CN105869138B - Hanging sonar azimuth correction method based on image matching - Google Patents
Hanging sonar azimuth correction method based on image matching Download PDFInfo
- Publication number
- CN105869138B CN105869138B CN201510061565.5A CN201510061565A CN105869138B CN 105869138 B CN105869138 B CN 105869138B CN 201510061565 A CN201510061565 A CN 201510061565A CN 105869138 B CN105869138 B CN 105869138B
- Authority
- CN
- China
- Prior art keywords
- image
- sonar
- value
- template
- column
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a hanging sonar azimuth correction method based on image matching, which is characterized by comprising the following steps: step 1, carrying out image grid processing on the sonar image: step 2, performing binary segmentation on the output of image grid processing; step 3, selecting a reference template, wherein the reference template is any one frame on a sonar image time sequence or is obtained by counting continuous multiframes on the time sequence; step 4, calculating a correlation matrix of the image after binary segmentation and a reference template according to a similarity measure principle; step 5, judging the offset information of the current image according to the correlation matrix; and 6, correcting the binary-segmented image according to the offset information. The method of the invention effectively corrects the orientation of the hanging sonar.
Description
Technical Field
The invention relates to the technical field of underwater active detection, in particular to a hanging sonar azimuth correction method based on image matching.
Background
The suspended sonar is widely applied to helicopters, ships and docks and can be used for monitoring and detecting marine environments, marine organisms and underwater human activities. Due to the limitation of a platform and cost, the suspended sonar is not provided with a pointing compass and a posture server, and is influenced by surge, so that the sonar inevitably rotates and swings underwater. In the absence of attitude compensation, the sonar image will have a phenomenon of shaking left and right. The current image course offset is obtained through an image matching technology, and the phenomenon of left-right shaking of the hanging sonar image can be effectively corrected.
Timeliness and matching accuracy are two important indexes of an image matching algorithm. Current image matching methods are basically classified into 3 categories: region-based matching methods, feature-based matching methods, interpretation-based matching methods. In the method based on the region, the timeliness of the cross correlation method is higher, but the matching precision is poorer; the mutual information interaction variance method is high in matching precision and poor in timeliness. The characteristic-based method is high in timeliness, but the matching accuracy depends on the quality of characteristic extraction. The interpretation-based image matching method is based on correct interpretation of the image, so that the requirements on the quality and a priori knowledge of the image are high. Considering the suspended sonar, besides the course change, there is a change in the pitch and roll, which causes the beam to irradiate to the water bottom or water surface, causing the reverberation background of the sonar image to change dramatically. By combining the advantages of different methods, the image feature matching method based on binary segmentation is considered and higher matching precision and timeliness are guaranteed through image segmentation.
Disclosure of Invention
The invention aims to solve the technical problem that the sonar image shakes left and right due to unstable underwater postures of the existing hanging sonar. The invention provides a hanging sonar azimuth correction method based on image matching to solve the problems.
The invention provides a hanging sonar azimuth correction method based on image matching, which comprises the following steps:
step 1, carrying out image grid processing on the sonar image:
h 1(.) is an image segmentation function,f i,j (r,c) Is a sonar image (r,c) Is a point at the upper left and has a size ofR×CTo (1) aiLine, firstjThe sub-picture of the column region,g i,j is composed off i,j (r,c) The second of the image after the image segmentation processingiLine and firstjColumn pixel values;
step 2, performing binary segmentation on the output of the image grid processing:
a i,j as second of the binary-divided imageiLine, firstjThe values of the pixels of the column are,dtin order to binarize the decision domain value,meanin the form of a function of the mean value,stdin the form of a function of the variance,Ain order to correct the value of the data,Athe value of (A) is between 0 and the signal-to-noise ratio of the sonar image, the larger the signal-to-noise ratio is,Athe larger the value of (A) is;
step 3, selecting a reference template, wherein the reference template is any one frame on a sonar image time sequence or is obtained by counting continuous multiframes on the time sequence;
step 4, calculating a correlation matrix of the image after binary segmentation and a reference template according to a similarity measure principle;
step 5, judging the offset information of the current image according to the correlation matrix;
and 6, correcting the binary-segmented image according to the offset information.
Further, when the reference template is obtained by counting a plurality of continuous frames in the time sequence of the sonar image, the reference template is in the ith row, and the pixel value in the jth column is:
；
is the pixel value of the ith row and the jth column of any frame in time series.
Further, the calculation formula of the correlation matrix of the binary-segmented image and the reference template is
，
；b ’ Is the image of the template and is,
is a first of a templatem-iLine and firstn-jColumn pixel values;cas a template pieceAnd (3) matching an output correlation matrix,c m,n is the first of the correlation matrixmLine and firstnA column correlation value; (cv,cr,cc) Means thatcHas a maximum value ofcv，cvIn thatcTo (1) acrLine and firstccColumns;maxthe maximum value is taken.
Further, in the present invention,a 0is a certain frame of image in the image sequence whenb ’ ∈a 0Time, i.e. image templateb ’ As an imagea 0In a step ofi 0,j 0) Is the size of the upper left dot isM×NThe area of (a) is,a 0has a size ofW×HThe limiting case may be b ’ ＝a 0And then:
(cv 0,cr 0,cc 0) Representing imagesa 0And the area thereof b ’ Is self-correlation matrix ofc 0Has a maximum value ofcv 0，cv 0In the correlation matrixc 0To (1) acr 0Line, firstcc 0Columns; according to the relevant principle, whencv 0Satisfy threshold valueDT(.), the image shift amount is:
rmfor real-time imagesaThe amount of the row offset is such that,cmfor real-time imagesaThe column offset.
The image-matching-based hanging sonar azimuth correction method according to claim 5, wherein the first image after correctioniLine, firstjThe column pixel values are:
。
compared with the prior art, the invention has the advantages that:
(1) the invention can improve the adaptability of the equipment, does not need to install a pointing compass and an attitude server under the same condition, and reduces the development cost of the equipment.
(2) The method has higher timeliness and precision, the accuracy of correcting the image orientation with slowly changing background reaches 100%, and the accuracy of correcting the image orientation with violently changing background is better than 80%.
Drawings
FIG. 1 is a process flow diagram of the present invention.
FIG. 2 is a schematic diagram of the simulation results of the principle of the present invention, wherein (a) is the template, (b) is the real-time image, (c) is the correlation center, (40, 30) is the correlation center, and (d) is the corrected image.
FIG. 3 is a diagram showing the simulation results of artificial deviation of actual data according to the present invention, wherein (a) is a template, (b) is a real-time image, (c) is a correlation center, (10, 0) is the correlation center, and (d) is a corrected image;
FIG. 4 is a graph illustrating actual data steady state background single template processing results of the present invention; (a) template, (b) 56 th frame image, (c) correlation center which is (-4, 0), and (d) 56 th frame image modified image;
FIG. 5 is a graph illustrating the actual data non-stationary background single template processing results of the present invention; (a) template, (b) 1412 th frame image, (c) correlation center, (1, 0) corrected 1412 th frame image, and (d) corrected 1412 th frame image;
FIG. 6 is a diagram illustrating the actual data non-stationary background time-varying template processing results of the present invention; (a) template, (b) real-time image, (c) correlation center (-3, 0), and (d) corrected image.
Detailed Description
The invention is further illustrated below with reference to the figures and examples.
Referring to fig. 1, the method for correcting the orientation of a suspended sonar based on image matching provided by the present invention includes:
step 1, acquiring a two-dimensional sonar image sequence, and performing image grid processing on any sonar image:
h 1(.) is an image segmentation function,f i,j (r,c) Is a sonar image (r,c) Is a point at the upper left and has a size ofR×CTo (1) aiLine, firstjThe sub-picture of the column region,g i,j is composed off i,j (r,c) Image segmentation of the first imageiLine and firstjColumn pixel values.
Step 2, performing binary segmentation on the output after image grid processing:
a i,j as second of the binary-divided imageiLine, firstjThe values of the pixels of the column are,dtin order to binarize the decision domain value,meanin the form of a function of the mean value,stdin the form of a function of the variance,
in order to correct the value of the data,Athe value of (A) is between 0 and the signal-to-noise ratio of the sonar image, the larger the signal-to-noise ratio is,Athe larger the value of (A), the more practical the skilled person can select
The value of (a).
Step 3, the reference template can select any frame on the sonar image time sequence, and can also be obtained by counting continuous multiframes on the time sequence:
b ’ i,j for successive multiframesb i,j To (1) aiLine, firstjThe statistical value of the pixels of a column,
is the second correction value.
And 4, calculating a correlation matrix of the current image and the reference template according to a similarity measure principle:
b ’ is the image of the template and is,cthe correlation matrix is output for the template matching,c m,n is the first of the correlation matrixmLine and firstnThe value of the column correlation is set,
is a first of a templatem-iLine and firstn-jColumn pixel values; (cv,cr,cc) Means thatcHas a maximum value ofcv，cvIn thatcTo (1) acrLine and firstccThe columns of the image data are,maxtaking the maximum value.
Step 5, searching a peak value of the correlation matrix, and judging the offset information of the current image according to the correlation matrix:
a 0is a certain frame of image in the image sequence whenb ’ ∈a 0When it is a templateb ’ As an imagea 0In a step ofi 0,j 0) Is the size of the upper left dot isM×NThe area of (a) is,a 0has a size ofW×HThe limiting case may beb ’ ＝a 0And then:
(cv 0,cr 0,cc 0) As an imagea 0And the area thereofb ’ Is self-correlation matrix ofc 0Maximum value ofcv 0In the first placecr 0Line, firstcc 0And (4) columns. According to the relevant principle, whencv 0Satisfy threshold valueDT(.), the image shift amount is:
rmfor real-time imagesaThe amount of the row offset is such that,cmfor real-time imagesaThe column offset.
And 6, image correction:
a ’ i,j is the corrected imageiLine, firstjColumn pixel values.
The image dithering correction is carried out on a certain type of hanging anti-frogman sonar by using the method, the image matrix is 192 wave beams (360 degrees), 1000 distance pixels (400 meters), and h1 (.) is realized by selecting a median filter, the processing effect is shown in a graph from 1 to 5, and the performance statistics is shown in a table 1.
TABLE 1 Multi-set actual data processing Performance statistics
As can be seen from the table 1, the method provided by the invention can correct the image deformation caused by sonar jitter, and the correction accuracy rate meets the application requirements.
Fig. 2 shows that (a) is a selected template, (b) is a real-time image, (c) is a schematic diagram of a correlation matrix peak value, and (d) is a corrected image, and it can be seen from fig. 2 that the correlation matrix peak value is effectively corrected by the method provided by the present invention.
In fig. 3, an arbitrary frame image (a) on the sonar image time sequence is selected as a template, then the template image is deviated by 10 degrees to the left to obtain a real-time image (b), then the method provided by the present invention is used for correction to obtain a corrected image (d), and the left deviation can be corrected by comparing (a) and (d).
FIG. 4 is a schematic diagram of the method of the present invention applied to a steady state background.
Fig. 5 and 6 are schematic diagrams of the method provided by the invention applied to an unsteady background.
Although the present invention has been described with reference to the preferred embodiments, it is not intended to limit the present invention, and those skilled in the art can make variations and modifications of the present invention without departing from the spirit and scope of the present invention by using the methods and technical contents disclosed above.
Claims (5)
1. A hanging sonar azimuth correction method based on image matching is characterized by comprising the following steps:
step 1, carrying out image grid processing on the sonar image:
gi,j＝h1(fk,l(r,c))
h1(. is an image segmentation function, fk,l(R, C) is the sonar image, (R, C) is the upper left point, the k-th row with the size R x C, the l-th column area subgraph, gi,jIs fk,l(r, c) pixel values of ith row and jth column of the image after the image segmentation processing;
step 2, performing binary segmentation on the output of the image grid processing:
dt＝mean(g)+std(g)+A
ai,jthe image is a binary segmented image, wherein the ith row and the jth column of the image are pixel values, dt is a binary decision domain value, mean is a mean function, std is a variance function, A is a correction value, the value of A is between 0 and the signal-to-noise ratio of the sonar image, and the larger the signal-to-noise ratio is, the larger the value of A is;
step 3, selecting a reference template, wherein the reference template is any one frame on a sonar image time sequence or is obtained by counting continuous multiframes on the time sequence;
step 4, calculating a correlation matrix of the image after binary segmentation and a reference template according to a similarity measure principle;
step 5, judging the offset information of the current image according to the correlation matrix;
and 6, correcting the binary-segmented image according to the offset information.
2. The hanging sonar azimuth correction method based on image matching according to claim 1, wherein when the reference template is obtained by counting a plurality of consecutive frames in a sonar image time sequence, the reference template is in an ith row, and a jth column has a pixel value: b'i,j＝mean(bi,j)+std(bi,j)+A；bi,jIs the pixel value of the ith row and the jth column of any frame in time series.
3. The hanging sonar azimuth correction method based on image matching according to claim 2, wherein a calculation formula of a correlation matrix of the binary-segmented image and the reference template is
(cv, cr, cc) ═ max (c); b 'is a template image, b'm-i,n-jPixel values of the m-i th row and the n-j th column of the template are taken as the pixel values; c is a template matching output correlation matrix, cm,nThe correlation value of the mth row and the nth column of the correlation matrix; (cv, cr, cc) means that the maximum value of c is cv, and cv is on the cr row and the cc column of c; max is taken to be the maximum value.
4. The hanging sonar azimuth correction method based on image matching according to claim 3, wherein a0For a certain frame of image in the image sequence, when b' belongs to a0When the image template b' is the image a0Middle school with (i)0,j0) Is the region of size M × N of the upper left dot, a0The dimension is W × H, and the limit may be b' ═ a0And then:
cr0＝i0+M
cc0＝j0+N
cv0＝c0(cr0，cc0)
(cv0,cr0,cc0) Representation image a0Autocorrelation matrix c with its region b0Maximum value of (2) is cv0，cv0In the correlation matrix c0Cr of (2)0Line, cc0Columns; according to the relevant principle, when cv0When the threshold value DT (·) is satisfied, the image shift amount is:
rm＝cr-cr0
cm＝cc-cc0
rm is the offset of a row of the live image, and cm is the offset of a column of the live image.
5. The hanging sonar azimuth correction method based on image matching according to claim 4, wherein the pixel values of the ith row and the jth column of the corrected image are as follows: a'i,j＝ai+rm,j+cm。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510061565.5A CN105869138B (en) | 2015-02-06 | 2015-02-06 | Hanging sonar azimuth correction method based on image matching |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510061565.5A CN105869138B (en) | 2015-02-06 | 2015-02-06 | Hanging sonar azimuth correction method based on image matching |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105869138A CN105869138A (en) | 2016-08-17 |
| CN105869138B true CN105869138B (en) | 2021-09-10 |
Family
ID=56623589
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510061565.5A Expired - Fee Related CN105869138B (en) | 2015-02-06 | 2015-02-06 | Hanging sonar azimuth correction method based on image matching |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105869138B (en) |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5155706A (en) * | 1991-10-10 | 1992-10-13 | Westinghouse Electric Corp. | Automatic feature detection and side scan sonar overlap navigation via sonar image matching |
| DE10054814A1 (en) * | 2000-11-04 | 2002-05-08 | Philips Corp Intellectual Pty | Images registration method e.g. for medical image matching, magnetic resonance tomography, involves transforming discrete images into registered images |
| CN101034476A (en) * | 2007-03-29 | 2007-09-12 | 上海大学 | Method for generating underwater non-shadow sonar remote sensing orthographic digital image by computer |
| CN101770582B (en) * | 2008-12-26 | 2013-05-08 | 鸿富锦精密工业（深圳）有限公司 | Image matching system and method |
| CN102096069B (en) * | 2010-12-17 | 2012-10-03 | 浙江大学 | Real-time processing system and method for phased array three-dimensional acoustic camera sonar |
| CN102313547B (en) * | 2011-05-26 | 2013-02-13 | 东南大学 | Vision navigation method of mobile robot based on hand-drawn outline semantic map |
| CN102622732A (en) * | 2012-03-14 | 2012-08-01 | 上海大学 | Front-scan sonar image splicing method |
| CN103226828B (en) * | 2013-04-09 | 2015-09-30 | 哈尔滨工程大学 | The method for registering images of a kind of underwater voice 3 D visual imaging |
| CN104268894B (en) * | 2014-10-17 | 2017-08-25 | 盐城工学院 | A kind of sectional slice method for registering images judged based on object pixel projection |
-
2015
- 2015-02-06 CN CN201510061565.5A patent/CN105869138B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN105869138A (en) | 2016-08-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11488308B2 (en) | Three-dimensional object detection method and system based on weighted channel features of a point cloud | |
| CN106960446B (en) | Unmanned ship application-oriented water surface target detection and tracking integrated method | |
| CN108108746B (en) | License plate character recognition method based on Caffe deep learning framework | |
| CN109344702B (en) | Pedestrian detection method and device based on depth image and color image | |
| CN108960135B (en) | Dense ship target accurate detection method based on high-resolution remote sensing image | |
| CN107680116B (en) | Method for monitoring moving target in video image | |
| CN112837303A (en) | Defect detection method, device, equipment and medium for mold monitoring | |
| CN103955930B (en) | Motion parameter estimation method based on gray integral projection cross-correlation function characteristics | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN103455991A (en) | Multi-focus image fusion method | |
| CN107248174A (en) | A kind of method for tracking target based on TLD algorithms | |
| WO2017113692A1 (en) | Method and device for image matching | |
| CN104079800A (en) | Shaking preventing method for video image in video surveillance | |
| CN111178193A (en) | Lane line detection method, lane line detection device and computer-readable storage medium | |
| CN112652020A (en) | Visual SLAM method based on AdaLAM algorithm | |
| CN108648205A (en) | A kind of sub-pixel edge detection method | |
| CN106886764B (en) | Panic degree calculation method and device based on deep learning | |
| CN113899349A (en) | Sea wave parameter detection method, equipment and storage medium | |
| CN110490170B (en) | Face candidate frame extraction method | |
| CN105869138B (en) | Hanging sonar azimuth correction method based on image matching | |
| CN107067397A (en) | A kind of image grading method based on infrared image complexity | |
| CN109087334B (en) | Target tracking method based on significant context sparse representation | |
| CN114863258A (en) | Method for detecting small target based on visual angle conversion in sea-sky-line scene | |
| CN112419407B (en) | Cloud cluster displacement vector calculation method and device based on cloud cluster edge recognition | |
| Okarma et al. | A fast image analysis technique for the line tracking robots |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210910 Termination date: 20220206 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |