CN101634705B - Method for detecting target changes of SAR images based on direction information measure - Google Patents
Method for detecting target changes of SAR images based on direction information measure Download PDFInfo
- Publication number
- CN101634705B CN101634705B CN2009100236367A CN200910023636A CN101634705B CN 101634705 B CN101634705 B CN 101634705B CN 2009100236367 A CN2009100236367 A CN 2009100236367A CN 200910023636 A CN200910023636 A CN 200910023636A CN 101634705 B CN101634705 B CN 101634705B
- Authority
- CN
- China
- Prior art keywords
- target
- information
- variation
- difference
- region unit
- 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
Images
Landscapes
- Radar Systems Or Details Thereof (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method for detecting target changes of SAR (synthetic aperture radar) images on the basis of direction information measure, mainly overcoming the disadvantages that the conventional method for detecting target changes containing high-intensify noise has poor noise immunity and low detection accuracy. The method comprises the following implementation procedures: (1) extracting the direction information from two time phases by using the direction measure; (2) constructing a difference image by the difference in the direction information measure of the two time phases; (3) obtaining the general outline of the changed target from the difference image by using the fuzzy C-means clustering method; (4) determining the centre point of the target and the area blocks containing the target information according to the general outline of the target; and (5) filtering the area blocks containing the changed target information, subtracting to obtain difference area blocks, and extracting the changed target of each block by using the method of maximum inter-class threshold value. The invention has the advantages of strong noise immunity and high accuracy of changed targets detection. Therefore, the invention can be applied to detect targets of SAR multitemporal images changes.
Description
Technical field
The invention belongs to technical field of image processing, relate to the image object change detecting method, be applicable to that two width of cloth do not contain object variations information in the phase SAR image simultaneously, by the change-detection of noise severe contamination.
Background technology
Change-detection is intended to obtain interested atural object change information by the difference between the image of areal different times, it is the gordian technique of carrying out in the earth observations application such as forest inventory investigation, soil utilization, covering variation research, environmental hazard assessment, city planning and the monitoring of national defence military situation, has exigence and application prospects.Synthetic-aperture radar SAR has round-the-clock, and the characteristics of round-the-clock are good change-detection information sources, and research SAR Image Change Detection technology has boundless application prospect.
Change-detection is the emphasis and the hot issue of remote sensing area research, and many scholars have carried out classification, analysis from different angles to existing change detecting method.Change detecting method is mainly considered the detection of image to image, is the detection method that grows up on pixel level level.Existing change detecting method can reduce two big classes: a class is the supervision detection method, and another kind of is non-supervision detection method.The former is meant the training field that obtains region of variation according to ground truth, thereby carries out change-detection; The latter is directly to two Data Detection of phase and without any need for extra information simultaneously not.Because the real information on ground is not easy to obtain, therefore the change detecting method of non-supervision is the change detecting method of using always.The common way of non-supervision detection method be directly relatively same position not simultaneously the pixel eigenwert of phase come change detected, usually adopt the mode of mathematic(al) manipulation to produce not alternate simultaneously difference image, again difference image is carried out thresholding and handle, therefrom extract region of variation.In present stage, the change detecting method of image and digital line layout figure, image and map, DEM, DOM and image has been proposed also from the angle of data source.From image registration angle whether, first registration image change-detection and change-detection and the Image registration method of carrying out has simultaneously again been proposed.From the angle of algorithm, many scholars have proposed the comprehensive use of several different methods, and these methods comprise end user's artificial neural networks, markov random file, mathematical morphology and fuzzy logic etc.
Although various change detecting methods have been applied to numerous areas widely, yet overall change-detection still exists many difficulties and problem and remains deep research and solution.Select as the threshold value that changes, and not high to artificial target detection degree of accuracy under the stronger situation of some noises, and be that semi-automatic and manual detection method is main mostly.
Summary of the invention
The objective of the invention is to overcome the shortcoming of above-mentioned existing problem, proposed a kind of SAR image object change detecting method, detect, and improve the accuracy of detection of target with automatic realization SAR image object based on direction information measure.
