CN109977965A - Method and device for determining detection target in remote sensing airport image - Google Patents
Method and device for determining detection target in remote sensing airport image Download PDFInfo
- Publication number
- CN109977965A CN109977965A CN201910152523.0A CN201910152523A CN109977965A CN 109977965 A CN109977965 A CN 109977965A CN 201910152523 A CN201910152523 A CN 201910152523A CN 109977965 A CN109977965 A CN 109977965A
- Authority
- CN
- China
- Prior art keywords
- image
- airport
- key point
- detection target
- area image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 83
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000012216 screening Methods 0.000 claims abstract description 12
- 238000004891 communication Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 30
- 238000003860 storage Methods 0.000 claims description 10
- 230000001052 transient effect Effects 0.000 claims description 7
- 238000005728 strengthening Methods 0.000 claims description 4
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 claims description 2
- 239000010931 gold Substances 0.000 claims description 2
- 229910052737 gold Inorganic materials 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000031068 symbiosis, encompassing mutualism through parasitism Effects 0.000 description 2
- 108091092195 Intron Proteins 0.000 description 1
- 230000002902 bimodal effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
- G06T7/44—Analysis of texture based on statistical description of texture using image operators, e.g. filters, edge density metrics or local histograms
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides a method and a device for determining a detection target in a remote sensing airport image, wherein the method comprises the following steps: screening a region image in the remote sensing airport image to obtain a non-flat region image; regular area images in the non-flat area images are removed to obtain airport candidate area images; the regular area image is an image corresponding to a communication area with a large regular shape; determining a detection target according to the airport candidate area image and a preset model; the preset model comprises local context information describing key points of the detection target. The device performs the above method. The method and the device for determining the detection target in the remote sensing airport image provided by the embodiment of the invention can efficiently and accurately determine the detection target in the remote sensing airport image.
Description
Technical field
The present embodiments relate to technical field of remote sensing image processing, more particularly to one kind to determine in remote sensing Airport Images
Detect the method and device of target.
Background technique
Currently, Remote Sensing Image Processing Technology using more and more extensive, the target detection especially in remote sensing images is each
A field all plays an important role.
Detection target in remote sensing Airport Images, such as aircraft, since the outer dimension of aircraft differs greatly, and detection is deposited
In difficulty；In addition, remote sensing Airport Images region is big, scene is complicated, there is largely building similar with aircraft signature attribute office
The interference of portion region, easily generates a large amount of false-alarms.These factors make existing methods adaptability limited and timeliness needs to be mentioned
It is high.
Therefore, how drawbacks described above is avoided, detection target can be efficiently and accurately determined in remote sensing Airport Images, at
For that need solve the problems, such as.
Summary of the invention
In view of the problems of the existing technology, the embodiment of the present invention provides one kind and determines detection mesh in remote sensing Airport Images
Calibration method and device.
The embodiment of the present invention provides one kind and determines detection mesh calibration method in remote sensing Airport Images, comprising:
The area image in remote sensing Airport Images is screened, to obtain non-gentle area image；
The regular area image in the non-gentle area image is rejected, to obtain airport candidate regions image；It is described regular
Area image is the corresponding image in connection region with large stretch of regular shape；
According to the airport candidate regions image and preset model, detection target is determined；The preset model includes to be described
The local contextual information of the key point of the detection target.
The embodiment of the present invention provides a kind of device that detection target is determined in remote sensing Airport Images, comprising:
Screening unit, for screening the area image in remote sensing Airport Images, to obtain non-gentle area image；
Culling unit, for rejecting the regular area image in the non-gentle area image, to obtain airport candidate regions
Image；The regular area image is the corresponding image in connection region with large stretch of regular shape；
Determination unit, for determining detection target according to the airport candidate regions image and preset model；The default mould
Type includes the local contextual information for being described the key point of the detection target.
The embodiment of the present invention provides a kind of electronic equipment, comprising: processor, memory and bus, wherein
The processor and the memory complete mutual communication by the bus；
The memory is stored with the program instruction that can be executed by the processor, and the processor calls described program to refer to
Order is able to carry out following method:
The area image in remote sensing Airport Images is screened, to obtain non-gentle area image；
The regular area image in the non-gentle area image is rejected, to obtain airport candidate regions image；It is described regular
Area image is the corresponding image in connection region with large stretch of regular shape；
According to the airport candidate regions image and preset model, detection target is determined；The preset model includes to be described
The local contextual information of the key point of the detection target.
The embodiment of the present invention provides a kind of non-transient computer readable storage medium, comprising:
The non-transient computer readable storage medium stores computer instruction, and the computer instruction makes the computer
Execute following method:
The area image in remote sensing Airport Images is screened, to obtain non-gentle area image；
The regular area image in the non-gentle area image is rejected, to obtain airport candidate regions image；It is described regular
Area image is the corresponding image in connection region with large stretch of regular shape；
According to the airport candidate regions image and preset model, detection target is determined；The preset model includes to be described
The local contextual information of the key point of the detection target.
The method and device provided in an embodiment of the present invention that detection target is determined in remote sensing Airport Images, by successively dividing
Non- gentle area image, airport candidate regions image are not obtained, and according on the local of the key point comprising being described detection target
The preset model of context information detects the airport candidate regions image, so that it is determined that detection target, it can be in remote sensing Airport Images
Efficiently and accurately determine detection target.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the embodiment of the method flow chart that the present invention determines detection target in remote sensing Airport Images；
Fig. 2 is the Installation practice structural schematic diagram that the present invention determines detection target in remote sensing Airport Images；
Fig. 3 is electronic equipment entity structure schematic diagram provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
Fig. 1 is the embodiment of the method flow chart that the present invention determines detection target in remote sensing Airport Images, as shown in Figure 1,
One kind provided in an embodiment of the present invention determines detection mesh calibration method in remote sensing Airport Images, comprising the following steps:
S101: the area image in screening remote sensing Airport Images, to obtain non-gentle area image.
Specifically, the area image in device screening remote sensing Airport Images, to obtain non-gentle area image.Remote sensing airport
Image f iotaeld-of-view is big, but detection target therein, such as aircraft, and accounting is small, distribution is sparse, is largely non-detection target in visual field
Corresponding region.Since aircraft is frequently found in the flat sites such as airfield runway, airplane parking area, and such flat site is only flying
The regional area that machine occurs has significant gradient texture to change, and therefore, screens the area image in remote sensing Airport Images, Ke Yijin
One step includes rejecting the corresponding image of flat site (i.e. rejecting shoulder image), to obtain non-gentle area image.
It is possible to further obtain non-gentle area image based on gradient integrogram: remote sensing Airport Images can be calculated
Then Gradient Features image generates SAT (sum area table) look-up table, carry out integrogram to Gradient Features image and redraw,
And carry out bimodal mean value adaptivenon-uniform sampling, that is, realize the acquisition of non-shoulder.The attribute of this bianry image vegetarian refreshments is ternary
Group P [x, y, l], wherein (x, y) indicates that the coordinate position of pixel, l indicate pixel point feature.Non- shoulder l=1, gently
Region l=0.
S102: the regular area image in the non-gentle area image is rejected, to obtain airport candidate regions image；It is described
Regular area image is the corresponding image in connection region with large stretch of regular shape.
Specifically, device rejects the regular area image in the non-gentle area image, to obtain airport candidate regions figure
Picture；The regular area image is the corresponding image in connection region with large stretch of regular shape.In non-gentle area image
It further include the very intensive region of the large area such as airport heavy construction texture, it is to be appreciated that: these regions (such as boat station
Building, cargo transport area, control tower etc.) there is no detection targets.These regions are regular area image, by gradient integrogram
After reason, regular area image is usually the connection region of large stretch of regular shape, wherein large stretch of regular shape can be rectangle, no
Make specific limit.For the different feature of two o'clock attributive character before and after the pixel of above-mentioned non-gentle area image, described in rejecting
Regular area image in non-gentle area image can first pass through the cross-carried out line by line to non-gentle area image and indulge
Double dimension scannings, get regular area image, then rejected.The airport candidate regions image can be understood as the time there are aircraft
The corresponding image of favored area.
S103: according to the airport candidate regions image and preset model, detection target is determined；The preset model includes
The local contextual information of the key point of the detection target is described.
Specifically, device determines detection target according to the airport candidate regions image and preset model；The preset model
Local contextual information comprising being described the key point of the detection target.Key point can be understood as that aircraft wheel can be embodied
Wide angle point, marginal point etc..Preset model may include local context HOG feature pyramid, wherein histograms of oriented gradients
(Histogram of Oriented Gradient, referred to as " HOG ") is characterized in one kind in computer vision and image procossing
For carrying out the Feature Descriptor of object detection.HOG feature is by calculating the gradient direction histogram with statistical picture regional area
Figure carrys out constitutive characteristic.Airport candidate regions image can be directly inputted into local context HOG feature pyramid, local context
The pyramidal output result of HOG feature is to detect target.
