CN105303566A - Target contour clipping-based SAR image target azimuth estimation method - Google Patents
Target contour clipping-based SAR image target azimuth estimation method Download PDFInfo
- Publication number
- CN105303566A CN105303566A CN201510666064.XA CN201510666064A CN105303566A CN 105303566 A CN105303566 A CN 105303566A CN 201510666064 A CN201510666064 A CN 201510666064A CN 105303566 A CN105303566 A CN 105303566A
- Authority
- CN
- China
- Prior art keywords
- target
- straight line
- azimuth
- matrix
- 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
Classifications
-
- 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
- G06T2207/10044—Radar image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20116—Active contour; Active surface; Snakes
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a target contour clipping-based SAR image target azimuth estimation method. The method includes the following steps that: S1, an SAR target is extracted from an input image; S2, the contour of the object is extracted, and the information of the contour of the object is expressed as a binary contour matrix; S3, a first nonzero element and the next element in each column of elements in the contour matrix are set to zero, so that clipping on the contour can be realized; and S4, a straight line having the most contact points with the clipped contour is searched, and an included angle between the straight line and a horizontal direction is adopted as a currently estimated target azimuth. According to the target contour clipping-based SAR image target azimuth estimation method of the invention, requirements for the shape of the target are not so many, so that dependence on a target extraction process is decreased; a contour clipping method is adopted, estimation accuracy under the condition that the difference of the length of a long leading edge and a short leading edge is small can be improved; and the estimated azimuth is corrected, so that defects of a traditional method in distinguishing a vertical azimuth and a horizontal azimuth can be eliminated, and estimation accuracy can be improved. The target contour clipping-based SAR image target azimuth estimation method of the invention has the advantages of low computing complexity and fast estimation.
Description
Technical field
The invention belongs in technical field of image processing, relate to image steganalysis method, be specifically related to the method for estimation of synthetic-aperture radar (SyntheticApertureRadar, the SAR) image goal position angle of a kind of based target profile cutting.
Background technology
Synthetic-aperture radar has the advantages such as round-the-clock, all-weather and strong penetration capacity, has become a kind of important military investigation.In recent years, the research utilizing high-resolution SAR image to carry out automatic target detection (AutomaticTargetRecognition, ATR) continues to bring out.
SAR target image is very responsive to the orientation of radar imagery, and the image difference that same target obtains in different azimuth is very large.The SAR template image of a large amount of different azimuth is stored, by target to be identified and template are carried out the identification of mating to come realize target in traditional SARATR system.Therefore, pre-estimate out the position angle of target, effectively can reduce the quantity of searching image, improve recognition efficiency and the accuracy rate of ATR system.
The estimation procedure of SAR azimuth of target generally includes Objective extraction and angle estimation two links.Objective extraction is extracted as far as possible exactly from image SAR target; Angle estimation is then to extracting target analysis, estimates the position angle of target.At present, main SAR azimuth of target method of estimation has: method of principal axis, boundary rectangle method and primary edge method.
Method of principal axis estimates azimuthal by the main shaft of target scattering center.Because the mathematical model that relates to is simple, calculation of complex is little, and for target occlusion to a certain degree, hidden and hinged there is robustness.But these class methods basic assumption is SAR target is about main axisymmetric.And in practical situations both, due to the impact by imaging scene or object construction, symmetry hypothesis is not necessarily set up, position angle is caused to be estimated inaccurate.Boundary rectangle method utilizes minimum enclosed rectangle fit object, always azimuth of target is determined according to walking of boundary rectangle, the shape of the method to target has higher requirement, namely needs the target extracted to have more regular shape, and irregular shape may produce larger angle estimation deviation; In addition, the method relates to series of rectangular and rotates, and calculated amount is larger.Primary edge rule is that the primary edge by detecting target estimates position angle.Comparatively first two method is high for the estimated accuracy of the method, and the acquisition of primary edge does not need complicated matrix rotation, speed, its major defect is when the short primary edge of target is vertical with radar beam, the estimation of vertical and level orientation produces and obscures, this problem fails to be well solved always, have impact on widely using of algorithm; And the method, when long and short primary edge length difference is less, easily produces larger estimated bias.
