CN106407902A - Geometric difference-based airplane object identification method - Google Patents
Geometric difference-based airplane object identification method Download PDFInfo
- Publication number
- CN106407902A CN106407902A CN201610789468.2A CN201610789468A CN106407902A CN 106407902 A CN106407902 A CN 106407902A CN 201610789468 A CN201610789468 A CN 201610789468A CN 106407902 A CN106407902 A CN 106407902A
- Authority
- CN
- China
- Prior art keywords
- polygon
- aircraft
- profile
- identified
- point
- 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
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Library & Information Science (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a geometric difference-based airplane object identification method. A contour is extracted after an image of an object is subjected to image processing operation, an approximation polygon of the contour of the object is constructed via a minimum perimeter polygon algorithm, the area of the approximation polygon is enabled to be equal to template area after the approximation polygon is enlarged or shrunk according to a template database standard object, the approximation polygon is enabled to be partially superposed with template database figures via an arrangement algorithm, and identification operation can be conducted. That the method is simple and easy to implement is verified via a simulation experiment, the object can be identified quickly, high accuracy can be realized, and the geometric difference-based airplane object identification method has anti-geometric deformation and anti-noise performance.
Description
Technical field
The invention belongs to field of target recognition, relate generally to the extraction of aircraft geometric properties and the identification of Aircraft Targets.
Background technology
Aircraft Target Identification is by being analyzed in real time to the photo captured by imaging radar or reconnaissance satellite, to determine
Aircraft Targets.Existing most of recognition methodss need to carry out feature extraction to target, further according to the feature extracted using each
Plant algorithm to be mated with aircraft signature in template base, be primarily present problems with:
(1) target characteristic extracting requires in principle with invariance, but sends out when shooting azimuth, gray scale, contrast etc.
Changing can change often, that is, it is non-existent for having absolute Invariance feature vector, needs extraction to have relatively stable
The characteristic vector of property;
(2) the extraction room and time complexity of complex characteristic is high, calculates complicated, reduces recognition efficiency, therefore will extract
Feature preferably simply easily extracts feature;
(3) with the appearance of type aircraft, the three dimensional structure of aircraft becomes more complicated, and aircraft shoots under different attitudes
Image difference very big, to feature extraction and aircraft identification bring bigger challenge.
Therefore, set up an accuracy rate height, to have anti-geometric deformation and the Aircraft Recognition of noiseproof feature be to have emphatically
Big meaning.
Content of the invention
In order to overcome the deficiencies in the prior art, the present invention provides a kind of Aircraft Target Recognition based on disparity,
Target can fast and effeciently be identified.
The technical solution adopted for the present invention to solve the technical problems comprises the following steps:
(1) utilize certain direction view outline polygon of aircraft, obtain the profile of other direction views by affine transformation
Polygon;Travel through each model aircraft, set up aircraft multi-pose model library;
(2) normalizing is carried out to the aircraft profile in the aircraft profile in aircraft brake disc to be identified and aircraft multi-pose model library
Change and process so that the aircraft profile in aircraft brake disc to be identified has identical with the aircraft profile in aircraft multi-pose model library
Area;
(3) aircraft outline polygon overlap processing, comprises the following steps:
(3.1) the aircraft profile in aircraft brake disc to be identified is designated as polygon A, the aircraft in aircraft multi-pose model library
Profile is designated as polygon B;Calculate center of gravity O of polygon A respectivelyABarycentric coodinates O with polygon BB;
(3.2) polygon B is rotated and translated certain a line each bar with polygon A successively so that polygon B
Side is overlapping and midpoint overlaps, each bar side of traversal polygon B, calculates weighing for i-th of polygon A in j-th with polygon B
O when foldedA、OBApart from dI, j;
(3.3) will be apart from dI, jWith default threshold value d*Compare, record meets dI, j<d*When corresponding polygon B attitude;
(4) identify target aircraft, comprise the following steps:
(4.1) ask for the common factor of polygon A and polygon B, obtain common factor polygon;
(4.2) ask for the polygonal area S that occurs simultaneouslyIntersectionAnd the area S of polygon A;
(4.3) seek template similarity coefficient ξ=SIntersection/S;
(4.4) the aircraft profile in the maximum aircraft multi-pose model library of template similarity coefficient, corresponding aircraft class are chosen
Type is target to be identified and corresponds to target.
Before described step (2), Image semantic classification is carried out to aircraft brake disc to be identified, including image gray processing, figure
As smooth, image filtering, image enhaucament, Image Edge-Detection and polygonal segments.
