality of the images (i.e., M)<N2Applicable to the case of the present invention), the eigenvector u of the covariance matrixlIt can be expressed as:
Figure GDA0001494154760000062
wherein v islM eigenvectors of the L matrix as follows:
Figure GDA0001494154760000063
and 8, carrying out normalized resampling on the test set samples according to the requirements of the step 3.
And 9, projecting the characteristic space, and automatically judging the military and civil attributes of the ship target in the test set sample.
Wherein, the step 9 specifically comprises the following steps:
(a) projecting the training set samples to the feature space of step 7;
that is, each training image is represented by a feature vector, where the kth weight is:
Figure GDA0001494154760000064
m, the M weights may form a vector that characterizes the position of the training image in the feature space:
ΩT＝[ω1,ω2,,ωM]
(b) projecting the test sample to the feature space in the step 7, and performing the same calculation method as the step (a);
(c) obtaining a training sample closest to the test sample in the characteristic space through distance calculation, automatically judging the military and civil attributes of the test sample according to the military and civil attributes of the training sample, and measuring the distance by adopting the Euclidean distance:
εk＝||Ω-Ωk||2
where Ω denotes the ship to be distinguished, ΩkRepresenting an image of a ship within the training set. When the distance is smaller than the threshold value, the ship to be judged is closest to the kth ship in the training set, and the ship to be judged and the kth ship in the training set have consistent military and civil attributes, so that the automatic identification of the military and civil attributes of the test sample is realized.
The method for automatically distinguishing the military and civil attributes of the optical remote sensing image ship target provided by the invention constructs a military and civil attribute distinguishing feature space of the ship target by obtaining a training sample, and performing a series of operations of manual distinguishing, normalization and resampling, generation of a collection, calculation of an average image, calculation of a difference image, calculation of an orthogonal vector and the like on the training sample, projects a test image to the space, and can automatically distinguish the military and civil attributes of the ship target in the test image through Euclidean distance measurement. The method only needs to construct the civil and military attribute discrimination feature space once, so that civil and military attributes of the ship targets found in the test image can be automatically discriminated by using the feature space in the future work, and the ship targets selected when constructing the civil and military attribute discrimination feature space need to cover all ship target types as much as possible. By the method, the civil and military attributes of the ship target in the newly found optical remote sensing image can be automatically distinguished, the brain of professional interpreters is liberated, and the working efficiency of ship target identification is greatly improved.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are only exemplary embodiments of the present invention and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (6)

1. A method for automatically judging the civil and military attributes of an optical remote sensing image ship target is characterized by comprising the following steps:
step 1, automatically rotating a ship target according to a ship target detection result, and obtaining a horizontal sample target slice of the ship target by using a minimum external rectangle;
step 2, M ship target slices are selected as training samples, and the ship target slices in the training samples are subjected to civil and military attribute manual judgment, wherein M is a positive integer and is not lower than the number of ship target categories;
step 3, carrying out normalization resampling on the ship target slices in the training sample to a fixed size and a gray level;
step 4, acquiring a set S containing M ship target slices, wherein each line of numerical values in the S is a one-dimensional vector of a training sample after the ship target slices are unfolded;
step 5, adding the M ship target slices in the S set in corresponding dimensions, then averaging, and calculating an average image psi;
step 6, calculating a difference image of each ship target slice and the average image psi in the S set to form a difference image phi;
step 7, calculating M orthogonal unit vectors according to the difference image phi to describe the distribution of phi and form a characteristic space for judging the military and civil attributes of the ship target;
step 8, carrying out normalization resampling on the test sample to a fixed size and a gray level;
and 9, respectively carrying out characteristic space projection on the test sample and the training sample, calculating the space distance between the test sample and the training sample, and determining the military and civil attributes of the ship target according to the space distance.
2. The method for automatically distinguishing the military and civil attributes of the optical remote sensing image ship target according to claim 1, wherein the step 1 specifically comprises the following steps:
according to a ship target detection result, a rectangular area containing a ship target is intercepted, the ship target is rotated by- α degrees based on a ship target orientation angle α obtained through ship target detection, a horizontal ship target sample is obtained, the horizontal ship target is enveloped by the minimum rectangle, and the minimum rectangle is intercepted, so that a horizontal sample target slice of the ship target is obtained.
3. The method for automatically distinguishing the military and civil attributes of the optical remote sensing image ship target according to claim 1, wherein the step 3 specifically comprises the following steps: each ship target slice in the training sample is resampled to a 256-level gray map of fixed pixel size.
4. The method for automatically distinguishing the military and civil attributes of the optical remote sensing image ship target according to claim 1, wherein the step 4 specifically comprises the following steps: each ship target slice in the training sample is saved as a vector Γ () of dimension 1 × N, then the S set is M × N:
S＝{Γ1,Γ2,Γ3,……,ΓMwhere N is the size of the fixed pixel and M is the ship mesh in the training sampleThe number of slices is marked.
5. The method for automatically distinguishing the military and civil attributes of an optical remote sensing image ship target according to claim 1, wherein in the step 7, M orthogonal unit vectors u are adoptedlExpressed as:
Figure FDA0001445818190000021
wherein v islM eigenvectors of the L matrix as follows:
Figure FDA0001445818190000022
where Φ is the difference image.
6. The method for automatically distinguishing the military and civil attributes of the optical remote sensing image ship target according to claim 1, wherein the step 9 specifically comprises the following steps:
(a) projecting the training samples to the feature space of step 7;
that is, the image in each training sample is represented by a feature vector, wherein the kth weight is:
Figure FDA0001445818190000023
wherein k is 1,2.. M;
the M weights form a vector, and represent the positions of the images in the training sample in the feature space:
ΩT＝[ω1,ω2,……,ωM]
(b) projecting the test sample to the feature space in the step 7, and performing the same calculation method as the step (a);
(c) obtaining a training sample closest to the test sample in the characteristic space through distance calculation, and automatically judging the military and civil attributes of the test sample according to the military and civil attributes of the training sample, wherein the distance measurement is carried out by adopting the Euclidean distance:
εk＝||Ω-Ωk||2
where Ω denotes the ship to be distinguished, ΩkRepresents the kth ship target in the training set when the distance is epsilonkAnd when the number of the warships is less than the threshold value, the warship to be judged is closest to the kth warship in the training set, and the warship and the kth warship have consistent military and civil attributes.
CN201711012734.1A 2017-10-26 2017-10-26 Method for automatically judging civil and military attributes of optical remote sensing image ship target Active CN107784285B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711012734.1A CN107784285B (en) 2017-10-26 2017-10-26 Method for automatically judging civil and military attributes of optical remote sensing image ship target

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711012734.1A CN107784285B (en) 2017-10-26 2017-10-26 Method for automatically judging civil and military attributes of optical remote sensing image ship target

Publications (2)

Publication Number Publication Date
CN107784285A CN107784285A (en) 2018-03-09
CN107784285B true CN107784285B (en) 2020-06-09

Family

ID=61433934

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711012734.1A Active CN107784285B (en) 2017-10-26 2017-10-26 Method for automatically judging civil and military attributes of optical remote sensing image ship target

Country Status (1)

Country Link
CN (1) CN107784285B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109241658A (en) * 2018-09-27 2019-01-18 中国电子科技集团公司第五十四研究所 Satellite dish Analysis method based on remote sensing image
CN111340040B (en) * 2020-02-26 2023-09-12 五八有限公司 Paper character recognition method and device, electronic equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102609523A (en) * 2012-02-10 2012-07-25 上海视畅信息科技有限公司 Collaborative filtering recommendation algorithm based on article sorting and user sorting
CN102881006A (en) * 2012-08-03 2013-01-16 吉林禹硕动漫游戏科技股份有限公司 Method for splicing and fusing image in multi-projection display system
CN104182755A (en) * 2014-08-30 2014-12-03 西安电子科技大学 Mammary gland molybdenum target X-ray image block feature extraction method based on tower-shaped principal component analysis (PCA)

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8600135B2 (en) * 2007-06-28 2013-12-03 Mayo Foundation For Medical Education And Research System and method for automatically generating sample points from a series of medical images and identifying a significant region

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102609523A (en) * 2012-02-10 2012-07-25 上海视畅信息科技有限公司 Collaborative filtering recommendation algorithm based on article sorting and user sorting
CN102881006A (en) * 2012-08-03 2013-01-16 吉林禹硕动漫游戏科技股份有限公司 Method for splicing and fusing image in multi-projection display system
CN104182755A (en) * 2014-08-30 2014-12-03 西安电子科技大学 Mammary gland molybdenum target X-ray image block feature extraction method based on tower-shaped principal component analysis (PCA)

Also Published As

Publication number Publication date
CN107784285A (en) 2018-03-09

Similar Documents

Publication Publication Date Title
Williams et al. Mine classification with imbalanced data
Xu et al. Automatic detection of inshore ships in high-resolution remote sensing images using robust invariant generalized Hough transform
CN103400156B (en) Based on the High Resolution SAR image Ship Detection of CFAR and rarefaction representation
CN109871902B (en) SAR small sample identification method based on super-resolution countermeasure generation cascade network
US20120206438A1 (en) Method for Representing Objects with Concentric Ring Signature Descriptors for Detecting 3D Objects in Range Images
US11681036B2 (en) Image processing apparatus, image processing method, and non-transitory computer readable medium storing image processing program
CN106295498B (en) Remote sensing image target area detection device and method
CN108564092A (en) Sunflower disease recognition method based on SIFT feature extraction algorithm
CN109816051B (en) Hazardous chemical cargo feature point matching method and system
Xu et al. Ship detection and extraction using visual saliency and histogram of oriented gradient
Yang et al. Ship detection in SAR images based on lognormal $\rho $-Metric
CN114821358A (en) Optical remote sensing image marine ship target extraction and identification method
Zhang et al. Saliency-driven oil tank detection based on multidimensional feature vector clustering for SAR images
CN107784285B (en) Method for automatically judging civil and military attributes of optical remote sensing image ship target
Shi et al. Adaptive Laplacian eigenmap-based dimension reduction for ocean target discrimination
CN107369163B (en) Rapid SAR image target detection method based on optimal entropy dual-threshold segmentation
CN110298855B (en) Sea-sky-line detection method based on Gaussian mixture model and texture analysis
Picard et al. Seafloor description in sonar images using the monogenic signal and the intrinsic dimensionality
Ch et al. Classification and Segmentation of Marine Related Remote Sensing Imagery Data Using Deep Learning
CN117115436A (en) Ship attitude detection method and device, electronic equipment and storage medium
CN109934147B (en) Target detection method, system and device based on deep neural network
Haigang et al. A novel ship detection method for large-scale optical satellite images based on visual LBP feature and visual attention model
CN115272856B (en) Ship target fine-grained identification method and equipment
Matasci et al. Deep learning for vessel detection and identification from spaceborne optical imagery
CN114332512B (en) Security check card port detection method and system based on big data

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant