CN108133218A - Infrared target detection method, equipment and medium - Google Patents
Infrared target detection method, equipment and medium Download PDFInfo
- Publication number
- CN108133218A CN108133218A CN201711338926.1A CN201711338926A CN108133218A CN 108133218 A CN108133218 A CN 108133218A CN 201711338926 A CN201711338926 A CN 201711338926A CN 108133218 A CN108133218 A CN 108133218A
- Authority
- CN
- China
- Prior art keywords
- image
- background
- infrared
- target
- obtains
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- 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
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The present invention relates to a kind of infrared target detection method, equipment and medium, the method includes:Infrared image progress super-pixel segmentation is obtained into the first image, and extract the boundary super-pixel feature of described first image as background sample collection;Dense error reconstruct is carried out to the background sample collection, obtains background notable figure;The undirected graph model based on the infrared image is built, and the sparse features for extracting the undirected graph model obtain target significant image;The background notable figure and the target significant image are merged using Bayesian formula, obtains infrared target.Infrared target detection method provided by the invention, equipment and medium, on the basis of super-pixel segmentation is carried out to infrared image, by non-directed graph model extraction sparse features, salient region is more accurately protruded, and dense error reconstructing method is combined, further improve the accuracy of testing result.
Description
Technical field
The present invention relates to a kind of methods of technical field of image processing, and in particular to a kind of infrared target detection method is set
Standby and medium.
Background technology
Infrared target detection is the core technology of infrared signal processing, is applied to infrared search and tracking (Infrared
Search and track, IRST) multiple fields such as system, precise guidance system, targeted surveillance system, satellite remote sensing system.
In recent years, the intelligent information processing method of view-based access control model attention mechanism becomes a big research hotspot.Significance analysis method due to
Its quick stationkeeping ability, is widely used in infrared target detection field.
Significance analysis method is by the way that characteristics of image and direction information measure are combined, to reduce target search range.
In terms of it is widely used in many civilian and militaries, such as image segmentation, image detection, object identification, video tracking.Conspicuousness
Detection algorithm is generally divided into visual attention detection and conspicuousness target detection.The former is notes during estimation eye-observation piece image
The variation track of viewpoint, the latter are attempt to extract entire well-marked target region.
The detection process of marking area can be converted into prospect, the segmentation problem of background.Algorithm based on background priori will scheme
As boundary is considered as pseudo- background area, the saliency value of block of pixels is obtained by the correlation for calculating block of pixels and pseudo- background area.Closely
Nian Lai, the significance analysis algorithm based on graph model attract attention when with high detection accuracy rate.Harel detects conspicuousness
Be defined as binary segmentation problem, and propose based on graph theory vision significance (graph-based visual saliency,
GBVS) model measures saliency value by being based on global characteristics method.However, the model efficiency is relatively low, it can not be clear
Recognize objective contour.Jiang assesses the conspicuousness of pixel by the uniqueness of target and the characteristic of spatial distribution of background area.
This process simplify the position fixing process of target, but it depends on borderline properties of image, and the accuracy of identification under complex background is poor.
Yang et al. calculates the manifold ranking based on background seed point, and as base by the correlation between block of pixels and background
Plinth assesses the conspicuousness of block of pixels.Manifold ranking algorithm based on graph model can obtain the reliable estimation of prospect.But
Its notable figure is relatively fuzzy and excessively emphasizes edge.
The low-level feature of image is often only utilized in the above-mentioned saliency algorithm represented based on figure, in complicated scene
Middle Boundary Extraction effect is unsatisfactory.Its testing result produces inevitable false judgment, therefore, only relies on single conspicuousness
Detection model is difficult the complete boundary of extraction.The defects of in order to overcome single conspicuousness detection algorithm, Borji is extracted RGB respectively
With the part in LAB color spaces and global image block characteristics, and it is incorporated into generating final notable figure.Similarly,
Qian describes the boundary of target area, office respectively using sparse reconstructed error, spatial domain frequency modulation contrast and Color-spatial distribution figure
Portion and global characteristics, and these characteristic informations are merged using condition random field, generate final notable figure.Gao is according to graph model structure
Asymmetric transition probability matrix is made, and is solved using Markov Random Walk Algorithms, initial Saliency maps is obtained, realizes
The fusion of low layer and high-level characteristic.Above-mentioned more conspicuousness blending algorithms make full use of the position of well-marked target, shape information, pass through
This mode obtains the notable figure that edge keeps good and prominent well-marked target entirety.However, complicated fusion rule limits
The scope of application of more conspicuousness methods.
Invention content
For the defects in the prior art, infrared target detection method provided by the invention, equipment and medium, to infrared
On the basis of image carries out super-pixel segmentation, by non-directed graph model extraction sparse features, conspicuousness area is more accurately protruded
Domain, and dense error reconstructing method is combined, further improve the accuracy of testing result.
In a first aspect, the present invention provides a kind of infrared target detection method, including:
Infrared image progress super-pixel segmentation is obtained into the first image, and the boundary super-pixel for extracting described first image is special
Sign is as background sample collection;
Dense error reconstruct is carried out to the background sample collection, obtains background notable figure;
The undirected graph model based on the infrared image is built, and the sparse features for extracting the undirected graph model obtain mesh
Mark notable figure;
The background notable figure and the target significant image are merged using Bayesian formula, obtains infrared target.
Preferably, it is described that infrared image progress super-pixel segmentation is obtained into the first image, and extract described first image
Boundary super-pixel feature as background sample collection, including:
Super-pixel segmentation is carried out to infrared image using simple linear Iterative Clustering and obtains the first image, and extract institute
The boundary super-pixel feature of the first image is stated as background sample collection.
Preferably, it is described that dense error reconstruct is carried out to the background sample collection, background notable figure is obtained, including:
Dimension-reduction treatment is carried out to the background sample collection using principal component analytical method, obtains feature vector;
Dense reconstructed error is calculated using described eigenvector;
Smoothing processing based on context is carried out to the dense reconstructed error, obtains background notable figure.
Preferably, the undirected graph model of the structure based on the infrared image, and extract the dilute of the undirected graph model
Thin feature obtains target significant image, including:
The infrared image is configured to the undirected graph model using super-pixel as node;
Row is optimized to the node in the undirected graph model using the sequence equation of non-normalized Laplce's form
Sequence;
The sparse features of undirected graph model after extraction Optimal scheduling obtain target significant image.
Preferably, it is described to merge the background notable figure and the target significant image using Bayesian formula, it obtains infrared
Target, including:
Extract the color characteristic of pixel;
The pixel number and difference color histogram in background area are counted according to the background notable figure, and from institute
The pixel number for obtaining including the color characteristic in difference color histogram is stated, obtains background observation likelihood probability;
The pixel number and color of object histogram in target area are counted according to the target significant image, and from institute
The pixel number for obtaining including the color characteristic in color of object histogram is stated, obtains target observation likelihood probability;
Likelihood probability and the target observation likelihood probability are observed according to the background, obtained using Bayesian formula infrared
Target.
Second aspect, the present invention provides a kind of infrared target detection equipment, including:At least one processor, at least one
A memory and the computer program instructions being stored in the memory, when the computer program instructions are by the processing
The method as described in any one of first aspect is realized when device performs.
The third aspect, the present invention provides a kind of computer readable storage mediums, are stored thereon with computer program, the journey
Any method in first aspect is realized when sequence is executed by processor.
Description of the drawings
The flow chart of infrared target detection method that Fig. 1 is provided by the embodiment of the present invention;
Fig. 2 is the method for the present invention and an example of the design sketch of the conspicuousness Comparison between detecting methods of other several classics;
Fig. 3 is the method for the present invention and an example of the design sketch of the conspicuousness Comparison between detecting methods of other several classics;
Fig. 4 is the method for the present invention and an example of the design sketch of the conspicuousness Comparison between detecting methods of other several classics;
Fig. 5 is the method for the present invention and an example of the design sketch of the conspicuousness Comparison between detecting methods of other several classics;
Fig. 6 is the hardware architecture diagram of infrared target detection equipment provided in an embodiment of the present invention.
Specific embodiment
The embodiment of technical solution of the present invention is described in detail below in conjunction with attached drawing.Following embodiment is only used for
Clearly illustrate technical scheme of the present invention, therefore be intended only as example, and the protection of the present invention cannot be limited with this
Range.
It should be noted that unless otherwise indicated, technical term or scientific terminology used in this application should be this hair
The ordinary meaning that bright one of ordinary skill in the art are understood.
As shown in Figure 1, a kind of infrared target detection method is present embodiments provided, including:
Infrared image progress super-pixel segmentation is obtained the first image, and extract the boundary of described first image by step S1
Super-pixel feature is as background sample collection.
Step S2 carries out dense error reconstruct to the background sample collection, obtains background notable figure.
Step S3 builds the undirected graph model based on the infrared image, and extracts the sparse spy of the undirected graph model
Obtain target significant image.
Step S4 merges the background notable figure and the target significant image using Bayesian formula, obtains infrared target.
Further, the concrete methods of realizing of the step S1 includes:(Simple is clustered using simple linear iteration
Linear Iterative Clustering, SLIC) algorithm to infrared image carry out super-pixel segmentation obtain the first image, and
The boundary super-pixel feature of described first image is extracted as background sample collection, the background sample collection is denoted as B=[b1,
b2,...,bM]。
Further, the concrete methods of realizing of the step S2 includes:
Step S201 carries out dimension-reduction treatment to the background sample collection using principal component analytical method (PCA), obtains feature
Vector is denoted as UB=[u1,u2,...,uD];
Step S202 calculates dense reconstructed error using described eigenvector;
Wherein, dense reconstructed error is calculated by the following formula:
Wherein,The average characteristics of all super-pixel are represented, by PCA substrates UBCalculate the dense of image block i (i ∈ [1, N])
Expression coefficient is βi, then pass through βiCalculate the dense reconstructed error of image block iReconstructed errorTo be normalized to [0,
1] in threshold range, consequently facilitating for describing the significance of image block.
Step S203 carries out the smoothing processing based on context to the dense reconstructed error, obtains background notable figure.
Further, the concrete methods of realizing of the step S3 includes:
The infrared image is configured to the undirected graph model using super-pixel as node by step S301.
Wherein, undirected artwork is denoted as G (V, E), and V represents node set, and E represents non-edge set.Undirected graph model G (V,
E) be a sparse connection structure, this that is, in the similarity matrix W of figure most elements be 0.If wij(wij∈W)
Weight for two neighbouring super pixels sides:
Wherein, ciAnd cjMean values of the super-pixel i and j in CIELab color spaces, σ are represented respectivelywThe intensity of control weight.
Step S302, using non-normalized Laplce's form sequence equation to the node in the undirected graph model into
Row Optimal scheduling.Specifically optimization method is:
f*=(1- α) (D- α W)-1Y,
Wherein, to angle matrix D=diag { d11,...,dnn, diagonal element dii=∑jwij, α is threshold constant, is used for
Control the influence of initial ranking value and neighborhood propagation to final ranking value.Matrix (1- α) (D- α W)-1It is used sort algorithm
The optimal similarity matrix practised, compared to original similarity matrix W, can be better described the degree of association between node.I-th of node
Ranking value f*(i) it is matrix (1- α) (D- α W)-1The i-th row and vector y inner product.Due to vectorial y be two-value label to
Amount, f*(i) value is the sum of i-th of node and all query node degrees of association.The matrix (D- α W) being calculated-1To no longer be
Sparse matrix, that is to say, that all there are a degrees of association between each two node.Feature extraction is improved by Optimal scheduling to imitate
Rate enhances the degree of association of undirected node of graph by Optimal scheduling, improve feature extraction precision.
Step S303, the sparse features for extracting the undirected graph model after Optimal scheduling obtain target significant image.
Further, the concrete methods of realizing of the step S4 includes:
Step S401 extracts the color characteristic of pixel.
Wherein, the color characteristic of pixel z is denoted as r (z).
Step S402 counts pixel number and background color Nogata in background area according to the background notable figure
Scheme, and obtain including the pixel number of the color characteristic from the difference color histogram, obtain background observation likelihood
Probability.Background observation likelihood probability is obtained especially by the following formula:
Wherein, NBRepresent the pixel number in background area, NbB(r(z))(r ∈ { L, a, b }) represents that color characteristic is fallen
The pixel number of feature r (z) is included in difference color histogram.
Step S403 counts pixel number and color of object Nogata in target area according to the target significant image
Scheme, and obtain including the pixel number of the color characteristic from the color of object histogram, obtain target observation likelihood
Probability.Target observation likelihood probability is obtained especially by the following formula:
Wherein, NFRepresent the pixel number in target area, NbF(r(z))(r ∈ { L, a, b }) represents that color characteristic is fallen
The pixel number of feature r (z) is included in color of object histogram.
Step S404 observes likelihood probability and the target observation likelihood probability according to the background, utilizes Bayes's public affairs
Formula obtains infrared target.Bayesian formula is as follows:
Wherein, p (F) is a constant, and H (z) then represents the feature vector of pixel z.
The embodiment of the present invention proposes a kind of infrared target detection side being combined based on graph model and the reconstruct of dense error
Infrared image first, closed loop figure is configured to using super-pixel segmentation by method;Secondly, it is calculated by the manifold ranking based on graph model
Target area and background are obtained two notable figures by method and the reconstruct of dense error respectively;Finally, the notable of Bayesian model is utilized
Figure syncretizing mechanism gets up two notable figure effective integrations, and fusion results are insensitive to background clutter interference, can be clearly
Position target area.When target appears near border, dense error reconstruct can reduce the possibility of erroneous judgement, enhance algorithm
Robustness.Extract for stable characterization sparse features and accurately the complete profile of target, graph model of the embodiment of the present invention
Its sparse features is extracted, target area is further accurately positioned.Therefore, the method for the present embodiment can will be shown under complex background
It writes target and background to be precisely separated, precision is better than existing algorithm.The testing result that method through this embodiment obtains can not only
Target is enough highlighted, inhibits background clutter interference, and keep the integrality of object edge.
As shown in Fig. 2~Fig. 5, the method for the present invention and the effect of the conspicuousness Comparison between detecting methods of other several classics.Fig. 2
(a), Fig. 3 (a), Fig. 4 (a), Fig. 5 (a) be infrared image, Fig. 2 (b)~Fig. 2 (h) be respectively the dense models of DSR, ITTI, GBVS,
The detection result figure of SUN, PQFT, SR and the method proposed, Fig. 3 (b)~Fig. 3 (h) be respectively the dense models of DSR, ITTI,
The detection result figure of GBVS, SUN, PQFT, SR and the method proposed, Fig. 4 (b)~Fig. 4 (h) be respectively the dense models of DSR,
The detection result figure of ITTI, GBVS, SUN, PQFT, SR and the method proposed, Fig. 5 (b)~Fig. 5 (h) are respectively DSR dense
The detection result figure of model, ITTI, GBVS, SUN, PQFT, SR and the method proposed.By comparing as can be seen that utilizing this
Inventive method, which carries out target detection, not only can significantly inhibit background clutter, and can keep the integrality of object edge.
In the objective evaluation Indexes Comparison table shown in 1~table of table 2, by Re (Recall), Sp (Specificity),
FPR (False Positive Rate), FNR (False Negative Rate) and Pre (Precision) indexs are weighed not
With the picture quality of detection method detection, Re, Sp and Pre are bigger, and FPR and FNR are smaller, illustrate that the result of detection is more accurate, effect
Fruit is better.ITTI, GBVS, SUN and PQFT model only probably determine target institute it can be seen from the data in 1~table of table 2
In position, so its Pre index is relatively low;The robustness of the dense models of DSR is poor;The target of SR model inspections is imperfect, so
Pre indexs are relatively low.In contrast, method proposed by the invention has not only highlighted target, but also maintains the complete of object edge
Whole property, index are better than other control methods, illustrate that the method for the present invention has higher accuracy and robustness.
1 first group of evaluation index of table
2 second groups of evaluation indexes of table
In addition, the infrared target detection method of the embodiment of the present invention can be realized by infrared target detection equipment.Fig. 6 shows
The hardware architecture diagram of infrared target detection equipment provided in an embodiment of the present invention is gone out.
Infrared target detection equipment can include processor 401 and be stored with the memory 402 of computer program instructions.
Specifically, above-mentioned processor 401 can include central processing unit (CPU) or specific integrated circuit
It (Application Specific Integrated Circuit, ASIC) or may be configured to implement implementation of the present invention
One or more integrated circuits of example.
Memory 402 can include the mass storage for data or instruction.For example it is unrestricted, memory
402 may include hard disk drive (Hard Disk Drive, HDD), floppy disk, flash memory, CD, magneto-optic disk, tape or logical
With the combination of universal serial bus (Universal Serial Bus, USB) driver or two or more the above.It is closing
In the case of suitable, memory 402 may include can be removed or the medium of non-removable (or fixed).In a suitable case, it stores
Device 402 can be inside or outside data processing equipment.In a particular embodiment, memory 402 is nonvolatile solid state storage
Device.In a particular embodiment, memory 402 includes read-only memory (ROM).In a suitable case, which can be mask
The ROM of programming, programming ROM (PROM), erasable PROM (EPROM), electric erasable PROM (EEPROM), electrically-alterable ROM
(EAROM) or the combination of flash memory or two or more the above.
Processor 401 is by reading and performing the computer program instructions stored in memory 402, to realize above-mentioned implementation
Any one infrared target detection method in example.
In one example, infrared target detection equipment may also include communication interface 403 and bus 410.Wherein, such as Fig. 6
Shown, processor 401, memory 402, communication interface 403 are connected by bus 410 and complete mutual communication.
Communication interface 403 is mainly used for realizing in the embodiment of the present invention between each module, device, unit and/or equipment
Communication.
Bus 410 includes hardware, software or both, and the component of infrared target detection equipment is coupled to each other together.It lifts
Unrestricted for example, bus may include accelerated graphics port (AGP) or other graphics bus, enhancing Industry Standard Architecture
(EISA) bus, Front Side Bus (FSB), super transmission (HT) interconnection, Industry Standard Architecture (ISA) bus, infinite bandwidth interconnect, are low
Number of pins (LPC) bus, memory bus, micro- channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-
Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association part (VLB) bus or
The combination of other suitable buses or two or more the above.In a suitable case, bus 410 may include one
Or multiple buses.Although specific bus has been described and illustrated in the embodiment of the present invention, the present invention considers any suitable bus
Or interconnection.
In addition, with reference to the infrared target detection method in above-described embodiment, the embodiment of the present invention can provide a kind of computer
Readable storage medium storing program for executing is realized.Computer program instructions are stored on the computer readable storage medium;The computer program refers to
Enable any one the infrared target detection method realized when being executed by processor in above-described embodiment.
Finally it should be noted that:The above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations;To the greatest extent
Pipe is described in detail the present invention with reference to foregoing embodiments, it will be understood by those of ordinary skill in the art that:Its according to
Can so modify to the technical solution recorded in foregoing embodiments either to which part or all technical features into
Row equivalent replacement;And these modifications or replacement, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme should all cover in the claim of the present invention and the range of specification.
Claims (7)
- A kind of 1. infrared target detection method, which is characterized in that including:Infrared image progress super-pixel segmentation is obtained into the first image, and the boundary super-pixel feature for extracting described first image is made For background sample collection;Dense error reconstruct is carried out to the background sample collection, obtains background notable figure;The undirected graph model based on the infrared image is built, and the sparse features for extracting the undirected graph model obtain target and show Write figure;The background notable figure and the target significant image are merged using Bayesian formula, obtains infrared target.
- 2. according to the method described in claim 1, it is characterized in that, described obtain first by infrared image progress super-pixel segmentation Image, and the boundary super-pixel feature of described first image is extracted as background sample collection, including:Super-pixel segmentation is carried out to infrared image using simple linear Iterative Clustering and obtains the first image, and extract described the The boundary super-pixel feature of one image is as background sample collection.
- 3. according to the method described in claim 1, it is characterized in that, described carry out dense error weight to the background sample collection Structure obtains background notable figure, including:Dimension-reduction treatment is carried out to the background sample collection using principal component analytical method, obtains feature vector;Dense reconstructed error is calculated using described eigenvector;Smoothing processing based on context is carried out to the dense reconstructed error, obtains background notable figure.
- 4. the according to the method described in claim 1, it is characterized in that, undirected artwork of the structure based on the infrared image Type, and the sparse features for extracting the undirected graph model obtain target significant image, including:The infrared image is configured to the undirected graph model using super-pixel as node;Sequence is optimized to the node in the undirected graph model using the sequence equation of non-normalized Laplce's form;The sparse features of undirected graph model after extraction Optimal scheduling obtain target significant image.
- 5. according to the method described in claim 1, it is characterized in that, described merge the background notable figure using Bayesian formula With the target significant image, infrared target is obtained, including:Extract the color characteristic of pixel;The pixel number and difference color histogram in background area are counted according to the background notable figure, and from the back of the body It obtains including the pixel number of the color characteristic in scape color histogram, obtains background observation likelihood probability;The pixel number and color of object histogram in target area are counted according to the target significant image, and from the mesh It obtains including the pixel number of the color characteristic in mark color histogram, obtains target observation likelihood probability;Likelihood probability and the target observation likelihood probability are observed according to the background, infrared mesh is obtained using Bayesian formula Mark.
- 6. a kind of infrared target detection equipment, which is characterized in that including:At least one processor, at least one processor and The computer program instructions being stored in the memory are realized when the computer program instructions are performed by the processor Method as described in any one of claim 1-5.
- 7. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is held by processor The method described in any one of claim 1-5 is realized during row.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711338926.1A CN108133218A (en) | 2017-12-14 | 2017-12-14 | Infrared target detection method, equipment and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711338926.1A CN108133218A (en) | 2017-12-14 | 2017-12-14 | Infrared target detection method, equipment and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108133218A true CN108133218A (en) | 2018-06-08 |
Family
ID=62389362
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711338926.1A Pending CN108133218A (en) | 2017-12-14 | 2017-12-14 | Infrared target detection method, equipment and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108133218A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109389593A (en) * | 2018-09-30 | 2019-02-26 | 内蒙古科技大学 | A kind of detection method, device, medium and the equipment of infrared image Small object |
CN109657610A (en) * | 2018-12-18 | 2019-04-19 | 北京航天泰坦科技股份有限公司 | A kind of land use change survey detection method of high-resolution multi-source Remote Sensing Images |
WO2020102988A1 (en) * | 2018-11-20 | 2020-05-28 | 西安电子科技大学 | Feature fusion and dense connection based infrared plane target detection method |
CN113177592A (en) * | 2021-04-28 | 2021-07-27 | 上海硕恩网络科技股份有限公司 | Image segmentation method and device, computer equipment and storage medium |
CN114376562A (en) * | 2021-09-10 | 2022-04-22 | 北京福乐云数据科技有限公司 | Multi-parameter artificial intelligence detector |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104751463A (en) * | 2015-03-31 | 2015-07-01 | 梁爽 | Three-dimensional model optimal visual angle selection method based on sketch outline features |
CN105427314A (en) * | 2015-11-23 | 2016-03-23 | 西安电子科技大学 | Bayesian saliency based SAR image target detection method |
CN105654475A (en) * | 2015-12-25 | 2016-06-08 | 中国人民解放军理工大学 | Image saliency detection method and image saliency detection device based on distinguishable boundaries and weight contrast |
-
2017
- 2017-12-14 CN CN201711338926.1A patent/CN108133218A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104751463A (en) * | 2015-03-31 | 2015-07-01 | 梁爽 | Three-dimensional model optimal visual angle selection method based on sketch outline features |
CN105427314A (en) * | 2015-11-23 | 2016-03-23 | 西安电子科技大学 | Bayesian saliency based SAR image target detection method |
CN105654475A (en) * | 2015-12-25 | 2016-06-08 | 中国人民解放军理工大学 | Image saliency detection method and image saliency detection device based on distinguishable boundaries and weight contrast |
Non-Patent Citations (2)
Title |
---|
CHUAN YANG等: "Saliency Detection via Graph-Based Manifold Ranking", 《2013 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
XIAOHUI LI等: "Saliency Detection via Dense and Sparse Reconstruction", 《2013 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109389593A (en) * | 2018-09-30 | 2019-02-26 | 内蒙古科技大学 | A kind of detection method, device, medium and the equipment of infrared image Small object |
WO2020102988A1 (en) * | 2018-11-20 | 2020-05-28 | 西安电子科技大学 | Feature fusion and dense connection based infrared plane target detection method |
US11055574B2 (en) | 2018-11-20 | 2021-07-06 | Xidian University | Feature fusion and dense connection-based method for infrared plane object detection |
CN109657610A (en) * | 2018-12-18 | 2019-04-19 | 北京航天泰坦科技股份有限公司 | A kind of land use change survey detection method of high-resolution multi-source Remote Sensing Images |
CN113177592A (en) * | 2021-04-28 | 2021-07-27 | 上海硕恩网络科技股份有限公司 | Image segmentation method and device, computer equipment and storage medium |
CN114376562A (en) * | 2021-09-10 | 2022-04-22 | 北京福乐云数据科技有限公司 | Multi-parameter artificial intelligence detector |
CN114376562B (en) * | 2021-09-10 | 2022-07-29 | 北京福乐云数据科技有限公司 | Multi-parameter artificial intelligence detector |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | Uncertainty estimation for stereo matching based on evidential deep learning | |
Xing et al. | An automatic learning-based framework for robust nucleus segmentation | |
CN109522908B (en) | Image significance detection method based on region label fusion | |
CN108133218A (en) | Infrared target detection method, equipment and medium | |
Li et al. | Change detection based on Gabor wavelet features for very high resolution remote sensing images | |
CN111553397B (en) | Cross-domain target detection method based on regional full convolution network and self-adaption | |
CN107633226B (en) | Human body motion tracking feature processing method | |
CN106338733B (en) | Forward-Looking Sonar method for tracking target based on frogeye visual characteristic | |
CN106815323B (en) | Cross-domain visual retrieval method based on significance detection | |
CN111476251A (en) | Remote sensing image matching method and device | |
Wang et al. | The poor generalization of deep convolutional networks to aerial imagery from new geographic locations: an empirical study with solar array detection | |
Yuan et al. | Efficient cloud detection in remote sensing images using edge-aware segmentation network and easy-to-hard training strategy | |
Xue et al. | Unsupervised change detection using multiscale and multiresolution Gaussian-mixture-model guided by saliency enhancement | |
CN111860823A (en) | Neural network training method, neural network training device, neural network image processing method, neural network image processing device, neural network image processing equipment and storage medium | |
Elkhateeb et al. | A novel coarse-to-Fine Sea-land segmentation technique based on superpixel fuzzy C-means clustering and modified Chan-Vese model | |
Alsanad et al. | Real-time fuel truck detection algorithm based on deep convolutional neural network | |
Yang et al. | An efficient automatic SAR image segmentation framework in AIS using kernel clustering index and histogram statistics | |
Liang et al. | Adaptive multiple kernel fusion model using spatial-statistical information for high resolution SAR image classification | |
Karavasilis et al. | Visual tracking using spatially weighted likelihood of Gaussian mixtures | |
Koyun et al. | Adversarial nuclei segmentation on H&E stained histopathology images | |
Abujayyab et al. | Integrating object-based and pixel-based segmentation for building footprint extraction from satellite images | |
Yang et al. | ESVC-based extraction and segmentation of texture features | |
CN116342653A (en) | Target tracking method, system, equipment and medium based on correlation filter | |
Tang et al. | Salient object detection via two-stage absorbing Markov chain based on background and foreground | |
Alahmari et al. | A review of nuclei detection and segmentation on microscopy images using deep learning with applications to unbiased stereology counting |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180608 |
|
RJ01 | Rejection of invention patent application after publication |