Described step (2) comprises the following steps:
(2.1) the aircraft profile in aircraft brake disc to be identified is designated as polygon A, the aircraft in aircraft multi-pose model library
Profile is designated as polygon B;Calculate the area S of polygon A and polygon B respectivelyA、SB;Likelihood ratio k=B/A of polygon A, B;
(2.2) take up an official post from polygon A place plane and take each summit of 1 point of O, junction point O and polygon A, calculate every company
Line length li, i=1,2,3,4 ..., n, n represent polygon vertex number, take point B on every line or its extended line respectively1、
B2、B3、…、BnSo that each point and point O are apart from Li=k*li;
(2.3) junction point B1、B2、B3、…、Bn, obtain the polygon after normalization.
The invention has the beneficial effects as follows:The Target Photo getting is extracted profile after image procossing, and with Xiao Zhou
Long polygon algorithm constructs the approximate polygon of objective contour;Then zoomed in or out after process according to template base standard target
Make its area equal with template area;Reuse and put algorithm and so that it is overlapped with template base visuals;And a kind of improvement is proposed
Type doubly linked list algorithm seeks polygon intersection, reaches the purpose of identification image by calculating intersection size.This
It is simple that invention demonstrates the method through emulation experiment, can quickly identify target, and accuracy rate is high, has anti-geometric deformation
And noiseproof feature.
Brief description
Fig. 1 is method of the present invention flow chart;
Fig. 2 is intersection point schematic diagram in embodiment;
Fig. 3 is aircraft various putting position schematic diagram in embodiment;
Fig. 4 is the method flow diagram traveling through each intersection point in embodiment.
Specific embodiment
The present invention is further described with reference to the accompanying drawings and examples, and the present invention includes but are not limited to following enforcements
Example.
The present invention provides a kind of Aircraft Target Identification algorithm based on disparity, and its step is:
(1) aircraft single-view multi-pose model library is set up:Using certain direction view outline polygon of aircraft, with affine
Conversion obtains other multi-angled view it is established that multiple view model storehouse;
(2) target aircraft Image semantic classification:To aircraft brake disc to be identified, carry out pretreatment first, to acquired original figure
As being processed, reduce the impacts to subsequent treatment for the invalid information such as noise, strengthen useful information, improve picture contrast;
(3) aircraft profile standardization:With template base multi-pose aircraft, similar polygon principle is utilized to target aircraft profile
Do normalized so as to have identical area;
(4) aircraft outline polygon overlap processing:Make the aircraft polygon in aircraft to be identified and template base by a set pattern
Then overlapping, and by center of gravity restriction method, obtained overlap mode data base is simplified;
(5) identify target aircraft:Travel through storage intersection point using doubly linked list and ask for aircraft profile intersection of polygon, according to friendship
Collection area identifies aircraft with the area ratio of former outline polygon.
As a further improvement on the present invention, the idiographic flow of described step (1) is:
(1.1) collect and obtain all kinds of aircraft pictures, obtain the outline polygon of all kinds of aircrafts after processing;
(1.2) model aircraft image outline is done the attitude figure that affine transformation can be obtained by its each attitude, thus setting up
Play aircraft multi-pose model library.
As a further improvement on the present invention, the idiographic flow of described step (1.2) is:
(1.2.1) set up affine Transform Model
Wherein,Det (A) ≠ 0,
(1.2.2) rotation and translation transformation set:a1=b2=cosj ,-a2=b1=sin j;
(1.2.3) homothetic conversion (change of scale in image) sets:a1=b2=T, a2=b1=0, a=Z=0;
(1.2.4) shear transformation sets:a1=b2=1, a2=0, b1=k, a=Z=0.
As a further improvement on the present invention, the idiographic flow of described step (3) is:
(3.1) calculate the area S of polygon A and polygon BA、SB.The likelihood ratio of polygon A, B is k (B/A);
(3.2) take up an official post from plane and take 1 point of O, connect each summit of this point and polygon A, calculate every wire length li(i
=1,2,3,4 ..., n) (n represents polygon vertex number), pressed the direction prolongation away from O and (worked as k<In O and summit line when 1
On take a little) make Li=k*li, obtain point B1、B2、B3、…、Bn;
(3.3) junction point B1、B2、B3、…、BnIt is amplified the polygon after (or reducing).
As a further improvement on the present invention, the idiographic flow of described step (4) is:
(4.1) calculate center of gravity O of polygon AAAnd barycentric coodinates O of polygon BB;
(4.2) make B through rotation, after translation from a line start successively with A first, second ... until last
A line is overlapping and midpoint overlaps, and calculates and put rear O each timeA、OBApart from dI, j(i, j) (represents that i-th side of A falls
On j-th side of B);
(4.3) centroidal distance threshold value d is set*;
(4.4) d will be metI, j<d*Adjustment putting position polygon be sequentially output.
As a further improvement on the present invention, the idiographic flow of described step (5) is:
(5.1) may have access to forerunner using doubly linked list and the construction featuress of descendant node and easy insertion point ask for be measured flying
Machine polygon and template base aircraft intersection of polygon;
(5.2) ask for common factor area of a polygon SIntersectionAnd aircraft area of a polygon S to be measured;
(5.3) seek template similarity coefficient ξ=SIntersection/S;
(5.4) compare similarity coefficient, what similarity coefficient was maximum is target correspondence target to be identified.
As a further improvement on the present invention, the idiographic flow of described step (5.1) is:
(5.1.1) defining polygon A is aircraft polygon to be measured, and polygon B is template base aircraft polygon;
(5.1.2) A vertex sequence is put into chain Table A according to sequence counter-clockwise, by B vertex sequence according to counter-clockwise
Put into chained list B;
(5.1.3) A and B is made to be put according to described in step (4);
(5.1.4) obtain all intersecting point coordinates of A, B, and intersection point is inserted into A, B chained list;
(5.1.5) travel through intersection point;
(5.1.6) judge whether that all intersection points are all traversed, without then going to step (5.1.5), otherwise output is handed over
Collection polygon.
As shown in figure 4, the idiographic flow of described step (5.1.5) is:
(5.1.5.1) an access point intersection point entering into B from A is selected to begin stepping through chained list;
(5.1.5.2) when reaching next intersection point, judge whether intersection point is former chained list polygon vertex, go to another if not
One chained list continues traversal, otherwise judges whether this intersection point is another chained list polygon vertex again, if otherwise along former polygonal chain
Table continues traversal, otherwise judges whether with intersection point as starting point two polygonal sides overlap, goes to another polygon if not
Chained list continues traversal, and otherwise continuing traversal to next intersection point along former polygon chained list is overlapping line segment end, then judges along the inverse time
When pin moves towards, whether intersection point is access point with respect to another polygon, if it is continues traversal along former polygon chained list, otherwise turns
To another polygon traversal;
(5.1.5.3) judging whether intersection point is intersection point selected by (5.1.5.1), if otherwise going to (5.1.5.2), otherwise tying
Bundle traversal, obtains a polygonal common factor chained list.
Embodiments of the invention comprise the following steps:
Step 1, aircraft single-view multi-pose model library are set up.Using certain direction view outline polygon of aircraft, use
Affine transformation obtains other multi-angled view it is established that multiple view model storehouse;
Concrete steps are carried out as follows:
1.1 collections obtain all kinds of aircraft pictures, obtain the outline polygon of all kinds of aircrafts after processing;
Model aircraft image outline is done the attitude figure that affine transformation can be obtained by its each attitude, thus setting up by 1.2
Aircraft multi-pose model library.
Step 2, target aircraft Image semantic classification.To aircraft brake disc to be identified, carry out pretreatment first, to acquired original
Image is processed, and reduces the impacts to subsequent treatment for the invalid information such as noise, strengthens useful information, improves picture contrast.
Step 3, aircraft profile standardization:Utilize similar polygon former in target aircraft profile and template base multi-pose aircraft
Reason does normalized so as to have identical area;
Concrete steps are carried out as follows:
The 3.1 area S calculating polygon A and polygon BA、SB.The likelihood ratio of polygon A, B is k (B/A);
3.2 take up an official post from plane takes 1 point of O, connects each summit of this point and polygon A, calculates every wire length li(i=
1,2,3,4 ..., n) (n represents polygon vertex number), pressed the direction prolongation away from O and (worked as k<When 1 on O with summit line
Take a little) make Li=k*li, obtain point B1、B2、B3、…、Bn;
3.3 junction point B1、B2、B3、…、BnIt is amplified the polygon after (or reducing).
