CN112001239B - SAR image target detection optimization method based on multiple SVD saliency fusion - Google Patents
SAR image target detection optimization method based on multiple SVD saliency fusion Download PDFInfo
- Publication number
- CN112001239B CN112001239B CN202010679460.7A CN202010679460A CN112001239B CN 112001239 B CN112001239 B CN 112001239B CN 202010679460 A CN202010679460 A CN 202010679460A CN 112001239 B CN112001239 B CN 112001239B
- Authority
- CN
- China
- Prior art keywords
- image
- characteristic
- sar image
- approximate
- sar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000005457 optimization Methods 0.000 title claims abstract description 9
- 230000004927 fusion Effects 0.000 title claims abstract description 7
- 239000010410 layer Substances 0.000 claims abstract description 15
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 13
- 239000011229 interlayer Substances 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 42
- 238000004364 calculation method Methods 0.000 claims description 24
- 238000010586 diagram Methods 0.000 claims description 15
- 230000003014 reinforcing effect Effects 0.000 claims description 15
- 230000002093 peripheral effect Effects 0.000 claims description 6
- 238000012546 transfer Methods 0.000 abstract description 3
- 241000657949 Elderberry carlavirus D Species 0.000 description 16
- 238000005516 engineering process Methods 0.000 description 5
- 238000003384 imaging method Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention discloses a SAR image target detection optimization method based on multiple SVD significance fusion, which comprises the following implementation steps: performing multiple SVD decomposition on an original SAR image, extracting intensity features, outlier features and consistency features from the SAR image and an approximate image thereof, respectively performing interlayer addition operation on the SAR image under different characteristics, and performing intra-layer image mean value extraction and other operations to obtain a total saliency map; and taking the pixel point with the highest salient value in the total salient map as a center, and taking the size of the required detection target in the image as a radius to obtain the transfer track and distribution of the salient region. The SAR image target detection method can improve the detection rate of the SAR image target detection algorithm under the condition of high resolution, and has certain robustness to noise.
Description
Technical Field
The invention belongs to radar remote sensing or image processing technology, namely an image processing technology is used for analyzing radar observation information, and particularly relates to application of a method for combining multiple SVDs, SAR image features and saliency in SAR image target detection.
Background
With the continuous development of synthetic aperture radar (Synthetic Aperture Radar, SAR) imaging technologies, the application range of SAR systems has become wide, including detection and identification of military targets and facilities, classification and identification of crops, detection and positioning of traffic facilities such as roads and bridges, and urban building distribution analysis. However, the rapid development of SAR imaging technology and the widespread use of SAR systems also present significant challenges for the interpretation work of SAR images. How to quickly and accurately find the needed information from massive SAR image data becomes a difficulty of the current SAR image interpretation research work.
SAR image target detection is a key problem in SAR image interpretation work. In particular, in the field of military, detection and positioning of military targets such as tanks, airplanes, ships and the like in SAR images have been the focus of research. At present, the most widely applied SAR image target detection method is a constant false alarm rate target detection method, and the target detection method is mainly developed along the improvement and innovation of the background clutter estimation method and the detector design. However, the constant false alarm rate target detection method often processes the pixel points in the SAR image, that is, determines whether the pixel points in the image are target pixel points or background pixel points according to a certain criterion. With the development of the SAR imaging technology, the resolution of the SAR image is higher, the details of a target and a background in the SAR image are clearer, and meanwhile, the local areas in the target and the background are fluctuant, so that when a constant false alarm rate detection method is used for judging single pixel points, the accuracy is reduced, namely, the pixel points in a local high-intensity area in the background are easily judged as the target, the false alarm rate is improved, and the pixel points in a local low-intensity area in the target are easily judged as the background, and the detection omission is caused. In addition, the wide application of the SAR system also complicates the imaging environment, so that the imaging result is more severely interfered by noise, namely, the SAR image has strong noise. The constant false alarm rate detection method has poor robustness to noise, so that the detection method has a trend of decreasing detection rate in a detection result along with the enhancement of image noise.
Therefore, aiming at the problems, the SAR image target detection optimization method based on multiple SVD significance fusion is provided, the optimization method can improve the detection rate of an SAR image target detection algorithm under the condition of high resolution, the low false alarm rate is maintained, and meanwhile, the optimization detection method has certain robustness to noise.
Disclosure of Invention
The invention aims to overcome the defect of the existing SAR image target detection method in processing the SAR image under the high resolution condition, so as to improve the detection rate of SAR image target detection, reduce the false alarm rate and the omission rate, and improve the robustness of the SAR image target detection method to noise.
The detailed technical scheme of the invention is as follows:
a SAR image target detection optimization method based on multiple SVD saliency fusion comprises the following steps:
step 1: carrying out multiple SVD decomposition on an original SAR image to obtain an original SAR image and approximate images of a plurality of original SAR images, wherein the method comprises the following steps of:
1.1: for a SAR image I of size p x q, first a first SVD decomposition is performed, i.e. the SAR image I is decomposed into a matrix U of size p x p multiplied by a diagonal matrix of size p x qMultiplied by a combination of matrices V of size q x q.
1.2: will diagonal matrixAll elements on the diagonal of (a) are ordered from big to small to generate a new diagonal matrix +.>
1.3: will diagonal matrixAll non-zero elements on the mid-diagonal are averaged mu 0 Mu 0 All diagonals will be less than μ for the threshold 0 The non-zero element value of (2) is assigned 0, forming a new diagonal matrix +.>
1.4: from diagonal matrixAnd U, V to recover an approximate SAR image I that approximates the original SAR image I 1 。
1.5: repeating the previous steps 1.3 and 1.4, i.e. by diagonal matrixAll non-zero elements on the middle diagonal are averaged to obtain mu i Mu i All diagonals will be less than μ for the threshold i The non-zero element value of (2) is assigned 0 to obtain a new diagonal matrixAnd is made up of->Recovering an approximate SAR image I i+1 Until the number of non-zero diagonal elements in the diagonal matrix is 1.
Wherein, the variable I i Representing an I-th approximate SAR image, when i=0, I 0 Is the original SAR image I; variable(s)To approximate SAR image I i Performing SVD decomposition on the corresponding diagonal matrix; variable mu i For diagonal matrix->The mean of all non-zero elements on the mid-diagonal.
Step 2: extracting features of an original SAR image and an approximate SAR image thereof obtained based on multiple SVD decomposition, wherein the extracted features comprise intensity features, outlier features and consistency features, and the steps are as follows:
2.1: for the original SAR image and n approximate SAR images obtained after the operation in the step 1, extracting intensity features, wherein the intensity feature extraction result is an image gray value because the SAR image is a gray image, so that the intensity feature I of the ith approximate SAR image int (i) The calculation is as formula (1):
2.2, extracting outlier characteristics of the original SAR image and n approximate SAR images obtained after the operation of the step 1, and extracting outlier characteristics I of the ith approximate SAR image ol (i) The calculation method comprises the following steps: firstly, counting the intensity value distribution of all pixel points in the whole image, generating a gray level histogram of the image, and counting the number k of the pixel points with each intensity value in gray levels 0-255 through the histogram j J e (0, 255); then utilize k j Solving probability f of any gray level in image by using total number N of pixels of image j The method comprises the steps of carrying out a first treatment on the surface of the And finally traversing all pixel points in the image, calculating an outlier olv (m, n) of each pixel point in the image, forming an outlier characteristic of the image by the outlier of all pixel points in the image, and calculating the outlier characteristic as shown in a formula (2):
I ol (i)＝R(olv(m,n)) (2)
the variable R (·) represents that the element in the bracket generates a corresponding matrix according to the coordinates thereof; the variable olv (m, n) is an outlier of the pixel point I (m, n) at the position (m, n) in the arbitrary SAR image I', and the calculation formula is shown as formula (3)
olv(m,n)＝-log(f j ),f j ＝k j /N (3)。
And 2.3, extracting consistency characteristics of the original SAR image and the n approximate SAR images obtained after the operation in the step 1.
Step 3: and (3) performing the following operation on the multiple SAR images under the same characteristic obtained in the step (2) so as to obtain characteristic saliency maps of the intensity characteristic, the outlier characteristic and the consistency characteristic respectively.
3.1, carrying out interlayer addition operation on the characteristic graphs of the intensity characteristic, the outlier characteristic and the consistency characteristic, wherein the calculation modes are shown in the formula (4), the formula (5) and the formula (6):
wherein the variable m is the m-th approximate SAR image selected as the middle layer, and the variable s l For the s selected l Amplitude approximation SAR image as peripheral layer s l =m+l, l= ±2, ±3; sign symbolAn inter-layer addition operation symbol, which indicates inter-layer image addition; variable fm int 、fm ol And fm cons An intensity feature enhancement map, an outlier feature enhancement map, and a consistency feature enhancement map of the image, respectively.
3.2, respectively solving the mean value of the intensity characteristic reinforcing diagram, the outlier characteristic reinforcing diagram and the consistency characteristic reinforcing diagram to obtain a central mean value characteristic reinforcing diagram of the corresponding characteristic reinforcing diagram;
where the variable l' is the total number of peripheral and central layers used to calculate the corresponding feature enhancement map.
Respectively making differences between the corresponding central mean characteristic enhancement map and the central layer characteristic enhancement map of the intensity characteristic, the outlier characteristic and the consistency characteristic to obtain a characteristic significant map of the corresponding characteristic;
FM int ＝|Fm int -I int (m)| (10)
FM ol ＝|Fm ol -I ol (m)| (11)
FM cons ＝|Fm cons -I cons (m)| (12)。
step 4: the feature saliency maps are weighted the same, the sum and the average are calculated to obtain a total saliency map, and the calculation formula is formula (13):
step 5: obtaining a total saliency map F through the step 4 all Then, taking the pixel point with the highest saliency value in the graph as the center, taking the size of the required detection target in the image as the radius to obtain a saliency area, and then assigning the saliency value of the saliency area to be 0 to obtain a new total saliency graph F' all 。
Step 6: for the new total saliency map F' all Repeating the operation of the step 5 to obtain a new salient region until no salient region exists in the total salient map; thereby acquiring the transfer track of the salient region and obtaining the distribution of all salient regions.
Wherein step 2.1 is for the consistency characteristic I of the ith approximate SAR image cons (i) The calculation method comprises the following steps: first, a structure tensor matrix St (i) of the SAR image is calculated, and the calculation formula thereof is formula (14):
wherein the variable R hh (i)＝R h (i)·R h (i)，R vv (i)＝R v (i)·R v (i)，R hv (i)＝R h (i)·R v (i) Variable R h (i) And R is v (i) The i-th approximate SAR image gradient in the horizontal direction and the vertical direction respectively;
according to the consistency characteristic that the partial result of the matrix St (i) can reach the image, the calculation formula is formula (15):
compared with the prior art, the invention has the following advantages:
1. by adopting the method with the multiple SVD structures, the method can effectively reserve important information in SAR image data, so that the structural information of a target in the SAR image is reserved more completely in a detection result, and the possibility of missed detection in the detection method is reduced.
2. On the basis of the multiple SVD structure, the intensity characteristic, the outlier characteristic and the consistency characteristic are adopted to obtain the salient region which accords with the target characteristic in the SAR image, the method can effectively improve the detection rate of the detection method and keep the low false alarm rate on the target detection of the high-resolution SAR image, and meanwhile, the salient method based on the multiple SVD structure has certain robustness to noise.
Drawings
FIG. 1 is a raw SAR image under test;
FIG. 2 is a graph showing intensity profile;
FIG. 3 is an outlier feature saliency map;
FIG. 4 is a consistent feature saliency map;
fig. 5 is a total saliency map.