Summary of the invention
The object of the invention is to overcome the deficiencies in the prior art, provide a kind of fast and the SAR image azimuth of target method of estimation of the higher based target profile cutting of estimated accuracy.
The object of the invention is to be achieved through the following technical solutions: the SAR image azimuth of target method of estimation of a kind of based target profile cutting, comprises the following steps:
S1, extraction target, with image partition method by the SAR Objective extraction in input picture out, be expressed as a binary object matrix, the line number of this matrix and columns equal width and the length of image respectively, a pixel of each some correspondence image of matrix, its value is this pixel of 1 expression is impact point, be this pixel of 0 expression is non-targeted point;
S2, extraction profile, to each point in objective matrix, add up the number of its peripheral region internal object point, if impact point number is less than default adjacent region threshold, then judges that this point is as point, otherwise be non-point; Objective contour information is expressed as a two-value profile matrix, the line number of this matrix and columns equal width and the length of image respectively, a pixel of each some correspondence image of matrix, its value is this pixel of 1 expression is point, be this pixel of 0 expression is non-point;
S3, cutting profile, successively each column element of scanning profile matrix, by the nonzero element of first in every column element and next element zero setting thereof;
S4, estimation position angle, comprise following sub-step:
S41, make in target image plane the straight line { L that one group of width is two pixels
θ, d, wherein θ is the angle of straight line and vertical direction, 0 ° of < θ≤180 °, and the sampling interval of θ is R
θ, 0 < R
θ≤ 1; D is the distance of top left corner apex to straight line of image,
wherein p, q are respectively line number and the columns of profile matrix, and the sampling interval of d is R
d, 0 < R
d≤ 1;
S42, add up every bar straight line L
θ, dwith the contact point number k of cutting rear profile
θ, d, ask the maximal value k of contact point number
max;
If S43 is k
maxcorresponding straight line only has one, then remember that this straight line is L
max; Otherwise, select k
maxstraight line minimum with horizontal direction angle in corresponding each straight line, and remember that this straight line is L
max;
S44, determine L
maxangle be the azimuth of target θ of current estimation
est.
Further, described position angle method of estimation also comprises step S5, a corrected azimuth, specifically comprises following sub-step:
The maximal value of the contact point number of S51, the straight line asking angle to be 180 ° and cutting rear profile, is designated as k
v, and calculate k
vwith maximal value k
maxrelative variation, be designated as Δ
k:
If S52 relative variation Δ
kbe less than the relative change threshold of contact point of setting
and straight line L
maxwith the angle Δ of horizontal direction
θbe less than the angle threshold of setting
then put θ
est=180.
Further, in described step S2, when adding up the number of each peripheral region internal object point in objective matrix, the area size of statistics is 5 × 5 pixels.
Further, described step S3 concrete methods of realizing is: establish profile matrix A=[A
1a
2... A
q], size is p × q, wherein A
ifor the column vector of p dimension, i=1,2 ..., q; Scan A successively from top to bottom
ieach element a
j,i, j=1,2 ..., p, if a
m,ifor first nonzero element scanned, then put a
m,i=0, if m < p now, then put a again
m,inext element a
(m+1), i=0.
The invention has the beneficial effects as follows: the inventive method does not do too much requirement to the shape of target, thus decrease the dependence to Objective extraction process; Adopt the method for profile cutting, improve the estimation accuracy in the less situation of long and short primary edge length difference, by revising the position angle estimated, compensate for classic method distinguishing the deficiency in vertical and level orientation, improve estimated accuracy; In addition, the inventive method computation complexity is low, estimates fast.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of position angle of the present invention method of estimation embodiment one;
Fig. 2 is the process flow diagram of position angle of the present invention method of estimation embodiment two;
Fig. 3 is for adopting method of the present invention to HB19462.001 result figure;
Fig. 4 is the estimated result schematic diagram of each method to the azimuth of target of image HB19462.001;
Fig. 5 is the estimated result schematic diagram of each method to the azimuth of target of image HB03808.004.
Embodiment
Technical scheme of the present invention is further illustrated below in conjunction with accompanying drawing.
The test pattern of the embodiment of the present invention all comes from MSTA (MovingandStationaryTargetAcquisitionandRecognition) standard data set of the advanced research project office (DARPA) of U.S. national defense.Be described for image HB19462.001 in MSTAR (as Suo Shi Fig. 3 (a)).The corresponding model of this image is the BRDM terrain object of E71, and the angle of pitch of imaging is 17 °.
Embodiment one: as shown in Figure 1, the SAR image azimuth of target method of estimation of a kind of based target profile cutting, comprises the following steps:
S1, extraction target, the size of the present embodiment input picture is 128 × 128, with image partition method by the SAR Objective extraction in input picture, is expressed as the binary object matrix O that a size is 128 × 128; The line number of objective matrix and columns equal width and the length of image respectively, a pixel of each some correspondence image of objective matrix, its value point (impact point) that to be this pixel of 1 expression be in target is this pixel of 0 expression is non-targeted point;
S2, extraction profile, to each point in O in objective matrix, add up the number n of (peripheral region selected by the present embodiment is 5 × 5 pixels) impact point in its peripheral region, if impact point number n is less than default adjacent region threshold T
n=22 (22≤T
n≤ 24), then judge that this is as the point (point) on profile, otherwise be non-point; Objective contour information is expressed as the two-value profile matrix A that a size is 128 × 128, the line number of this matrix and columns equal width and the length of image respectively, a pixel of each some correspondence image of A, its value is this pixel of 1 expression is point, be this pixel of 0 expression is non-point; The objective contour that the present embodiment extracts is as shown in Fig. 3 (b);
S3, cutting profile, successively each column element of scanning profile matrix, by the nonzero element of first in every column element and next element zero setting thereof, realize the cutting to profile with this; Specific implementation method is: establish profile matrix A=[A
1a
2... A
q], size is 128 × 128, wherein A
ibe the column vector of 128 dimensions, i=1,2 ..., 128; Scan A successively from top to bottom
ieach element a
j,i, j=1,2 ..., 128, if a
m,ifor first nonzero element scanned, then put a
m,i=0, if m < 128 now, then put a again
m,inext element a
(m+1), i=0; As the 57th column data A
57containing nonzero element, first nonzero element is positioned at the 75th row, then put a
75,57=0 and subsequent a
76,57=0; Contour images after the present embodiment cutting is as shown in Fig. 3 (c);
S4, estimation position angle, comprise following sub-step:
S41, make in target image plane the straight line { L that one group of width is two pixels
θ, d, wherein θ is the angle of straight line and vertical direction, 0 ° of < θ≤180 °, and the sampling interval of θ is R
θ=0.5 (0 < R
θ≤ 1); D is the distance of top left corner apex to straight line of image,
the sampling interval of d is R
d=1 (0 < R
d≤ 1);
S42, add up every bar straight line L
θ, dwith the contact point number k of cutting rear profile
θ, d, ask the maximal value k of contact point number
max=20;
Contact point number maximal value k in S43, the present embodiment
maxthe straight line of=20 has many, therefore contact point number is 20 and the straight line minimum with the angle (horizontal angle) of horizontal direction is designated as L
max, now L
maxcorresponding horizontal angle Δ
θbe 5.500;
S44, determine L
maxangle be the azimuth of target θ of current estimation
est, because of L
maxangle theta=95.500, then the azimuth of target θ of current estimation is set
estit is 95.500 degree.
Embodiment two: embodiment one has completed the estimation of image goal position angle, in order to obtain more accurate azimuth of target further, as shown in Figure 2, the present invention is after completing steps S1 ~ S4, also comprise step S5, a corrected azimuth, specifically comprise following sub-step:
The maximal value of the contact point number of S51, the straight line asking angle to be 180 ° and cutting rear profile, is designated as k
v, try to achieve k
v=18; Calculate k
vwith maximal value k
maxrelative variation (relative variation), be designated as Δ
k:
S52, the relative change threshold of contact point is set
angle threshold
because meeting simultaneously
and
the azimuth of target θ then will estimated
estbe modified to 180 degree; So far, the azimuth of target θ estimated
est=180.The position angle that the present embodiment estimates is as shown in Fig. 3 (d), and wherein azimuthal direction straight line that target is surveyed outward represents.
Below effect of the present invention and boundary rectangle method and primary edge method are compared.Test machine is Intel (R) i5-5300U processor, and dominant frequency is 2.3GHz.Test point free hand drawing test and integration test two step are carried out.In all tests, the first step all adopts conventional threshold segmentation method to extract target.
(1) free hand drawing test
Free hand drawing test is described for image HB19462.001 (as Suo Shi Fig. 4 (a)) and image HB03808.004 (as Suo Shi Fig. 5 (a)).The corresponding model of image HB03808.004 is the BTR70 terrain object of c71, and the angle of pitch of imaging is 17 °.
Table 1 is the inventive method and boundary rectangle method, the position angle estimated result of primary edge method compares.Can see, in the accuracy estimated, the inventive method is obviously better than boundary rectangle method and primary edge method.Especially, to HB19462.001 image, boundary rectangle method and primary edge method create in the estimation of vertical and level orientation to be obscured, thus causes very large evaluated error, and the present invention overcomes this problem, obtains and estimates comparatively accurately.
The each method of table 1 compares single image azimuth of target estimated result
Fig. 4 is the estimated result schematic diagram (azimuth direction that estimate straight line that target outward survey represent) of each method to the azimuth of target of HB19462.001 image, wherein Fig. 4 (a) is original HB19462.001 image, Fig. 4 (b) is for boundary rectangle method is to the estimated result of HB19462.001 image, Fig. 4 (c) takes the estimated result of boundary method to HB19462.001 image as the leading factor, and Fig. 4 (d) is for the inventive method is to the estimated result of HB19462.001 image.
Fig. 5 is the estimated result schematic diagram (azimuth direction that estimate straight line that target outward survey represent) of each method to the azimuth of target of HB03808.004 image, wherein Fig. 5 (a) is original HB03808.004 image, Fig. 5 (b) is for boundary rectangle method is to the estimated result of HB03808.004 image, Fig. 5 (c) takes the estimated result of boundary method to HB03808.004 image as the leading factor, and Fig. 5 (d) is for the inventive method is to the estimated result of HB03808.004 image.
(2) integration test
It is that BRDM, BTR70 and BMP (SN9563) the three types target of 17 ° is tested that integration test have selected the angle of pitch in MSTAR database.Wherein BRDM type has 298 width target images, and BTR70 type and BMP (SN9563) type respectively have 233 width.Carry out angle estimation to the method that these test pattern the inventive method and boundary rectangle method, primary edge method and 2011 " computer utility " upper " the SAR azimuth of target method of estimation based on primary edge Radon converts " (document 1) issued propose, its result is as shown in table 2.List the target numbers of absolute error within the scope of 5 ° and 10 ° estimated at position angle in table and account for the ratio of general objective number, the average of absolute error and standard deviation, working time etc.From the accuracy estimated, the inventive method evaluated error is less, and much smaller than boundary rectangle method and primary edge method method.Be averaged to all images, the inventive method estimates that absolute error is less than 5 ° and is respectively 92.7% and 98.9% with the ratio being less than 10 °, is all obviously better than additive method.From working time, the estimating speed of the inventive method is the fastest in all methods, and obviously faster than boundary rectangle method.It should be noted that, in table, the data of " improvement primary edge " method are come in document 1.In theory, Ying Yuyuan consuming time " primary edge " method of the method quite consuming time, but in table, data are much larger than this experimental result, and this is mainly due in first step Objective extraction, and document 1 have employed more complicated method; In addition, document 1 is different from the machine that this experiment test uses, and result also in the difference of data.
The estimated result of each method of table 2 to sample image azimuth of target compares
Those of ordinary skill in the art will appreciate that, embodiment described here is to help reader understanding's principle of the present invention, should be understood to that protection scope of the present invention is not limited to so special statement and embodiment.Those of ordinary skill in the art can make various other various concrete distortion and combination of not departing from essence of the present invention according to these technology enlightenment disclosed by the invention, and these distortion and combination are still in protection scope of the present invention.
Claims (4)
1. a SAR image azimuth of target method of estimation for based target profile cutting, is characterized in that, comprise the following steps:
S1, extraction target, with image partition method by the SAR Objective extraction in input picture out, be expressed as a binary object matrix, the line number of this matrix and columns equal width and the length of image respectively, a pixel of each some correspondence image of matrix, its value is this pixel of 1 expression is impact point, be this pixel of 0 expression is non-targeted point;
S2, extraction profile, to each point in objective matrix, add up the number of its peripheral region internal object point, if impact point number is less than default adjacent region threshold, then judges that this point is as point, otherwise be non-point; Objective contour information is expressed as a two-value profile matrix, the line number of this matrix and columns equal width and the length of image respectively, a pixel of each some correspondence image of matrix, its value is this pixel of 1 expression is point, be this pixel of 0 expression is non-point;
S3, cutting profile, successively each column element of scanning profile matrix, by the nonzero element of first in every column element and next element zero setting thereof;
S4, estimation position angle, comprise following sub-step:
S41, make in target image plane the straight line { L that one group of width is two pixels
θ, d, wherein θ is the angle of straight line and vertical direction, 0 ° of < θ≤180 °, and the sampling interval of θ is R
θ, 0 < R
θ≤ 1; D is the distance of top left corner apex to straight line of image,
wherein p, q are respectively line number and the columns of profile matrix, and the sampling interval of d is R
d, 0 < R
d≤ 1;
S42, add up every bar straight line L
θ, dwith the contact point number k of cutting rear profile
θ, d, ask the maximal value k of contact point number
max;
If S43 is k
maxcorresponding straight line only has one, then remember that this straight line is L
max; Otherwise, select k
maxstraight line minimum with horizontal direction angle in corresponding each straight line, and remember that this straight line is L
max;
S44, determine L
maxangle be the azimuth of target θ of current estimation
est.
2. the SAR image azimuth of target method of estimation of based target profile according to claim 1 cutting, is characterized in that, described position angle method of estimation also comprises step S5, a corrected azimuth, specifically comprises following sub-step:
The maximal value of the contact point number of S51, the straight line asking angle to be 180 ° and cutting rear profile, is designated as k
v, and calculate k
vwith maximal value k
maxrelative variation, be designated as Δ
k:
If S52 relative variation Δ
kbe less than the relative change threshold of contact point of setting
and straight line L
maxwith the angle Δ of horizontal direction
θbe less than the angle threshold of setting
then put θ
est=180.
3. the SAR image azimuth of target method of estimation of based target profile according to claim 1 cutting, it is characterized in that, in described step S2, when adding up the number of each peripheral region internal object point in objective matrix, the area size of statistics is 5 × 5 pixels.