Further, in order to more accurately determine detection target, preset model can also include being based on direction pre-estimation
DPM, wherein deformable component model (Deformable Part Model, referred to as " DPM ") is a kind of component-based
Detection algorithm.According to the airport candidate regions image and preset model, determines detection target, may include:
According to the airport candidate regions image, the local context HOG feature pyramid and the DPM, determine described in
Target is detected, is specifically comprised the following steps:
The direction pre-estimation of local suspected target: the independent communication domain point of suspected target is carried out to airport candidate regions image first
From.Aircraft parked state is three kinds in airport: independently parking, parks side by side, connecting with shelter bridge.Wherein, the aircraft independently parked
It is individually present in a manner of connected domain after candidate regions extract, the aircraft connecting with shelter bridge then has baguette and surrounding objects
Connect, be based on the above characteristic, the opening operation expanded afterwards is first corroded to the progress of candidate regions image, the tiny stick that will connect with aircraft
It eliminates, aircraft is separated at very thin point, so that each Aircraft Targets become the connected domain of individualism.Secondly, to previous step
Candidate regions image after middle opening operation operation carries out circumferencial direction scanning and estimates direction to obtain.Aircraft itself has specific knot
Structure feature (head, tail, two sides wing) therefore carries out circumferencial direction scanning, border picture by the center of circle of aircraft geometric center point
Specific trend can be presented in plain amplitude arrangement, specifically: there are 4 peak values, 4 peak valleys in intersection.It will be obtained in previous step
Connected domain carry out circumferencial direction scanning, using amplitude dash forward most apparent 4 points of different feature phase angle as doubtful Aircraft Targets
Direction estimate.
DPM candidate's aircraft identifies: DPM model training process: initially setting up trained and validation database, constructs reasonable sample
This collection.Then m corresponding root filters are trained with standard SVM algorithm, they is joined together, and be iterated optimization.
Then the component for initializing each component model, finally by coordinate descent algorithm Optimized model parameter beta.
DPM detection process: in the local context HOG feature pyramid of above-mentioned foundation, with local suspected target direction
On the basis of the direction that pre-estimation obtains, detection and identification successively is carried out to Aircraft Targets.In sequence rotation detection process, if at certain
One angle detects Aircraft Targets, then stops at the identification of the connected domain.If Aircraft Targets are not detected in estimating direction,
Then between 0 °~360 ° with every 60 ° be intervening sequences rotational communication domain, detected.Calculate local context HOG feature gold
Every layer of score in word tower chooses effective score by the method for non-maxima suppression, and is mapped back image, can be obtained really
Set the goal the rectangle frame of position, that is, detects the testing result of target.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, by successively obtaining respectively
Non- gentle area image, airport candidate regions image, and believed according to the local context of the key point comprising being described detection target
The preset model of breath detects the airport candidate regions image, so that it is determined that detection target, it can be efficient in remote sensing Airport Images, quasi-
Really determine detection target.
On the basis of the above embodiments, the method also includes:
The preset model is constructed in advance.
Specifically, device constructs the preset model in advance.Further, the preset model may include local or more
Literary HOG feature pyramid, correspondingly, described construct the preset model in advance, comprising:
Filter out all key points.
Specifically, device filters out all key points.Since aircraft is clear-cut, local message is abundant, easily extracts and obtains
Key point, such as angle point, marginal point etc. carry out key point to the airport candidate regions image in above-mentioned steps using fast algorithm and mention
It takes, thus, efficiently obtain the local prominent position in the scene of remote sensing airport.Multiple dimensioned local mode can be extracted to key point
Histogram feature (the local contextual information that this feature can effectively describe key point), in conjunction with oc-svm classifier, filters out
It include all key points of aircraft local attribute.
Each key point is weighted, and generates the local context weight for indicating the local contextual information
Figure.
Specifically, device is weighted each key point, and generate the office for indicating the local contextual information
Domain context weight map.
Further, the local context weight map for indicating the local contextual information is generated, may include:
Construct the Gaussian function for strengthening the center characteristic of each key point.
Wherein, the Gaussian function can be with are as follows:
Wherein, Fk(x, y) is the Gaussian function, can indicate that the weighted value of some key point, k are some key point, xk
And ykRespectively the abscissa and ordinate, α of this key point are weighed intensities coefficient, σkFor Gaussian function coefficient, wherein add
The specific value of power strength factor and Gaussian function coefficient can be independently arranged according to the actual situation.
The weighted value of each key point is superimposed according to the Gaussian function.
Specifically, device is superimposed the weighted value of each key point according to the Gaussian function.
It can be superimposed the weighted value of each key point according to the following formula:
Wherein, F (x, y) is superimposed weighted value, k is some key point, N is the sum of key point, Fk(x, y) is institute
State Gaussian function.
The superimposed weighted value of normalized, to obtain the local context weight map.
Specifically, the superimposed weighted value of device normalized, to obtain the local context weight map.It needs
Bright: in weighting, point off density cluster center weight value easily generates local the larger value, and local value is excessive in order to prevent, influences
Global weight map characteristic.Min-Max Normalization normalized is carried out to superimposed weighted value, so that F (x, y)
Numerical value between 0~1, the F (x, y) after normalized is local context weight map.
The local context HOG feature pyramid is constructed according to the local context weight map.
Specifically, device constructs the local context HOG feature pyramid according to the local context weight map.It can
To use following steps:
With gradient operator [- 1,0,1] and [- 1,0,1]TEach pixel in the candidate regions image of airport under the former resolution ratio of calculating
The horizontal direction gradient and vertical direction gradient of point I (x, y) and its gradient magnitude and angle:
Wherein, horizontal direction gradient are as follows:
Gv(x, y)=I (x+1, y)-I (x-1, y)；
Vertical direction gradient are as follows:
Gh(x, y)=I (x, y+1)-I (x, y-1)；
Gradient magnitude are as follows:
Angle are as follows:
Gain calculating is carried out using local area context weight map:
GF(x, y)=F (x, y) * G (x, y)；
GF(x, y) is the gradient magnitude containing aircraft local contextual information.With the gradient magnitude gradient direction after gain
Histogram, in each cell in gradient magnitude gradient orientation histogram pixel gradient magnitude and direction count, and
The histogram of gradients of each cell in block block in gradient magnitude gradient orientation histogram is normalized, is collected
HOG feature in block block, is combined into feature vector, finally constructs local context HOG characteristic pattern.
The histograms of oriented gradients of image is calculated by HOG algorithm to construct L layers of feature pyramid, it is specified that λ is sampling
Specification, the number of plies for needing to walk downwards in order to obtain twice of resolution ratio of a certain layer in pyramid are λ.And pyramid top layer is
HOG feature of the image under former resolution ratio.
It should be understood that local context weight map is weighted by the key point to aircraft, description
Be aircraft self zone and peripheral region symbiosis context relation, can reflect a possibility that there are aircrafts at this size.HOG
Feature is then by calculating the gradient orientation histogram with statistical picture regional area come constitutive characteristic, to the edge contour of object
It is described.The local context weighted value of each pixel is multiplied with gradient magnitude, on the side of succeeding target identification modeling
The symbiosis that aircraft peripheral region is introduced in edge contour feature, establishes local context HOG feature pyramid.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, described in constructing in advance
Preset model is further able in remote sensing Airport Images efficiently and accurately determine detection target.
On the basis of the above embodiments, the preset model includes local context HOG feature pyramid；Correspondingly,
It is described to construct the preset model in advance, comprising:
Filter out all key points.
Specifically, device filters out all key points.The explanation that can refer to above-described embodiment, repeats no more.
Each key point is weighted, and generates the local context weight for indicating the local contextual information
Figure.
Specifically, device is weighted each key point, and generate the office for indicating the local contextual information
Domain context weight map.The explanation that can refer to above-described embodiment, repeats no more.
The local context HOG feature pyramid is constructed according to the local context weight map.
Specifically, device constructs the local context HOG feature pyramid according to the local context weight map.It can
Referring to the explanation of above-described embodiment, repeat no more.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, by generating local or more
Literary weight map simultaneously constructs local context HOG feature pyramid, can more reasonably construct preset model, be further able to
Detection target is efficiently and accurately determined in remote sensing Airport Images.
On the basis of the above embodiments, the local context generated for indicating the local contextual information is weighed
Multigraph, comprising:
Construct the Gaussian function for strengthening the center characteristic of each key point.
Specifically, device constructs the Gaussian function for strengthening the center characteristic of each key point.It can refer to above-mentioned
The explanation of embodiment, repeats no more.
The weighted value of each key point is superimposed according to the Gaussian function.
Specifically, device is superimposed the weighted value of each key point according to the Gaussian function.It can refer to above-described embodiment
Illustrate, repeats no more.
The superimposed weighted value of normalized, to obtain the local context weight map.
Specifically, the superimposed weighted value of device normalized, to obtain the local context weight map.It can refer to
The explanation of above-described embodiment, repeats no more.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, by each key point
Weighted value, advanced optimized the building process of preset model.
On the basis of the above embodiments, the Gaussian function are as follows:
Wherein, Fk(x, y) is the Gaussian function, k is some key point, xkAnd ykThe respectively horizontal seat of this key point
Mark and ordinate, α are weighed intensities coefficient, σkFor Gaussian function coefficient.
Specifically, the Gaussian function in device are as follows:
Wherein, Fk(x, y) is the Gaussian function, k is some key point, xkAnd ykThe respectively horizontal seat of this key point
Mark and ordinate, α are weighed intensities coefficient, σkFor Gaussian function coefficient.The explanation that can refer to above-described embodiment, repeats no more.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, pass through specific expression formula
Gaussian function is limited, is further able to more reasonably construct preset model.
On the basis of the above embodiments, the weighted value that each key point is superimposed according to the Gaussian function, comprising:
It is superimposed the weighted value of each key point according to the following formula:
Wherein, F (x, y) is superimposed weighted value, k is some key point, N is the sum of key point, Fk(x, y) is institute
State Gaussian function.
Specifically, device is superimposed the weighted value of each key point according to the following formula:
Wherein, F (x, y) is superimposed weighted value, k is some key point, N is the sum of key point, Fk(x, y) is institute
State Gaussian function.The explanation that can refer to above-described embodiment, repeats no more.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, pass through specific Superposition Formula
The superposition calculation for carrying out the weighted value of each key point is further able to more reasonably construct preset model.
On the basis of the above embodiments, preset model further includes the DPM based on direction pre-estimation；Correspondingly, described
According to the airport candidate regions image and preset model, detection target is determined, comprising:
According to the airport candidate regions image, the local context HOG feature pyramid and the DPM, determine described in
Detect target.
Specifically, device is according to the airport candidate regions image, the local context HOG feature pyramid and described
DPM determines the detection target.The explanation that can refer to above-described embodiment, repeats no more.
It is provided in an embodiment of the present invention that detection mesh calibration method is determined in remote sensing Airport Images, pass through local context
HOG feature pyramid and DPM determine the detection target, are further able to efficiently and accurately determine in remote sensing Airport Images
Detect target.
Fig. 2 is the Installation practice structural schematic diagram that the present invention determines detection target in remote sensing Airport Images, such as Fig. 2 institute
Show, the embodiment of the invention provides a kind of in remote sensing Airport Images determines the device of detection target, including screening unit 201,
Culling unit 202 and determination unit 203, in which:
Screening unit 201 is used to screen the area image in remote sensing Airport Images, to obtain non-gentle area image；It rejects
Unit 202 is used to reject the regular area image in the non-gentle area image, to obtain airport candidate regions image；The rule
Main plot area image is the corresponding image in connection region with large stretch of regular shape；Determination unit 203 is used for according to the airport
Candidate regions image and preset model determine detection target；The preset model includes the key point for being described the detection target
Local contextual information.
Specifically, screening unit 201 is used to screen the area image in remote sensing Airport Images, to obtain non-gentle administrative division map
Picture；Culling unit 202 is used to reject the regular area image in the non-gentle area image, to obtain airport candidate regions figure
Picture；The regular area image is the corresponding image in connection region with large stretch of regular shape；Determination unit 203 is used for basis
The airport candidate regions image and preset model determine detection target；The preset model includes to be described the detection target
Key point local contextual information.
The device provided in an embodiment of the present invention that detection target is determined in remote sensing Airport Images, by successively obtaining respectively
Non- gentle area image, airport candidate regions image, and believed according to the local context of the key point comprising being described detection target
The preset model of breath detects the airport candidate regions image, so that it is determined that detection target, it can be efficient in remote sensing Airport Images, quasi-
Really determine detection target.
The device provided in an embodiment of the present invention that detection target is determined in remote sensing Airport Images specifically can be used for executing
The process flow of above-mentioned each method embodiment, details are not described herein for function, is referred to retouching in detail for above method embodiment
It states.
Fig. 3 is electronic equipment entity structure schematic diagram provided in an embodiment of the present invention, as shown in figure 3, the electronic equipment
It include: processor (processor) 301, memory (memory) 302 and bus 303；
Wherein, the processor 301, memory 302 complete mutual communication by bus 303；
The processor 301 is used to call the program instruction in the memory 302, to execute above-mentioned each method embodiment
Provided method, for example, the area image in screening remote sensing Airport Images, to obtain non-gentle area image；It rejects
Regular area image in the non-gentle area image, to obtain airport candidate regions image；The regular area image is tool
There is the corresponding image in connection region of large stretch of regular shape；According to the airport candidate regions image and preset model, detection is determined
Target；The preset model includes the local contextual information for being described the key point of the detection target.
The present embodiment discloses a kind of computer program product, and the computer program product includes being stored in non-transient calculating
Computer program on machine readable storage medium storing program for executing, the computer program include program instruction, when described program instruction is calculated
When machine executes, computer is able to carry out method provided by above-mentioned each method embodiment, for example, screening remote sensing Airport Images
In area image, to obtain non-gentle area image；The regular area image in the non-gentle area image is rejected, to obtain
Take airport candidate regions image；The regular area image is the corresponding image in connection region with large stretch of regular shape；According to
The airport candidate regions image and preset model determine detection target；The preset model includes to be described the detection target
Key point local contextual information.
The present embodiment provides a kind of non-transient computer readable storage medium, the non-transient computer readable storage medium
Computer instruction is stored, the computer instruction makes the computer execute method provided by above-mentioned each method embodiment, example
It include: such as the area image screened in remote sensing Airport Images, to obtain non-gentle area image；Reject the non-gentle administrative division map
Regular area image as in, to obtain airport candidate regions image；The regular area image is with large stretch of regular shape
The corresponding image in connection region；According to the airport candidate regions image and preset model, detection target is determined；The preset model
Local contextual information comprising being described the key point of the detection target.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can be stored in a computer readable storage medium, the program
When being executed, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes: ROM, RAM, magnetic disk or light
The various media that can store program code such as disk.
The apparatus embodiments described above are merely exemplary, wherein described, unit can as illustrated by the separation member
It is physically separated with being or may not be, component shown as a unit may or may not be physics list
Member, it can it is in one place, or may be distributed over multiple network units.It can be selected according to the actual needs
In some or all of the modules achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying creativeness
Labour in the case where, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on
Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should
Computer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingers
It enables and using so that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementation
Method described in certain parts of example or embodiment.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (10)
1. one kind determines detection mesh calibration method in remote sensing Airport Images characterized by comprising
The area image in remote sensing Airport Images is screened, to obtain non-gentle area image；
The regular area image in the non-gentle area image is rejected, to obtain airport candidate regions image；The regular region
Image is the corresponding image in connection region with large stretch of regular shape；
According to the airport candidate regions image and preset model, detection target is determined；The preset model is described comprising being described
Detect the local contextual information of the key point of target.
2. the method according to claim 1, wherein the method also includes:
The preset model is constructed in advance.
3. according to the method described in claim 2, it is characterized in that, the preset model includes local context HOG feature gold
Word tower；Correspondingly, described construct the preset model in advance, comprising:
Filter out all key points；
Each key point is weighted, and generates the local context weight map for indicating the local contextual information；
The local context HOG feature pyramid is constructed according to the local context weight map.
4. according to the method described in claim 3, it is characterized in that, the generation is for indicating the local contextual information
Local context weight map, comprising:
Construct the Gaussian function for strengthening the center characteristic of each key point；
The weighted value of each key point is superimposed according to the Gaussian function；
The superimposed weighted value of normalized, to obtain the local context weight map.
5. according to the method described in claim 4, it is characterized in that, the Gaussian function are as follows:
Wherein, Fk(x, y) is the Gaussian function, k is some key point, xkAnd ykRespectively the abscissa of this key point and
Ordinate, α are weighed intensities coefficient, σkFor Gaussian function coefficient.
6. according to the method described in claim 4, it is characterized in that, described be superimposed each key point according to the Gaussian function
Weighted value, comprising:
It is superimposed the weighted value of each key point according to the following formula:
Wherein, F (x, y) is superimposed weighted value, k is some key point, N is the sum of key point, Fk(x, y) is the height
This function.