Detailed Description
The specific embodiment of the invention is as follows:
a SAR image target detection optimization method based on multiple SVD saliency fusion comprises the following steps:
step 1: carrying out multiple SVD decomposition on an original SAR image to obtain an original SAR image and approximate images of a plurality of original SAR images, wherein the method comprises the following steps of:
1.1: for a SAR image I of size p x q, first a first SVD decomposition is performed, i.e. the SAR image I is decomposed into a matrix U of size p x p multiplied by a diagonal matrix of size p x qMultiplying by a combination of matrix V of size q x q;
1.2: will diagonal matrixAll elements on the diagonal of (a) are ordered from big to small to generate a new diagonal matrix +.>
1.3: will diagonal matrixAll non-zero elements on the mid-diagonal are averaged mu 0 Mu 0 All diagonals will be less than μ for the threshold 0 The non-zero element value of (2) is assigned 0, forming a new diagonal matrix +.>
1.4: from diagonal matrixAnd U, V to recover an approximate SAR image I that approximates the original SAR image I 1 ；
1.5: repeating the previous steps 1.3 and 1.4, i.e. by diagonal matrixAll non-zero elements on the middle diagonal are averaged to obtain mu i Mu i All diagonals will be less than μ for the threshold i The non-zero element value of (2) is assigned 0 to obtain a new diagonal matrixAnd is made up of->Recovering an approximate SAR image I i+1 Until the number of non-zero diagonal elements in the diagonal matrix is 1;
wherein, the variable I i Representing an I-th approximate SAR image, when i=0, I 0 Is the original SAR image I; variable(s)To approximate SAR image I i Performing SVD decomposition on the corresponding diagonal matrix; variable mu i For diagonal matrix->The mean value of all non-zero elements on the mid-diagonal;
step 2: extracting features of an original SAR image and an approximate SAR image thereof obtained based on multiple SVD decomposition, wherein the extracted features comprise intensity features, outlier features and consistency features, and the steps are as follows:
2.1: for the original SAR image and n approximate SAR images obtained after the operation in the step 1, extracting intensity features, wherein the intensity feature extraction result is an image gray value because the SAR image is a gray image, so that the intensity feature I of the ith approximate SAR image int (i) The calculation is as formula (1):
2.2 for the raw SAR image obtained after step 1 operationExtracting outlier characteristics from the image and n approximate SAR images, and extracting outlier characteristics I of the ith approximate SAR image ol (i) The calculation method comprises the following steps: firstly, counting the intensity value distribution of all pixel points in the whole image, generating a gray level histogram of the image, and counting the number k of the pixel points with each intensity value in gray levels 0-255 through the histogram j J e (0, 255); then utilize k j Solving probability f of any gray level in image by using total number N of pixels of image j The method comprises the steps of carrying out a first treatment on the surface of the And finally traversing all pixel points in the image, calculating an outlier olv (m, n) of each pixel point in the image, forming an outlier characteristic of the image by the outlier of all pixel points in the image, and calculating the outlier characteristic as shown in a formula (2):
I ol (i)＝R(olv(m,n)) (2)
the variable olv (m, n) is an outlier of the pixel point I '(m, n) at the position (m, n) in the arbitrary SAR image I', and the calculation formula is shown as formula (3)
olv(m,n)＝-log(f j ),f j ＝k j /N (3)
2.3, extracting consistency characteristics of the original SAR image and n approximate SAR images obtained after the operation in the step 1;
consistency feature I for the ith approximate SAR image cons (i) The calculation method comprises the following steps: firstly, calculating a structure tensor matrix St (i) of the SAR image, wherein the calculation formula is formula (4):
wherein the variable R hh (i)＝R h (i)·R h (i)，R vv (i)＝R v (i)·R v (i)，R hv (i)＝R h (i)·R v (i) Variable R h (i) And R is v (i) The i-th approximate SAR image gradient in the horizontal direction and the vertical direction respectively;
according to the consistency characteristic that the partial result of the matrix St (i) can reach the image, the calculation formula is formula (5):
step 3: performing the following operation on a plurality of SAR images under the same characteristic obtained in the step 2, so as to obtain characteristic saliency maps of the intensity characteristic, the outlier characteristic and the consistency characteristic respectively;
3.1, carrying out interlayer addition operation on the characteristic graphs of the intensity characteristic, the outlier characteristic and the consistency characteristic, wherein the calculation modes are shown in a formula (6), a formula (7) and a formula (8):
wherein the variable m is the m-th approximate SAR image selected as the middle layer, and the variable s l For the s selected l Amplitude approximation SAR image as peripheral layer s l =m+l, l= ±2, ±3; sign symbolAn inter-layer addition operation symbol, which indicates inter-layer image addition; variable fm int 、fm ol And fm cons Respectively an intensity characteristic enhancement map, an outlier characteristic enhancement map and a consistency characteristic enhancement map of the image;
3.2, respectively solving the mean value of the intensity characteristic reinforcing diagram, the outlier characteristic reinforcing diagram and the consistency characteristic reinforcing diagram to obtain a central mean value characteristic reinforcing diagram of the corresponding characteristic reinforcing diagram;
wherein the variable l' is the total number of peripheral layers and central layers used to calculate the corresponding feature enhancement map;
respectively making differences between the corresponding central mean characteristic enhancement map and the central layer characteristic enhancement map of the intensity characteristic, the outlier characteristic and the consistency characteristic to obtain a characteristic significant map of the corresponding characteristic;
FM int ＝|Fm int -I int (m)| (12)
FM ol ＝|Fm ol -I ol (m)| (13)
FM cons ＝|Fm cons -I cons (m)| (14)
step 4: the feature saliency maps are weighted the same, the sum and the average are calculated to obtain a total saliency map, and the calculation formula is formula (15):
step 5: obtaining a total saliency map F through the step 4 all Then, taking the pixel point with the highest saliency value in the graph as the center, taking the size of the required detection target in the image as the radius to obtain a saliency area, and then assigning the saliency value of the saliency area to be 0 to obtain a new total saliency graph F' all ；
Step 6: for the new total saliency map F' all Repeating the operation of the step 5 to obtain a new salient region until no salient region exists in the total salient map; thereby acquiring the transfer track of the salient region and obtaining the distribution of all salient regions.
Claims (1)
1. A SAR image target detection optimization method based on multiple SVD saliency fusion comprises the following steps:
step 1: carrying out multiple SVD decomposition on an original SAR image to obtain an original SAR image and approximate images of a plurality of original SAR images, wherein the method comprises the following steps of:
1.1: for a SAR image I of size p x q, first a first SVD decomposition is performed, i.e. the SAR image I is decomposed into a matrix U of size p x p multiplied by a diagonal matrix of size p x qMultiplying by a combination of matrix V of size q x q;
1.2: will diagonal matrixAll elements on the diagonal of (a) are ordered from big to small to generate a new diagonal matrix +.>
1.3: will diagonal matrixAll non-zero elements on the mid-diagonal are averaged mu 0 Mu 0 All diagonals will be less than μ for the threshold 0 The non-zero element value of (2) is assigned 0, forming a new diagonal matrix +.>
1.4: from diagonal matrixAnd U, V to recover an approximate SAR image I that approximates the original SAR image I 1 ；
1.5: repeating the previous steps 1.3 and 1.4, i.e. by diagonal matrixAll non-zero elements on the middle diagonal are averaged to obtain mu i Mu i All diagonals will be less than μ for the threshold i The non-zero element value of (2) is assigned 0, resulting in a new diagonal matrix +.>And is made up of->Recovering an approximate SAR image I i+1 Until the number of non-zero diagonal elements in the diagonal matrix is 1;
wherein, the variable I i Representing an I-th approximate SAR image, when i=0, I 0 Is the original SAR image I; variable(s)To approximate SAR image I i Performing SVD decomposition on the corresponding diagonal matrix; variable mu i For diagonal matrix->The mean value of all non-zero elements on the mid-diagonal;
step 2: extracting features of an original SAR image and an approximate SAR image thereof obtained based on multiple SVD decomposition, wherein the extracted features comprise intensity features, outlier features and consistency features, and the steps are as follows:
2.1: for the original SAR image and n approximate SAR images obtained after the operation in the step 1, extracting intensity features, wherein the intensity feature extraction result is an image gray value because the SAR image is a gray image, so that the intensity feature I of the ith approximate SAR image int (i) The calculation is as formula (1):
for the firstConsistency feature I of I approximate SAR images cons (i) The calculation method comprises the following steps: first, a calculation formula for calculating the structure tensor matrix St (i) of the SAR image is formula (14):
wherein the variable R hh (i)＝R h (i)·R h (i)，R vv (i)＝R v (i)·R v (i)，R hv (i)＝R h (i)·R v (i) Variable R h (i) And R is v (i) The i-th approximate SAR image gradient in the horizontal direction and the vertical direction respectively;
according to the consistency characteristic that the partial result of the matrix St (i) can reach the image, the calculation formula is formula (15):
2.2, extracting outlier characteristics of the original SAR image and n approximate SAR images obtained after the operation of the step 1, and extracting outlier characteristics I of the ith approximate SAR image ol (i) The calculation method comprises the following steps: firstly, counting the intensity value distribution of all pixel points in the whole image, generating a gray level histogram of the image, and counting the number k of the pixel points with each intensity value in gray levels 0-255 through the histogram j J e (0, 255); then utilize k j Solving probability f of any gray level in image by using total number N of pixels of image j The method comprises the steps of carrying out a first treatment on the surface of the And finally traversing all pixel points in the image, calculating an outlier olv (m 1, n 1) of each pixel point in the image, forming an outlier characteristic of the image by the outlier values of all pixel points in the image, and calculating the outlier characteristic as shown in a formula (2):
I ol (i)＝R(olv(m1,n1)) (2)
the variable olv (m 1, n 1) is an outlier of the pixel point I '(m 1, n 1) at the position (m 1, n 1) in the arbitrary SAR image I', and the calculation formula is shown as formula (3)
olv(m1,n1)＝-log(f j ),f j ＝k j /N (3)
2.3, extracting consistency characteristics of the original SAR image and n approximate SAR images obtained after the operation in the step 1;
step 3: performing the following operation on a plurality of SAR images under the same characteristic obtained in the step 2, so as to obtain characteristic saliency maps of the intensity characteristic, the outlier characteristic and the consistency characteristic respectively;
3.1, carrying out interlayer addition operation on the characteristic graphs of the intensity characteristic, the outlier characteristic and the consistency characteristic, wherein the calculation modes are shown in the formula (4), the formula (5) and the formula (6):
wherein the variable m is the m-th approximate SAR image selected as the middle layer, and the variable s l For the s selected l Amplitude approximation SAR image as peripheral layer s l =m+l, l= ±2, ±3; sign symbolAn inter-layer addition operation symbol, which indicates inter-layer image addition; variable fm int 、fm ol And fm cons Respectively an intensity characteristic enhancement map, an outlier characteristic enhancement map and a consistency characteristic enhancement map of the image;
3.2, respectively solving the mean value of the intensity characteristic reinforcing diagram, the outlier characteristic reinforcing diagram and the consistency characteristic reinforcing diagram to obtain a central mean value characteristic reinforcing diagram of the corresponding characteristic reinforcing diagram;
wherein the variable l' is the total number of peripheral layers and central layers used to calculate the corresponding feature enhancement map;
respectively making differences between the corresponding central mean characteristic enhancement map and the central layer characteristic enhancement map of the intensity characteristic, the outlier characteristic and the consistency characteristic to obtain a characteristic significant map of the corresponding characteristic;
FM int ＝|Fm int -I int (m)| (10)
FM ol ＝|Fm ol -I ol (m)| (11)
FM cons ＝|Fm cons -I cons (m)| (12)
step 4: the feature saliency maps are weighted the same, the sum and the average are calculated to obtain a total saliency map, and the calculation formula is formula (13):
step 5: obtaining a total saliency map F through the step 4 all Then, taking the pixel point with the highest saliency value in the graph as the center, taking the size of the required detection target in the image as the radius to obtain a saliency area, and then assigning the saliency value of the saliency area to be 0 to obtain a new total saliency graph F' all ；
Step 6: for the new total saliency map F' all Repeating the operation of the step 5 to obtain a new salient region until no salient region exists in the total salient map; thereby acquiring a significant regionAnd a distribution of all salient regions is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010679460.7A CN112001239B (en) | 2020-07-15 | 2020-07-15 | SAR image target detection optimization method based on multiple SVD saliency fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010679460.7A CN112001239B (en) | 2020-07-15 | 2020-07-15 | SAR image target detection optimization method based on multiple SVD saliency fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112001239A CN112001239A (en) | 2020-11-27 |
| CN112001239B true CN112001239B (en) | 2023-11-28 |
Family
ID=73466463
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010679460.7A Active CN112001239B (en) | 2020-07-15 | 2020-07-15 | SAR image target detection optimization method based on multiple SVD saliency fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112001239B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103065136A (en) * | 2013-01-29 | 2013-04-24 | 中国电子科技集团公司第二十八研究所 | Method for recognizing collaborative target in SAR (Synthetic Aperture Radar) image based on visual attention mechanism |
| CN103413127A (en) * | 2013-09-10 | 2013-11-27 | 上海海事大学 | Marine target significance detection method based on spectrum singular value decomposition |
| CN106651937A (en) * | 2016-10-19 | 2017-05-10 | 成都电科智达科技有限公司 | Method for detecting small unmanned aerial vehicle target based on super-pixels and scene prediction |
| CN107680120A (en) * | 2017-09-05 | 2018-02-09 | 南京理工大学 | Tracking Method of IR Small Target based on rarefaction representation and transfer confined-particle filtering |
| WO2018197585A1 (en) * | 2017-04-26 | 2018-11-01 | Skansense S.L.U. | Identifying targets within images |
| CN108830883A (en) * | 2018-06-05 | 2018-11-16 | 成都信息工程大学 | Vision attention SAR image object detection method based on super-pixel structure |
| CN109146925A (en) * | 2018-08-23 | 2019-01-04 | 郑州航空工业管理学院 | Conspicuousness object detection method under a kind of dynamic scene |
| CN109800713A (en) * | 2019-01-21 | 2019-05-24 | 成都信息工程大学 | The remote sensing images cloud detection method of optic increased based on region |
| CN111062321A (en) * | 2019-12-17 | 2020-04-24 | 佛山科学技术学院 | SAR detection method and system based on deep convolutional network |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8559671B2 (en) * | 2008-12-18 | 2013-10-15 | The Regents Of The University Of California | Training-free generic object detection in 2-D and 3-D using locally adaptive regression kernels |
| US10339387B2 (en) * | 2016-03-03 | 2019-07-02 | Brigham Young University | Automated multiple target detection and tracking system |
-
2020
- 2020-07-15 CN CN202010679460.7A patent/CN112001239B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103065136A (en) * | 2013-01-29 | 2013-04-24 | 中国电子科技集团公司第二十八研究所 | Method for recognizing collaborative target in SAR (Synthetic Aperture Radar) image based on visual attention mechanism |
| CN103413127A (en) * | 2013-09-10 | 2013-11-27 | 上海海事大学 | Marine target significance detection method based on spectrum singular value decomposition |
| CN106651937A (en) * | 2016-10-19 | 2017-05-10 | 成都电科智达科技有限公司 | Method for detecting small unmanned aerial vehicle target based on super-pixels and scene prediction |
| WO2018197585A1 (en) * | 2017-04-26 | 2018-11-01 | Skansense S.L.U. | Identifying targets within images |
| CN107680120A (en) * | 2017-09-05 | 2018-02-09 | 南京理工大学 | Tracking Method of IR Small Target based on rarefaction representation and transfer confined-particle filtering |
| CN108830883A (en) * | 2018-06-05 | 2018-11-16 | 成都信息工程大学 | Vision attention SAR image object detection method based on super-pixel structure |
| CN109146925A (en) * | 2018-08-23 | 2019-01-04 | 郑州航空工业管理学院 | Conspicuousness object detection method under a kind of dynamic scene |
| CN109800713A (en) * | 2019-01-21 | 2019-05-24 | 成都信息工程大学 | The remote sensing images cloud detection method of optic increased based on region |
| CN111062321A (en) * | 2019-12-17 | 2020-04-24 | 佛山科学技术学院 | SAR detection method and system based on deep convolutional network |
Non-Patent Citations (5)
| Title |
|---|
| A target detection algorithm for SAR images based on regional probability statistics and saliency analysis;Baohua Zhang等;International Journal of Remote Sensing;第40卷(第4期);第1394-1410页 * |
| 一种基于深层次多尺度特征融合CNN的SAR图像舰船目标检测算法;杨龙等;光学学报;第40卷(第2期);第132-140页 * |
| 一种基于视觉显著性的SAR图像弱目标检测算法;张衡等;电子信息对抗技术(第2期);第3-8页 * |
| 低检测率SAR图像目标检测与识别理论研究;刘说;中国博士学位论文全文数据库信息科技辑(第1期);第三-五章 * |
| 基于高分辨距离像的雷达地面目标识别技术;李龙;中国博士学位论文全文数据库 信息科技辑(第12期);第I136-93页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112001239A (en) | 2020-11-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108983219B (en) | Fusion method and system for image information and radar information of traffic scene | |
| CN108596055B (en) | Airport target detection method of high-resolution remote sensing image under complex background | |
| Shim et al. | Road damage detection using super-resolution and semi-supervised learning with generative adversarial network | |
| CN111914686B (en) | SAR remote sensing image water area extraction method, device and system based on surrounding area association and pattern recognition | |
| Zhang et al. | Region of interest extraction in remote sensing images by saliency analysis with the normal directional lifting wavelet transform | |
| CN112257609A (en) | Vehicle detection method and device based on self-adaptive key point heat map | |
| Bharadwaj Sundra Murthy et al. | Detection of potholes in autonomous vehicle | |
| CN109977968B (en) | SAR change detection method based on deep learning classification comparison | |
| CN113920436A (en) | Remote sensing image marine vessel recognition system and method based on improved YOLOv4 algorithm | |
| CN113239953B (en) | SAR image rotation ship detection method based on directed Gaussian function | |
| CN111666801A (en) | Large-scene SAR image ship target detection method | |
| CN112802063A (en) | Satellite cloud picture prediction method and device and computer readable storage medium | |
| Qin et al. | Multilevel wavelet-SRNet for SAR target recognition | |
| Xu et al. | COCO-Net: A dual-supervised network with unified ROI-loss for low-resolution ship detection from optical satellite image sequences | |
| Li et al. | Ship target detection and recognition method on sea surface based on multi-level hybrid network | |
| Zuo et al. | A remote sensing image semantic segmentation method by combining deformable convolution with conditional random fields | |
| CN105139426B (en) | A kind of video moving object detection method based on undecimated wavelet transform and LBP | |
| EP4323952A1 (en) | Semantically accurate super-resolution generative adversarial networks | |
| CN112001239B (en) | SAR image target detection optimization method based on multiple SVD saliency fusion | |
| CN114332644A (en) | Large-view-field traffic density acquisition method based on video satellite data | |
| JP2018124963A (en) | Image processing device, image recognition device, image processing program, and image recognition program | |
| CN113887652B (en) | Remote sensing image weak and small target detection method based on morphology and multi-example learning | |
| WO2018143278A1 (en) | Image processing device, image recognition device, image processing program, and image recognition program | |
| CN112233042B (en) | Method for rapidly generating large-scene SAR image containing non-cooperative target | |
| CN114913504A (en) | Vehicle target identification method of remote sensing image fused with self-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 |