Realize that the object of the invention technical scheme is the strong target in mutually and extract corresponding impact point when detecting two width of cloth earlier, determines the situation of change of image again by the similarity degree of relative position relation between 2 width of cloth image object points.Its concrete steps comprise as follows:
(1) choose the not SAR image of phase simultaneously of two width of cloth, Shi Xiangzhong includes object variations information, and target information is subjected to the noise severe contamination;
(2) at the single image of choosing, set the yardstick of window, as this center neighborhood of a point, be divided into 0 ° by angle, 45 °, 90 ° with 135 ° of four different directions, window is divided into 2 equal regional f
1And f
2, two zones that each direction is divided into are one group;
(3) calculate the mean value mean (f of two zones of different pixel values in every group respectively
1) and mean (f
2);
(4) calculate the difference of these two averages:
diff(θ)=mean(f
1)
θ-mean(f
2)
θ,θ=0°,45°,90°,135°
Obtain 4 different differences according to 4 directions, choose 4 middle maximal values and estimate for: M=Max (| diff (θ) |) as the direction of this pixel
And the direction of this point is: θ=arg (Max (| diff (θ) |));
(5) utilize direction to estimate the direction information measure that extracts 2 o'clock phases respectively, by the difference structural differences figure of the direction information measure of 2 o'clock phases;
(6) at disparity map, adopt the method for fuzzy C-means clustering, disparity map is divided into 3 classes, and the information that will contain object variations concentrates on a class, obtain the general profile of variation targets by this class;
(7), determine the central point of target and comprise the region unit of target information according to the general profile of target;
(8) at two width of cloth not simultaneously in the phase SAR image, the region unit that contains variation targets information is carried out filtering, and filtered 2 o'clock alpha region pieces are subtracted each other obtain the difference region unit, each difference region unit is extracted the target of its variation with the method for maximum between-cluster variance threshold value.
The present invention has the following advantages compared with prior art:
1. because the directional information of target is affected by noise little, the present invention utilizes direction information measure to determine to comprise the region unit of object variations information, has stronger noise immunity;
2. after the present invention adopts the variation of direction information measure to determine to comprise the region unit of variation targets information, region unit is made change-detection more separately, the secondary change-detection has improved accuracy of detection;
3. the present invention determines that each comprises the region unit of variation targets, extract target at each region unit with threshold method separately, and general classic method is after obtaining disparity map, in order to obtain detecting target, the method of threshold value is to get threshold value at entire image, this threshold value is inappropriate, compares general classic method, has improved accuracy of detection.
Description of drawings
Fig. 1 is a realization flow synoptic diagram of the present invention;
Fig. 2 is 4 directional diagrams of direction information measure among the present invention;
Fig. 3 is the window figure of filtering among the present invention;
Fig. 4 is test two width of cloth phase remote sensing images simultaneously not among the present invention;
Fig. 5 is the direction information measure figure that trial image is extracted;
Fig. 6 is that the trial image direction is estimated disparity map;
Fig. 7 estimates the variation targets profile diagram that obtains after the disparity map classification to the trial image direction;
Fig. 8 is a region unit of determining to contain object variations information in original two width of cloth images;
Fig. 9 is the disparity map that contains the region unit of object variations information;
Figure 10 is the variation targets of extracting at each difference block;
Figure 11 is the final change-detection result of test.
Embodiment
As shown in Figure 1, specific implementation process of the present invention is as follows:
Step 1. is chosen the not SAR image of phase simultaneously of two width of cloth.
Experimental image is chosen is two width of cloth SAR image of phase simultaneously not of traffic pattern, as shown in Figure 4, wherein Fig. 4 (a) is first o'clock phase, Fig. 4 (b) is second o'clock phase, change information is the variation of Aircraft Target, do not include object variations information in the phase SAR image simultaneously, and target information is subjected to the noise severe contamination.
Step 2. pair choose two width of cloth not simultaneously the SAR image of phase extract directional information respectively.
2a) at the single image of choosing, set the yardstick of window,, be divided into 0 ° by angle as this center neighborhood of a point, 45 °, 90 ° with 135 ° of four different directions, as shown in Figure 2, wherein Fig. 2 (a) is 0 ° of direction window, Fig. 2 (b) is 45 ° of direction windows, Fig. 2 (c) is 90 ° of direction windows, and Fig. 2 (d) is 135 ° of direction windows, and window is divided into 2 equal regional f
1And f
2, two zones that each direction is divided into are one group;
2b) calculate the mean value mean (f of two zones of different pixel values in every group respectively
1) and mean (f
2);
2c) calculate the difference of these two averages:
diff(θ)=mean(f
1)
θ-mean(f
2)
θ,θ=0°,45°,90°,135°
Obtain 4 different differences according to 4 directions, choose 4 middle maximal values and estimate for: M=Max (| diff (θ) |), and the direction of this point is as the direction of this pixel: θ=arg (Max (| diff (θ) |)),
According to the method described above, extract the not directional information of phase SAR image simultaneously of two width of cloth respectively, it is 13x13 that the present invention extracts the window size of selecting for use when direction is estimated, from shown in Figure 5, Fig. 5 (a) is that the direction of first o'clock phase is estimated figure, and Fig. 5 (b) is that the direction of second o'clock phase is estimated figure, from Fig. 5 (a) and 5 (b) as can be seen, although contain high intensity noise among the former figure, direction is estimated affected by noise hardly.
Step 3. structure grain is estimated disparity map.
Utilize above-mentioned direction estimate to two width of cloth not simultaneously the SAR image of phase extract directional information, the direction that obtains is estimated figure X
1And X
2, by X
1And X
2The difference that direction is estimated obtains direction and estimates disparity map X
d, as shown in Figure 6, as can be seen from Figure 6, direction is estimated the profile information that significantly only keeps variation targets in the disparity map.
Step 4. determines to comprise the region unit of target information.
4a) direction is estimated disparity map, adopt the method for fuzzy C-means clustering, direction is estimated disparity map be divided into definite variation zone, uncertain variation zone and confusion region 3 classes, get and determine the general profile of variation zone as target, as shown in Figure 7, as can be seen from Figure 7, obtain the general profile that binary map has shown variation targets after the classification;
4b) in target general profile binary map, search for, from searching first white point opening entry coordinate, in the white point that searches is the window neighborhood of center 7*7 size, just continue search if contain white point, and the coordinate of record white point, continue moving window, in the window neighborhood that with the white point is center 7*7 size, finally search less than white point, write down the coordinate of all white points that this time search, find out the maximal value X of all horizontal ordinates of white point that this time search
MaxWith minimum value X
Min, (X by formula
Max+ X
MinThe horizontal ordinate of central point is determined in)/2, the ordinate of central point also can obtain by the method, from all white point coordinates that this searches, find out and contain maximal value and minimum value coordinate points respectively, finally determine to comprise the rectangular area piece of this target, according to said method, continue search, up to determining that all contain the region unit of target, as shown in Figure 8, Fig. 8 (a) comprised the region unit figure of object variations information at first o'clock in mutually, and Fig. 8 (b) comprised the region unit figure of object variations information at second o'clock in mutually.
Step 5. structure contains the region unit disparity map of object variations information.
At two width of cloth not simultaneously in the phase SAR image, the region unit that contains variation targets information is carried out filtering, in the present invention, adopt the trim-meaning filtering method, remove Gaussian noise and salt-pepper noise, filtering as shown in Figure 3, the pixel of current point is f (i, j), at the window W (m that with this pixel is central point
t* m
t) in, m
tBe the width of window, the window interior pixel is pressed the ascending ordering of gray-scale value
The value of filtering rear hatch central point is
Wherein λ is a filter factor, and getting λ in the present invention is 0.3, and window size is got 5*5,
The alpha region piece subtracts each other and obtains the difference region unit during at last with filtered two width of cloth, and as shown in Figure 9, as can be seen from Figure 9, the target of each variation all in a little region unit, therefore can be handled region unit separately, extracts variation targets.
Step 6. is to each difference region unit, employing maximum between-cluster variance threshold method extracts the variation targets in each piece, as shown in figure 10, as can be seen from Figure 10, detect the Aircraft Target of variation in each independent region unit accurately, in this group laboratory, detect the variation of 5 Aircraft Target altogether, after the Aircraft Target of the variation of extracting, according to the position of each variation targets in former figure, obtain final change-detection result, as shown in figure 11, from the result of Figure 11 more accurate Aircraft Target that detects variation of the present invention as can be seen, and the result is not subjected to noise.