4. the SAR image azimuth of target method of estimation of the based target profile cutting according to claims 1 to 3 any one, is characterized in that, described step S3 concrete methods of realizing is: establish profile matrix A=[A
1a
2... A
q], size is p × q, wherein A
ifor the column vector of p dimension, i=1,2 ..., q; Scan A successively from top to bottom
ieach element a
j,i, j=1,2 ..., p, if a
m,ifor first nonzero element scanned, then put a
m,i=0, if m < p now, then put a again
m,inext element a
(m+1), i=0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510666064.XA CN105303566B (en) | 2015-10-15 | 2015-10-15 | A kind of SAR image azimuth of target method of estimation cut based on objective contour |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510666064.XA CN105303566B (en) | 2015-10-15 | 2015-10-15 | A kind of SAR image azimuth of target method of estimation cut based on objective contour |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105303566A true CN105303566A (en) | 2016-02-03 |
| CN105303566B CN105303566B (en) | 2018-02-09 |
Family
ID=55200788
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510666064.XA Expired - Fee Related CN105303566B (en) | 2015-10-15 | 2015-10-15 | A kind of SAR image azimuth of target method of estimation cut based on objective contour |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105303566B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106874881A (en) * | 2017-02-23 | 2017-06-20 | 电子科技大学 | A kind of anti-joint sparse of part of multi-template space time correlation represents method for tracking target |
| CN107610131A (en) * | 2017-08-25 | 2018-01-19 | 百度在线网络技术（北京）有限公司 | A kind of image cropping method and image cropping device |
| CN108629786A (en) * | 2017-03-23 | 2018-10-09 | 展讯通信（上海）有限公司 | Method for detecting image edge and device |
| CN112070151A (en) * | 2020-09-07 | 2020-12-11 | 北京环境特性研究所 | Target classification and identification method of MSTAR data image |
| CN113344937A (en) * | 2021-04-25 | 2021-09-03 | 中国科学院空天信息创新研究院 | Method for automatically identifying and removing black edge of remote sensing image |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101710176A (en) * | 2009-12-23 | 2010-05-19 | 北京航空航天大学 | SAR image moving object attitude angle extraction method based on echoed data |
| US20110235897A1 (en) * | 2010-03-24 | 2011-09-29 | Nat'l Institute Of Advanced Industrial Science And Technology | Device and process for three-dimensional localization and pose estimation using stereo image, and computer-readable storage medium storing the program thereof |
| CN103065162A (en) * | 2013-01-31 | 2013-04-24 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) target azimuth angle estimation method based on sparse description |
| CN103177443A (en) * | 2013-03-07 | 2013-06-26 | 中国电子科技集团公司第十四研究所 | SAR (synthetic aperture radar) target attitude angle estimation method based on randomized hough transformations |
-
2015
- 2015-10-15 CN CN201510666064.XA patent/CN105303566B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101710176A (en) * | 2009-12-23 | 2010-05-19 | 北京航空航天大学 | SAR image moving object attitude angle extraction method based on echoed data |
| US20110235897A1 (en) * | 2010-03-24 | 2011-09-29 | Nat'l Institute Of Advanced Industrial Science And Technology | Device and process for three-dimensional localization and pose estimation using stereo image, and computer-readable storage medium storing the program thereof |
| CN103065162A (en) * | 2013-01-31 | 2013-04-24 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) target azimuth angle estimation method based on sparse description |
| CN103177443A (en) * | 2013-03-07 | 2013-06-26 | 中国电子科技集团公司第十四研究所 | SAR (synthetic aperture radar) target attitude angle estimation method based on randomized hough transformations |
Non-Patent Citations (2)
| Title |
|---|
| LIJUAN QIN ET AL.: "The Judgment Method for the Unique Solution of Real-time Pose Estimation from Particular Line Correspondences", 《PROCEEDINGS OF THE 2006 IEEE INTERNATIONAL CONFERENCE ON MECHATRONICS AND AUTOMATION》 * |
| 高贵 等: "SAR图像目标方位角估计方法综述", 《信号处理》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106874881A (en) * | 2017-02-23 | 2017-06-20 | 电子科技大学 | A kind of anti-joint sparse of part of multi-template space time correlation represents method for tracking target |
| CN106874881B (en) * | 2017-02-23 | 2019-09-24 | 电子科技大学 | A kind of anti-joint sparse expression method for tracking target in the part of multi-template space time correlation |
| CN108629786A (en) * | 2017-03-23 | 2018-10-09 | 展讯通信（上海）有限公司 | Method for detecting image edge and device |
| CN108629786B (en) * | 2017-03-23 | 2020-07-21 | 展讯通信（上海）有限公司 | Image edge detection method and device |
| CN107610131A (en) * | 2017-08-25 | 2018-01-19 | 百度在线网络技术（北京）有限公司 | A kind of image cropping method and image cropping device |
| CN107610131B (en) * | 2017-08-25 | 2020-05-12 | 百度在线网络技术（北京）有限公司 | Image clipping method and image clipping device |
| CN112070151A (en) * | 2020-09-07 | 2020-12-11 | 北京环境特性研究所 | Target classification and identification method of MSTAR data image |
| CN112070151B (en) * | 2020-09-07 | 2023-12-29 | 北京环境特性研究所 | Target classification and identification method for MSTAR data image |
| CN113344937A (en) * | 2021-04-25 | 2021-09-03 | 中国科学院空天信息创新研究院 | Method for automatically identifying and removing black edge of remote sensing image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105303566B (en) | 2018-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105303566A (en) | Target contour clipping-based SAR image target azimuth estimation method | |
| CN103886325B (en) | Cyclic matrix video tracking method with partition | |
| CN110097536A (en) | Hexagon bolt looseness detection method based on deep learning and Hough transformation | |
| CN105046271A (en) | MELF (Metal Electrode Leadless Face) component positioning and detecting method based on match template | |
| CN108596961A (en) | Point cloud registration method based on Three dimensional convolution neural network | |
| CN114677554A (en) | Statistical filtering infrared small target detection tracking method based on YOLOv5 and Deepsort | |
| CN101770583B (en) | Template matching method based on global features of scene | |
| CN110647836B (en) | Robust single-target tracking method based on deep learning | |
| CN102163333B (en) | Change detection method for synthetic aperture radar (SAR) images of spectral clustering | |
| Liang et al. | Robust sea-sky-line detection for complex sea background | |
| CN103778436A (en) | Pedestrian gesture inspecting method based on image processing | |
| CN103871039A (en) | Generation method for difference chart in SAR (Synthetic Aperture Radar) image change detection | |
| CN108319961B (en) | Image ROI rapid detection method based on local feature points | |
| CN116086484A (en) | Laser radar mileage calculation method based on ground plane constraint and loop detection | |
| CN116758049A (en) | Urban flood three-dimensional monitoring method based on active and passive satellite remote sensing | |
| CN107230210A (en) | A kind of fast partition method of remote sensing images harbour waterborne target | |
| CN104881670A (en) | Rapid target extraction method used for SAR azimuth estimation | |
| CN103093241B (en) | Based on the remote sensing image nonuniformity cloud layer method of discrimination of homogeneity process | |
| CN105205826A (en) | SAR image target azimuth angle estimation method based on direction straight line screening | |
| Liu et al. | A Multi-scale Feature Pyramid SAR Ship Detection Network with Robust Background Interference | |
| Yang et al. | Fast and accurate vanishing point detection in complex scenes | |
| CN114820712B (en) | Unmanned aerial vehicle tracking method based on self-adaptive target frame optimization | |
| CN104732190A (en) | Synthetic aperture sonar target detection method based on orthogonal texture correlation analysis | |
| Boufama et al. | Towards a fast and reliable dense matching algorithm | |
| Li et al. | Multi-level Pyramid Feature Extraction and Task Decoupling Network for SAR Ship Detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180209 Termination date: 20201015 |