Step 4, aircraft outline polygon overlap processing.Make the aircraft polygon in aircraft to be identified and template base by certain
Rule is overlapping, and by center of gravity restriction method, obtained overlap mode data base is simplified;
Concrete steps are carried out as follows:
Center of gravity O of 4.1 calculating polygon AAAnd barycentric coodinates O of polygon BB;
4.2 make B through rotation, after translation from a line start successively with A first, second ... until last
Bar side is overlapping and midpoint overlaps, and calculates and put rear O each timeA、OBApart from dI, j(i, j) (represents that i-th side of A falls in B
J-th side on);
4.3 setting centroidal distance threshold values d*=15;
4.4 will meet dI, j<d*Adjustment putting position polygon be sequentially output, as shown in figure 3, in figure is various putting
The parameter of position is as shown in the table:
(a1) | (a2) | (a3) | (a4) | |
Aircraft A overlap side | 5 | 6 | 12 | 16 |
Aircraft B overlap side | 4 | 2 | 17 | 34 |
Centroidal distance | 3.579 | 6.107 | 3.961 | 12.590 |
Common factor area of a polygon | 22231.6 | 17705.2 | 15653.0 | 17568.3 |
Aircraft area of a polygon | 25052 | 25052 | 25052 | 25052 |
Common factor ratio | 0.887 | 0.707 | 0.625 | 0.701 |
(b1) | (b2) | (b3) | (b4) | |
Aircraft A overlap side | 16 | 18 | 19 | 0 |
Aircraft B overlap side | 33 | 4 | 33 | 0 |
Centroidal distance | 7.911 | 8.734 | 8.734 | 0 |
Common factor area of a polygon | 17751.4 | 17704.0 | 17704.0 | 27149.9995 |
Aircraft area of a polygon | 27150 | 27150 | 27150 | 17150 |
Common factor ratio | 0.654 | 0.652 | 0.652 | 1 |
Step 5, identification target aircraft.Travel through storage intersection point using doubly linked list and ask for aircraft profile intersection of polygon, root
Area ratio identification aircraft according to common factor area and former outline polygon.Specifically carry out as follows:
5.1 ask for common factor polygon using doubly linked list.Specifically carry out as follows:
5.1.1 setting A is aircraft polygon to be measured, and B is template base polygon;
5.1.2 generate the corresponding chained list of polygon A, B:Chained list A={ a1, a2..., a38, chained list B={ b1, b2...,
b38};
5.1.3 according to the algorithm of putting limiting based on center of gravity, suitable disposing way, such as the 6th side of figure A and the 2nd of B are chosen
Overlapping (while start counting up from 0);
5.1.4 all intersection points of A are obtained:I={ I1, I2..., I17, intersection point is inserted in A, B chained list, obtains new A, B
Chained list:
A '={ a1..., a6, I1, a7, I2, I3, a8..., a11, I5, a12, a13(I4), a14, a17(I6), a19, I7, a20...,
a24(I8) ..., a26, I11, a27, I10, a28, I9..., a32(I12), I13, a33, I14, I15, I16, a34, I17... a38, a1};
B '={ b1..., b3(I2), I3, b4, b5(I4), I5, I6, I7, b6, b7, I8, b8(I9), I10, b9, I11, b10...,
b14, I12, b15(I13), b16..., b23, I14, b24, b30, b31(I15), b32, I16, b33, I17, b34, b35, I1, b36..., b38,
b1};
5.1.5 travel through intersection point, obtain common factor polygon chained list W:
W={ I1, a7, I2, I3, a8..., a11, I5, I6, I7, a20..., I8, I9, a29..., a31, I11, I12, b16, b23,
I13, I14, b32, I15, a34, I16, b34, b35, I1}∪{I10, b9, I11, a27}
5.2 ask for common factor area of a polygon
SIntersection=17705.2
Aircraft area of a polygon to be measured
S=25052;
5.3 seek template similarity coefficient
ξ=SIntersection/ S=0.707;
5.4 compare similarity coefficient identification target, select the corresponding aircraft of b4 template to be that aircraft to be measured corresponds to type in figure legends
Number.
The above is only the preferred embodiment of the present invention, protection scope of the present invention is not limited to above-described embodiment, all
The technical scheme belonging under thinking of the present invention belongs to protection scope of the present invention.It should be pointed out that it is general for the art
For logical technical staff, some improvements and modifications without departing from the principles of the present invention, should be regarded as the protection model of the present invention
Enclose.
Claims (3)
1. a kind of Aircraft Target Recognition based on disparity is it is characterised in that comprise the steps:
(1) utilize certain direction view outline polygon of aircraft, the profile obtaining other direction views by affine transformation is polygon
Shape;Travel through each model aircraft, set up aircraft multi-pose model library;
(2) place is normalized to the aircraft profile in the aircraft profile in aircraft brake disc to be identified and aircraft multi-pose model library
Reason is so that the aircraft profile in aircraft brake disc to be identified has identical face with the aircraft profile in aircraft multi-pose model library
Long-pending;
(3) aircraft outline polygon overlap processing, comprises the following steps:
(3.1) the aircraft profile in aircraft brake disc to be identified is designated as polygon A, the aircraft profile in aircraft multi-pose model library
It is designated as polygon B;Calculate center of gravity O of polygon A respectivelyABarycentric coodinates O with polygon BB;
(3.2) polygon B is rotated and translated certain a line each bar side weight with polygon A successively so that polygon B
Folded and midpoint overlaps, each bar side of traversal polygon B, calculate i-th of polygon A overlapping in j-th with polygon B when
OA、OBApart from dI, j;
(3.3) will be apart from dI, jWith default threshold value d*Compare, record meets dI, j<d*When corresponding polygon B attitude;
(4) identify target aircraft, comprise the following steps:
(4.1) ask for the common factor of polygon A and polygon B, obtain common factor polygon;
(4.2) ask for the polygonal area S that occurs simultaneouslyIntersectionAnd the area S of polygon A;
(4.3) seek template similarity coefficient ξ=SIntersection/S;
(4.4) choose the aircraft profile in the maximum aircraft multi-pose model library of template similarity coefficient, corresponding type of airplane is
Correspond to target for target to be identified.
2. the Aircraft Target Recognition based on disparity according to claim 1 it is characterised in that:Described step
(2) before, Image semantic classification is carried out to aircraft brake disc to be identified, including image gray processing, image smoothing, image filtering, figure
Image intensifying, Image Edge-Detection and polygonal segments.
3. the Aircraft Target Recognition based on disparity according to claim 1 is it is characterised in that described step
(2) comprise the following steps:
(2.1) the aircraft profile in aircraft brake disc to be identified is designated as polygon A, the aircraft profile in aircraft multi-pose model library
It is designated as polygon B;Calculate the area S of polygon A and polygon B respectivelyA、SB;Likelihood ratio k=B/A of polygon A, B;
(2.2) take up an official post from polygon A place plane and take each summit of 1 point of O, junction point O and polygon A, calculate every line long
Degree li, i=1,2,3,4 ..., n, n represent polygon vertex number, take point B on every line or its extended line respectively1、B2、
B3、…、BnSo that each point and point O are apart from Li=k*li;
(2.3) junction point B1、B2、B3、…、Bn, obtain the polygon after normalization.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610789468.2A CN106407902A (en) | 2016-08-31 | 2016-08-31 | Geometric difference-based airplane object identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610789468.2A CN106407902A (en) | 2016-08-31 | 2016-08-31 | Geometric difference-based airplane object identification method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106407902A true CN106407902A (en) | 2017-02-15 |
Family
ID=58001248
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610789468.2A Pending CN106407902A (en) | 2016-08-31 | 2016-08-31 | Geometric difference-based airplane object identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106407902A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108052942A (en) * | 2017-12-28 | 2018-05-18 | 南京理工大学 | A kind of visual pattern recognition methods of aircraft flight attitude |
RU2708346C1 (en) * | 2017-03-22 | 2019-12-05 | Александр Иванович Стучилин | Method for identification of aircraft type by means of passive optical location and device for implementation thereof |
WO2021164328A1 (en) * | 2020-02-17 | 2021-08-26 | 腾讯科技(深圳)有限公司 | Image generation method, device, and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101877062A (en) * | 2009-12-21 | 2010-11-03 | 汉王科技股份有限公司 | Method for profile analysis in image layout area |
CN102930300A (en) * | 2012-11-21 | 2013-02-13 | 北京航空航天大学 | Method and system for identifying airplane target |
CN103914687A (en) * | 2014-03-14 | 2014-07-09 | 常州大学 | Rectangular-target identification method based on multiple channels and multiple threshold values |
CN104105439A (en) * | 2012-02-06 | 2014-10-15 | 皇家飞利浦有限公司 | Invisible bifurcation detection within vessel tree images |
CN105469094A (en) * | 2015-11-28 | 2016-04-06 | 重庆交通大学 | Edge vector line extraction algorithm of binary image of road surface |
CN107516095A (en) * | 2016-06-16 | 2017-12-26 | 阿里巴巴集团控股有限公司 | A kind of image-recognizing method and device |
-
2016
- 2016-08-31 CN CN201610789468.2A patent/CN106407902A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101877062A (en) * | 2009-12-21 | 2010-11-03 | 汉王科技股份有限公司 | Method for profile analysis in image layout area |
CN104105439A (en) * | 2012-02-06 | 2014-10-15 | 皇家飞利浦有限公司 | Invisible bifurcation detection within vessel tree images |
CN102930300A (en) * | 2012-11-21 | 2013-02-13 | 北京航空航天大学 | Method and system for identifying airplane target |
CN103914687A (en) * | 2014-03-14 | 2014-07-09 | 常州大学 | Rectangular-target identification method based on multiple channels and multiple threshold values |
CN105469094A (en) * | 2015-11-28 | 2016-04-06 | 重庆交通大学 | Edge vector line extraction algorithm of binary image of road surface |
CN107516095A (en) * | 2016-06-16 | 2017-12-26 | 阿里巴巴集团控股有限公司 | A kind of image-recognizing method and device |
Non-Patent Citations (1)
Title |
---|
王彦芳等: ""基于几何差异的目标识别算法"", 《计算机测量与控制》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
RU2708346C1 (en) * | 2017-03-22 | 2019-12-05 | Александр Иванович Стучилин | Method for identification of aircraft type by means of passive optical location and device for implementation thereof |
CN108052942A (en) * | 2017-12-28 | 2018-05-18 | 南京理工大学 | A kind of visual pattern recognition methods of aircraft flight attitude |
CN108052942B (en) * | 2017-12-28 | 2021-07-06 | 南京理工大学 | Visual image recognition method for aircraft flight attitude |
WO2021164328A1 (en) * | 2020-02-17 | 2021-08-26 | 腾讯科技(深圳)有限公司 | Image generation method, device, and storage medium |
US11847812B2 (en) | 2020-02-17 | 2023-12-19 | Tencent Technology (Shenzhen) Company Limited | Image generation method and apparatus, device, and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Papazov et al. | Real-time 3D head pose and facial landmark estimation from depth images using triangular surface patch features | |
CN107194341B (en) | Face recognition method and system based on fusion of Maxout multi-convolution neural network | |
WO2020063527A1 (en) | Human hairstyle generation method based on multi-feature retrieval and deformation | |
Glasner et al. | Aware object detection and pose estimation | |
CN108108764B (en) | Visual SLAM loop detection method based on random forest | |
CN107481315A (en) | A kind of monocular vision three-dimensional environment method for reconstructing based on Harris SIFT BRIEF algorithms | |
CN107122712B (en) | Palm print image identification method based on CNN and bidirectional VLAD | |
Yue et al. | Robust loop closure detection based on bag of superpoints and graph verification | |
CN104346608A (en) | Sparse depth map densing method and device | |
CN104850838A (en) | Three-dimensional face recognition method based on expression invariant regions | |
Wang et al. | Bottle detection in the wild using low-altitude unmanned aerial vehicles | |
CN102364497A (en) | Image semantic extraction method applied in electronic guidance system | |
CN104616247B (en) | A kind of method for map splicing of being taken photo by plane based on super-pixel SIFT | |
Takahashi et al. | Expandable YOLO: 3D object detection from RGB-D images | |
CN111340873A (en) | Method for measuring and calculating object minimum outer envelope size of multi-view image | |
CN108388902B (en) | Composite 3D descriptor construction method combining global framework point and local SHOT characteristics | |
CN105654122B (en) | Based on the matched spatial pyramid object identification method of kernel function | |
CN106407902A (en) | Geometric difference-based airplane object identification method | |
CN112686191B (en) | Living body anti-counterfeiting method, system, terminal and medium based on three-dimensional information of human face | |
Yin et al. | Cae-lo: Lidar odometry leveraging fully unsupervised convolutional auto-encoder for interest point detection and feature description | |
CN110348310A (en) | A kind of Hough ballot 3D colour point clouds recognition methods | |
CN104732247B (en) | A kind of human face characteristic positioning method | |
Lee | Symmetry-driven shape description for image retrieval | |
CN110969101A (en) | Face detection and tracking method based on HOG and feature descriptor | |
CN104376312A (en) | Face recognition method based on word bag compressed sensing feature extraction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170215 |
|
WD01 | Invention patent application deemed withdrawn after publication |