7. method according to any one of claims 1 to 6, which is characterized in that preset model further includes based on direction pre-estimation
DPM；Correspondingly, it is described according to the airport candidate regions image and preset model, determine detection target, comprising:
According to the airport candidate regions image, the local context HOG feature pyramid and the DPM, the detection is determined
Target.
8. a kind of device for determining detection target in remote sensing Airport Images characterized by comprising
Screening unit, for screening the area image in remote sensing Airport Images, to obtain non-gentle area image；
Culling unit, for rejecting the regular area image in the non-gentle area image, to obtain airport candidate regions image；
The regular area image is the corresponding image in connection region with large stretch of regular shape；
Determination unit, for determining detection target according to the airport candidate regions image and preset model；The preset model packet
Local contextual information containing the key point for being described the detection target.
9. a kind of electronic equipment characterized by comprising processor, memory and bus, wherein
The processor and the memory complete mutual communication by the bus；
The memory is stored with the program instruction that can be executed by the processor, and the processor calls described program to instruct energy
Enough methods executed as described in claim 1 to 7 is any.
10. a kind of non-transient computer readable storage medium, which is characterized in that the non-transient computer readable storage medium is deposited
Computer instruction is stored up, the computer instruction makes the computer execute the method as described in claim 1 to 7 is any.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910152523.0A CN109977965B (en) | 2019-02-28 | 2019-02-28 | Method and device for determining detection target in remote sensing airport image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910152523.0A CN109977965B (en) | 2019-02-28 | 2019-02-28 | Method and device for determining detection target in remote sensing airport image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109977965A true CN109977965A (en) | 2019-07-05 |
| CN109977965B CN109977965B (en) | 2021-04-13 |
Family
ID=67077656
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910152523.0A Active CN109977965B (en) | 2019-02-28 | 2019-02-28 | Method and device for determining detection target in remote sensing airport image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109977965B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705427A (en) * | 2019-09-25 | 2020-01-17 | 中国人民解放军61646部队 | Extraction processing method and device for remote sensing image target area |
| CN111862006A (en) * | 2020-06-30 | 2020-10-30 | 北京北方智图信息技术有限公司 | Detection method and device for small aircraft |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101650784A (en) * | 2009-09-23 | 2010-02-17 | 南京大学 | Method for matching images by utilizing structural context characteristics |
| US7853072B2 (en) * | 2006-07-20 | 2010-12-14 | Sarnoff Corporation | System and method for detecting still objects in images |
| CN102214298A (en) * | 2011-06-20 | 2011-10-12 | 复旦大学 | Method for detecting and identifying airport target by using remote sensing image based on selective visual attention mechanism |
| CN103186790A (en) * | 2011-12-27 | 2013-07-03 | 北京三星通信技术研究有限公司 | Object detecting system and object detecting method |
| CN103514448A (en) * | 2013-10-24 | 2014-01-15 | 北京国基科技股份有限公司 | Method and system for navicular identification |
| CN104200236A (en) * | 2014-08-22 | 2014-12-10 | 浙江生辉照明有限公司 | Quick target detection method based on DPM (deformable part model) |
| CN105046224A (en) * | 2015-07-16 | 2015-11-11 | 东华大学 | Block self-adaptive weighted histogram of orientation gradient feature based face recognition method |
| CN106067022A (en) * | 2016-05-28 | 2016-11-02 | 北方工业大学 | Remote sensing image harbor ship detection false alarm eliminating method based on DPM algorithm |
| CN106778835A (en) * | 2016-11-29 | 2017-05-31 | 武汉大学 | The airport target by using remote sensing image recognition methods of fusion scene information and depth characteristic |
| CN108446707A (en) * | 2018-03-06 | 2018-08-24 | 北方工业大学 | Remote sensing image airplane detection method based on key point screening and DPM confirmation |
| CN108564602A (en) * | 2018-04-16 | 2018-09-21 | 北方工业大学 | Airplane detection method based on airport remote sensing image |
-
2019
- 2019-02-28 CN CN201910152523.0A patent/CN109977965B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7853072B2 (en) * | 2006-07-20 | 2010-12-14 | Sarnoff Corporation | System and method for detecting still objects in images |
| CN101650784A (en) * | 2009-09-23 | 2010-02-17 | 南京大学 | Method for matching images by utilizing structural context characteristics |
| CN102214298A (en) * | 2011-06-20 | 2011-10-12 | 复旦大学 | Method for detecting and identifying airport target by using remote sensing image based on selective visual attention mechanism |
| CN103186790A (en) * | 2011-12-27 | 2013-07-03 | 北京三星通信技术研究有限公司 | Object detecting system and object detecting method |
| CN103514448A (en) * | 2013-10-24 | 2014-01-15 | 北京国基科技股份有限公司 | Method and system for navicular identification |
| CN104200236A (en) * | 2014-08-22 | 2014-12-10 | 浙江生辉照明有限公司 | Quick target detection method based on DPM (deformable part model) |
| CN105046224A (en) * | 2015-07-16 | 2015-11-11 | 东华大学 | Block self-adaptive weighted histogram of orientation gradient feature based face recognition method |
| CN106067022A (en) * | 2016-05-28 | 2016-11-02 | 北方工业大学 | Remote sensing image harbor ship detection false alarm eliminating method based on DPM algorithm |
| CN106778835A (en) * | 2016-11-29 | 2017-05-31 | 武汉大学 | The airport target by using remote sensing image recognition methods of fusion scene information and depth characteristic |
| CN108446707A (en) * | 2018-03-06 | 2018-08-24 | 北方工业大学 | Remote sensing image airplane detection method based on key point screening and DPM confirmation |
| CN108564602A (en) * | 2018-04-16 | 2018-09-21 | 北方工业大学 | Airplane detection method based on airport remote sensing image |
Non-Patent Citations (2)
| Title |
|---|
| JUNHYUK HYUN，AND ETC: "Proposing a Fast Circular HOG Descriptor for Detecting Rotated Objects", 《2015 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN)》 * |
| 尹宏鹏等: "基于视觉的目标检测与跟踪综述", 《自动化学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705427A (en) * | 2019-09-25 | 2020-01-17 | 中国人民解放军61646部队 | Extraction processing method and device for remote sensing image target area |
| CN111862006A (en) * | 2020-06-30 | 2020-10-30 | 北京北方智图信息技术有限公司 | Detection method and device for small aircraft |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109977965B (en) | 2021-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108596101B (en) | Remote sensing image multi-target detection method based on convolutional neural network | |
| CN108416307B (en) | Method, device and equipment for detecting pavement cracks of aerial images | |
| CN106874894B (en) | Human body target detection method based on regional full convolution neural network | |
| CN106599869B (en) | A kind of vehicle attribute recognition methods based on multitask convolutional neural networks | |
| CN111914795B (en) | Method for detecting rotating target in aerial image | |
| CN109165623B (en) | Rice disease spot detection method and system based on deep learning | |
| CN104166841B (en) | The quick detection recognition methods of pedestrian or vehicle is specified in a kind of video surveillance network | |
| CN108875600A (en) | A kind of information of vehicles detection and tracking method, apparatus and computer storage medium based on YOLO | |
| CN107742099A (en) | A kind of crowd density estimation based on full convolutional network, the method for demographics | |
| CN110930387A (en) | Fabric defect detection method based on depth separable convolutional neural network | |
| CN109670503A (en) | Label detection method, apparatus and electronic system | |
| CN110378297A (en) | A kind of Remote Sensing Target detection method based on deep learning | |
| CN109255286A (en) | A kind of quick detection recognition method of unmanned plane optics based on YOLO deep learning network frame | |
| CN109948593A (en) | Based on the MCNN people counting method for combining global density feature | |
| CN104537689B (en) | Method for tracking target based on local contrast conspicuousness union feature | |
| CN112070729A (en) | Anchor-free remote sensing image target detection method and system based on scene enhancement | |
| CN108446707A (en) | Remote sensing image airplane detection method based on key point screening and DPM confirmation | |
| CN108932273A (en) | Picture screening technique and device | |
| CN112990086A (en) | Remote sensing image building detection method and device and computer readable storage medium | |
| CN110969171A (en) | Image classification model, method and application based on improved convolutional neural network | |
| CN109886937A (en) | Defects of insulator detection method based on super-pixel segmentation image recognition | |
| CN102819747B (en) | Method for automatically classifying forestry service images | |
| CN115049948B (en) | Unmanned aerial vehicle inspection method and device based on neural network model and related equipment | |
| CN109977965A (en) | Method and device for determining detection target in remote sensing airport image | |
| CN112101189A (en) | SAR image target detection method and test platform based on attention mechanism |
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 |