Claims (2)
1. the SAR image object change detecting method based on direction information measure comprises the steps:
(1) choose the not SAR image of phase simultaneously of two width of cloth, Shi Xiangzhong includes object variations information, and target information is subjected to the noise severe contamination;
(2) at the single image of choosing, set the yardstick of window, as this center neighborhood of a point, be divided into 0 ° by angle, 45 °, 90 ° with 135 ° of four different directions, window is divided into 2 equal regional f
1And f
2, two zones that each direction is divided into are one group;
(3) calculate the mean value mean (f of two zones of different pixel values in every group respectively
1) and mean (f
2);
(4) calculate the difference of these two averages:
diff(θ)=mean(f
1)
θ-mean(f
2)
θ,θ=0°,45°,90°,135°
Obtain 4 different differences according to 4 directions, choose 4 middle maximal values and estimate for: M=Max (| diff (θ) |) as the direction of this pixel
And the direction of this point is: θ=arg (Max (| diff (θ) |));
(5) utilize direction to estimate the direction information measure that extracts 2 o'clock phases respectively, by the difference structural differences figure of the direction information measure of 2 o'clock phases;
(6) at disparity map, adopt the method for fuzzy C-means clustering, direction is estimated disparity map be divided into definite variation zone, uncertain variation zone and confusion region 3 classes, and the information that will contain object variations concentrates on a class, obtain the general profile of variation targets by this class;
(7), determine the central point of target and comprise the region unit of target information according to the general profile of target:
7a) in target general profile binary map, search for,, in the white point that searches is the window neighborhood of center 7*7 size, just continue search if contain white point from searching first white point opening entry coordinate, and the coordinate of record white point;
7b) continue moving window, in the window neighborhood that with the white point is center 7*7 size, finally search, write down the coordinate of all white points that this time search, find out the maximal value X of all horizontal ordinates of white point that this time search less than white point
MaxWith minimum value X
Min, (X by formula
Max+ X
MinThe horizontal ordinate of central point is determined in)/2, and the ordinate of central point also can obtain by the method;
7c) from all white point coordinates that this searches, find out and contain maximal value and minimum value coordinate points respectively, finally determine to comprise the rectangular area piece of this target, according to said method, continue search, up to determining that all contain the region unit of target;
(8) at two width of cloth not simultaneously in the phase SAR image, the region unit that contains variation targets information is carried out filtering, and the alpha region piece subtracts each other and obtains the difference region unit during with filtered two width of cloth, each difference region unit is extracted the target of its variation with the method for maximum between-cluster variance threshold value.
2. image object change detecting method according to claim 1, the described target of each difference region unit being extracted its variation with the method for maximum between-cluster variance threshold value of step (8) wherein, be in each piece, to extract target earlier, turn back among the former figure by the position of each piece in former figure again, obtain the change-detection result at last.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009100236367A CN101634705B (en) | 2009-08-19 | 2009-08-19 | Method for detecting target changes of SAR images based on direction information measure |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009100236367A CN101634705B (en) | 2009-08-19 | 2009-08-19 | Method for detecting target changes of SAR images based on direction information measure |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101634705A CN101634705A (en) | 2010-01-27 |
CN101634705B true CN101634705B (en) | 2011-12-07 |
Family
ID=41593954
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2009100236367A Active CN101634705B (en) | 2009-08-19 | 2009-08-19 | Method for detecting target changes of SAR images based on direction information measure |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101634705B (en) |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101770027B (en) * | 2010-02-05 | 2012-05-16 | 河海大学 | Ground surface three-dimensional deformation monitoring method based on InSAR and GPS data fusion |
CN102005049B (en) * | 2010-11-16 | 2012-05-09 | 西安电子科技大学 | Unilateral generalized gaussian model-based threshold method for SAR (Source Address Register) image change detection |
CN102081799B (en) * | 2011-01-10 | 2012-12-12 | 西安电子科技大学 | Method for detecting change of SAR images based on neighborhood similarity and double-window filtering |
CN103955943A (en) * | 2014-05-21 | 2014-07-30 | 西安电子科技大学 | Non-supervision change detection method based on fuse change detection operators and dimension driving |
CN105160666B (en) * | 2015-08-25 | 2018-03-06 | 西安电子科技大学 | SAR image change detection based on Non-Stationary Analysis and condition random field |
CN105389817B (en) * | 2015-11-04 | 2017-11-14 | 河海大学 | A kind of two phase remote sensing image variation detection methods |
CN106443672B (en) * | 2016-08-30 | 2019-03-29 | 西安电子科技大学 | A kind of orientation multichannel SAR signal adaptive reconstructing method |
CN107256409B (en) * | 2017-05-22 | 2021-01-01 | 西安电子科技大学 | High-resolution SAR image change detection method based on SAE and significance detection |
CN109801306B (en) * | 2019-01-22 | 2020-10-27 | 首都师范大学 | Tidal ditch extraction method based on high resolution remote sensing image |
CN113359133B (en) * | 2021-06-03 | 2022-03-15 | 电子科技大学 | Object-oriented change detection method for collaborative optical and radar remote sensing data |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5546084A (en) * | 1992-07-17 | 1996-08-13 | Trw Inc. | Synthetic aperture radar clutter reduction system |
CN101299243A (en) * | 2008-06-27 | 2008-11-05 | 西安电子科技大学 | Method of image segmentation based on immune spectrum clustering |
CN101498789A (en) * | 2009-02-25 | 2009-08-05 | 中国测绘科学研究院 | Ground object target classification method and apparatus based on polarimetric synthetic aperture radar |
-
2009
- 2009-08-19 CN CN2009100236367A patent/CN101634705B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5546084A (en) * | 1992-07-17 | 1996-08-13 | Trw Inc. | Synthetic aperture radar clutter reduction system |
CN101299243A (en) * | 2008-06-27 | 2008-11-05 | 西安电子科技大学 | Method of image segmentation based on immune spectrum clustering |
CN101498789A (en) * | 2009-02-25 | 2009-08-05 | 中国测绘科学研究院 | Ground object target classification method and apparatus based on polarimetric synthetic aperture radar |
Non-Patent Citations (2)
Title |
---|
田小林等.具有细节保护的自适应邻域SAR图像分割.《模式识别与人工智能》.2008,第21卷(第4期),第527页-534页. * |
田小林等.基于PSO优化空间约束聚类的SAR图像分割.《电子学报》.2008,第36卷(第3期),第453-457页. * |
Also Published As
Publication number | Publication date |
---|---|
CN101634705A (en) | 2010-01-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101634705B (en) | Method for detecting target changes of SAR images based on direction information measure | |
CN103077384B (en) | A kind of method and system of vehicle-logo location identification | |
Rottensteiner et al. | Using the Dempster–Shafer method for the fusion of LIDAR data and multi-spectral images for building detection | |
Guo et al. | Relevance of airborne lidar and multispectral image data for urban scene classification using Random Forests | |
CN102855622B (en) | A kind of infrared remote sensing image sea ship detection method based on significance analysis | |
CN104361582B (en) | Method of detecting flood disaster changes through object-level high-resolution SAR (synthetic aperture radar) images | |
CN105139015B (en) | A kind of remote sensing images Clean water withdraw method | |
CN101950364A (en) | Remote sensing image change detection method based on neighbourhood similarity and threshold segmentation | |
CN106128121B (en) | Vehicle queue length fast algorithm of detecting based on Local Features Analysis | |
CN102567994B (en) | Infrared small target detection method based on angular point gaussian characteristic analysis | |
Er-Sen et al. | An adaptive edge-detection method based on the canny operator | |
CN105976376B (en) | High-resolution SAR image target detection method based on component model | |
CN104766079B (en) | A kind of remote method for detecting infrared puniness target | |
CN103632363A (en) | Object-level high-resolution remote sensing image change detection method based on multi-scale fusion | |
CN101975940A (en) | Segmentation combination-based adaptive constant false alarm rate target detection method for SAR image | |
CN104809433A (en) | Zebra stripe detection method based on maximum stable region and random sampling | |
CN102289948A (en) | Multi-characteristic fusion multi-vehicle video tracking method under highway scene | |
CN101482969B (en) | SAR image speckle filtering method based on identical particle computation | |
CN103383733A (en) | Lane video detection method based on half-machine study | |
CN101551851A (en) | Infrared image target identification method | |
CN106204594A (en) | A kind of direction detection method of dispersivity moving object based on video image | |
CN101976504A (en) | Multi-vehicle video tracking method based on color space information | |
CN105469046A (en) | Vehicle model identification method based on PCA and SURF characteristic cascade | |
CN105787950A (en) | Infrared image sea-sky-line detection algorithm based on line gradient accumulation | |
CN103049788A (en) | Computer-vision-based system and method for detecting number of pedestrians waiting to cross crosswalk |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |