CN112991309A - Automatic detection method for internal wave region of satellite SAR image - Google Patents
Automatic detection method for internal wave region of satellite SAR image Download PDFInfo
- Publication number
- CN112991309A CN112991309A CN202110329744.8A CN202110329744A CN112991309A CN 112991309 A CN112991309 A CN 112991309A CN 202110329744 A CN202110329744 A CN 202110329744A CN 112991309 A CN112991309 A CN 112991309A
- Authority
- CN
- China
- Prior art keywords
- image
- internal wave
- sar
- internal
- wave
- 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.)
- Withdrawn
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 43
- 238000012545 processing Methods 0.000 claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 20
- 238000001228 spectrum Methods 0.000 claims description 29
- 229910052704 radon Inorganic materials 0.000 claims description 25
- SYUHGPGVQRZVTB-UHFFFAOYSA-N radon atom Chemical compound [Rn] SYUHGPGVQRZVTB-UHFFFAOYSA-N 0.000 claims description 25
- 238000001914 filtration Methods 0.000 claims description 11
- 230000009466 transformation Effects 0.000 claims description 8
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 7
- 230000000694 effects Effects 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 150000003257 radon Chemical class 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 10
- 238000012937 correction Methods 0.000 description 10
- 238000004088 simulation Methods 0.000 description 9
- 230000005855 radiation Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 238000003491 array Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000000903 blocking effect Effects 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013524 data verification Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 239000013535 sea water Substances 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Radar Systems Or Details Thereof (AREA)
- Image Processing (AREA)
Abstract
The invention provides an automatic detection method for an internal wave region of a satellite SAR image, namely a method for automatically detecting an internal wave existing region of ocean aiming at an SAR image of a large-scale ocean scene under the condition of no manual intervention. The method is oriented to SAR satellite on-satellite processing, the ocean internal wave existing region is rapidly identified, false alarms such as ocean front and sea waves are eliminated, the detection rate of the internal wave region is higher than 75%, strong internal wave existing region information can be rapidly provided for the ground, the method is a basic algorithm for internal wave region detection of an SAR satellite on-satellite processing system, and meanwhile, reference can be provided for internal wave ground processing. The method reduces false alarms as much as possible on the basis of ensuring the discovery rate performance, can realize on-orbit low false alarms, low missing detection and high-efficiency discovery, and can automatically detect the internal wave region.
Description
Technical Field
The invention belongs to the technical field of satellite picture processing and analysis, and particularly relates to an automatic detection method for an internal wave region of a satellite SAR image.
Background
The ocean internal wave refers to the fluctuation in the ocean and exists in a sea area with a stable seawater density jump layer, the maximum amplitude is generated below the sea surface, the amplitude is several meters or even dozens of meters, and the wavelength is hundreds of meters to dozens of kilometers. The internal waves cause the fluctuation of an isopycnic profile, so that the propagation size and direction of the sound velocity are changed, the sonar detection efficiency is influenced, the submarine concealment is influenced, meanwhile, the existence of the internal waves brings challenges to the submarine operation, and the large-amplitude internal waves can drive the submarine to cause the falling depth or the upward throwing. Therefore, the ocean internal wave phenomenon is an ocean natural phenomenon which has a large influence on the submarine, particularly, the large-amplitude internal wave has a great influence on the navigation safety of the underwater submarine, and the important guiding significance for timely mastering whether the internal wave exists in a task sea area or not on submarine command is achieved. The internal wave of the ocean can be well observed by utilizing the satellite-borne Synthetic Aperture Radar (SAR), but because the SAR image has large data volume, low transmission speed and long processing delay time for waiting the ground, the internal wave area in the SAR image needs to be identified on the satellite, and the geographic information is abstracted and transmitted to the ground in time, so that the later-period rapid application and response are facilitated.
Most of the previous internal wave detection methods are used for processing a certain image, for example, a certain internal wave section line in an SAR image is manually selected, the wavelength of the internal wave section line is analyzed, and the internal wave amplitude is inverted by combining a sea area density profile.
Disclosure of Invention
The invention aims to provide an automatic detection method for an internal wave region of a satellite SAR image, namely a method for automatically detecting an internal wave existing region of ocean aiming at an SAR image of a large-range ocean scene under the condition of no manual intervention. The method is oriented to SAR satellite on-satellite processing, the ocean internal wave existing region is rapidly identified, false alarms such as ocean front and sea waves are eliminated, the detection rate of the internal wave region is higher than 75%, strong internal wave existing region information can be rapidly provided for the ground, the method is a basic algorithm for internal wave region detection of an SAR satellite on-satellite processing system, and meanwhile, reference can be provided for internal wave ground processing.
The invention provides an automatic detection method for an internal wave region of a satellite SAR image, which comprises the following steps:
s1, removing land areas and areas with the water depth less than 30m of the SAR images of the satellites to be detected, and further processing the SAR images of the ocean areas with the water depth more than 30 m;
s2, smoothing the SAR image processed in the step S1 by utilizing a Gaussian window to estimate the gray scale large-scale brightness change of the SAR image caused by the change of the incidence angle, and dividing the input SAR image by the smoothed image to correct the fluctuation of the SAR image caused by the change of the incidence angle of different distance units;
as a specific description of the embodiment, wherein the size of the gaussian window is set to 5 km;
s3: carrying out filtering processing on the SAR image processed in the step S2 by adopting isotropic low-frequency filtering on smaller-scale random sea waves and SAR image noise;
wherein the cut-off space wavelength of the low-frequency filtering is 200 m;
s4: dividing the SAR image processed in the step S3 into a plurality of sub-blocks, wherein the sub-blocks are overlapped; taking the sub-blocks starting from the edge, forming a to-be-processed image block set { D1 }k},{D1kThe image processing method comprises the steps that in addition to image data information of each block, block position information and image resolution information of the block are also included;
s6, sequentially processing the image blocks of the image block set to be processed, calculating the SAR image power spectrum of the blocks by utilizing Fourier transform, and calculating the wave number-power spectrum; normalized energy in the wavelength range of the marine internal wave is used as marine internal spectrum characteristics;
as a specific description of the embodiment, the internal wave wavelength is defined as 600-2500 m, the corresponding wave number is 0.0025-0.0105, and the normalized energy in the wave number range of 0.0025-0.0105 is integrated as the marine internal spectrum feature.
S7, according to the extracted marine internal wave spectrum characteristics, carrying out coarse detection on the region possibly containing internal wave image subblocks to obtain a coarse selection internal wave image subblock set { D3 }k}；
S8 for image subblock set { D3kCoarse detection is carried out on the screened SAR image subblock areas to extract local characteristic lines, improved Radon transformation is utilized to detect linear characteristics in a window, and the direction and the position are determined;
the improved Radon transform comprises the following steps:
s81: the image block is subjected to a Radon transform,
s82: in order to remove diagonal effect, calculating the average value of the original image, forming a uniform image with the same size as the original image and the mean value of gray value, and performing Radon transformation on the image;
s83: performing point division on the Radon coefficient value obtained in the step S81 by using the Radon domain coefficient in the step S82 to obtain a normalized Radon coefficient value;
s84: setting a threshold T, finding out the position of a peak point which is greater than the threshold in a Radon domain, recording a linear equation of the peak point, and storing the inclination angle and intercept information of a straight line into a two-dimensional array;
s85: according to the linear equation of step S84, a straight line is drawn in the original image, and a threshold T2 is set, and truncation is performed when the gray scale at both ends of the straight line is less than T2.
S9, according to the local characteristic line two-dimensional array obtained by calculation of S8, grouping the local characteristic line two-dimensional array, then calculating the angle and the mutual distance of the local characteristic line sets in a grouping set, only reserving the local characteristic line sets with the inclination deviation less than 20 degrees and the distance between the intercepts ranging from 500 meters to 2500 meters, removing other characteristic line sets, and calculating to obtain the number of the final reserved characteristic line sets, if the number is more than 2, judging that the image block has internal waves, otherwise, judging that no internal waves exist;
s10: image block set { D3kS8-S9 are sequentially processed, and then the image block set { D3 is outputkAccording to the image block set { D3 }, and according to the image block setkAnd (4) dividing the image blocks into blocks, and outputting the positions of the areas containing the internal waves.
Preferably, a step S5 of removing a block containing a large area of invalid values is added between the steps S4 and S6, and in practical applications, the processed SAR image is often a geographically corrected image with a large number of invalid values at the edges of the image. When the area containing invalid values occupies 40% or more of the total block, the accuracy of the processing is affected, and thus these areas need to be removed. Counting the number of invalid values in the block and the total number of pixels in the block, and when the total number of invalid values is more than 40%, counting from { D1kRemoving the blocks to obtain a to-be-processed image block set { D2 }k}。
The method of the invention is divided into three key stages of pretreatment, feature descriptor and feature extraction and feature identification. In the preprocessing stage, firstly, a certain-size Gaussian window is utilized to smooth the SAR image so as to estimate the gray scale large-scale brightness change of the SAR image caused by the change of an incident angle, secondly, the noise characteristics of random sea waves and the SAR image are combined, the low-frequency filtering of isotropy is adopted to inhibit the random sea waves and the SAR image noise, and meanwhile, the texture characteristics of ocean internal waves are kept to the maximum extent. In the stage of feature descriptor and feature extraction, a dense window is adopted to scan the whole image, the local features of the ocean internal wave lower layer are extracted, and the local high-level features are further extracted. In the characteristic identification stage, the characteristics extracted in the last stage are distinguished from complex marine phenomena such as marine internal waves, sea waves, mesoscale vortexes, ocean fronts, shallow sea terrains, target trails, rainfall phenomena and the like by utilizing the slope and the mutual distance of characteristic lines in Radon domains, and the marine internal waves are extracted. Through comprehensive application of targeted strategies at each stage, the method reduces false alarms as much as possible on the basis of ensuring the discovery rate performance, and can realize on-track low false alarms, low missing detection, efficient discovery and automatic detection of the internal wave region.
Drawings
FIG. 1: an automatic detection processing flow chart of ocean internal waves;
FIG. 2: removing typical SAR image scenes including land and inner wave regions from land and shallow sea;
FIG. 3: a Radarsat 22010-year 7-month-sea south island east SAR image and a distance-direction non-uniform radiation correction effect graph;
FIG. 4: a Radarsat 22010-year 7-month-sea south island eastern SAR image and a noise and wave suppression effect graph;
FIG. 5: dividing a to-be-detected block of the SAR image into a map;
FIG. 6: an internal wave and background region power spectrum characteristic diagram;
FIG. 7: extracting a local characteristic line of an image subblock region;
FIG. 8: marine internal wave SAR image contrast signal map;
FIG. 9: a detection result graph of inner wave regions of SAR images of the east part of the south island of 7-month sea in Radarsat 22010 years;
FIG. 10: an ocean internal wave SAR simulation image detection result graph;
FIG. 11: an automatic detection processing result diagram of marine internal wave measured data, wherein (a) an ENVISAT data internal wave detection result diagram, and (b) a high-grade No. 3 data internal wave detection result diagram.
Detailed Description
The invention provides an automatic detection processing method of marine internal waves under the condition of no manual intervention, aiming at the application requirement of automatically detecting the existence area of the marine internal waves in an SAR image of a large-scale marine scene. A flow chart of the method of the invention is shown in figure 1. In fig. 1, input data is a multi-view SAR power image (with geometric correction information) which is projected by ground distance, corrected by radiation and uniform by a grid, and output is an area ocean internal wave thumbnail image with a geographic position.
As shown in fig. 1, the input data is a multi-view SAR power image (with geometric correction information) with ground distance projection, radiation correction and grid uniformity, and the output is a regional marine internal wave thumbnail image with a geographic position.
The present invention will be described in detail below with reference to examples and the accompanying drawings.
Example 1: steps of the method of the present invention
S1: removing land and shallow sea areas
By utilizing ETOPO topographic elevation Data issued by the U.S. National Geophysical Data Center (NGDC), land areas with the elevation larger than 0 and areas with the water depth smaller than 30m are removed from SAR images, the values of the land areas and the areas are set to NaN (non-significant number), only ocean area SAR images with the water depth larger than 30m are processed subsequently, and the effects of land removal and shallow water removal (smaller than 30m) on the SAR images with the land areas are shown in figure 2.
S2: range-based non-uniform radiation correction
The variation of the scattering intensity due to the different incidence angle of each range cell along the range direction observed by the SAR is corrected. Firstly, smoothing the SAR image by using a Gaussian window with a certain size to estimate the large-scale brightness change of the SAR image gray caused by the change of the incidence angle, wherein the size of the Gaussian window is preliminarily set to be 5 kilometers. The input SAR image is divided by the smoothed image to correct the SAR image fluctuation caused by the incident angle variation of different distance units, and a distance-to-non-uniform radiation correction effect graph is shown in fig. 3 by taking the SAR image of the east of the south island of the 7-month sea in rrarsatsat 22010 as an example.
The steps of using the Gaussian window function to carry out the non-uniform radiation correction are as follows:
s21: averaging the images along the distance direction to obtain a distance direction average value vector M1;
s22: calculating the distance-wise gaussian window size (400 pixels in the distance-wise gaussian window in the example of fig. 3) according to the image resolution (12.5 meters resolution of the SAR image of fig. 3);
s23: performing Gaussian smoothing on the mean vector M1, and estimating the change of the distance to the light and shade;
and S24, correcting the image along the distance direction by using the result of the step S23, and dividing the original pixel by S23 to obtain a correction value.
S25: the original image mean MM is calculated, and the image correction value obtained in S24 is multiplied by MM to obtain a distance-direction non-uniform radiation corrected image.
S3: low-pass filtering based sea wave and noise suppression
And (3) suppressing the noise of the small-scale random sea waves and the SAR image, filtering the noise of the small-scale random sea waves and the SAR image by adopting a Gaussian low-pass filter, and taking a filtering result of 200m from the cut-off space wavelength of the Gaussian low-pass filter as shown in figure 4.
S4: SAR image segmentation
Dividing the SAR image to be processed into a plurality of sub-blocks according to the characteristic scale of the internal wave, wherein the sub-blocks are not overlapped, when the image edge can not be 20 kilometers, in order to prevent missing detection, the sub-blocks from the edge are taken, at the moment, the sub-blocks in the middle of the image are overlapped to a certain extent, the size of the sub-blocks is preliminarily set to be 20 kilometers, the sub-blocks are not overlapped, the result of image blocking of the SAR image shown in figure 4 is shown in figure 5, the azimuth width is 80 kilometers in figure 4, the distance width is 86 kilometers in the azimuth direction, therefore, the number of the azimuth blocks is 4, the distance direction is 5, the distance direction is overlapped to a certain extent with the middle blocks in the image edge area, and the blocked blocks form a block set to be processedkD1, block set { D1kBesides the image data information of each block, the image data information also includes the block position information and the image resolution information of the block.
S5: and removing the blocks containing large-area invalid values.
In practical application, the processed SAR image is often an image after geographic correction, and a large number of invalid values exist at the edge of the image. When the area containing invalid values occupies 40% or more of the total block, the accuracy of the processing is affected, and thus these areas need to be removed. Counting the number of invalid values in the block and the total number of pixels in the block, and when the total number of invalid values is more than 40%, counting from { D1kRemoving the blocks to obtain a to-be-processed image block set { D2 }k}。
S6: image block SAR image spectrum estimation
For { D2kProcessing the image blocks one by one. Calculating the SAR image power spectrum by utilizing Fourier transform, and calculating a wave number-power spectrum according to the SAR image power spectrum; the non-uniform radiation correction by using the Gaussian window function comprises the following steps:
s61: two-dimensional fast Fourier transform is carried out on the image to obtain a frequency spectrum F (u, v),
s62: translating the frequency spectrum, and moving the origin of the frequency domain to the center of the image to obtain
S63: calculating its power spectrum
S64: moving the origin of coordinates to the center of the image, calculating the wave number k of the sector of each point of the half part image on the two-dimensional power spectrum, adding the value to P (t, k) to obtain a wave number-power spectrum P,
s65: low-frequency blocking, namely, the central region of the two-dimensional power spectrum, namely, the corresponding position of the initial 3 points of the wave number-power spectrum P is a low-frequency energy long region, the low frequency is set to zero, thereby reducing the influence of image energy and image brightness,
s66: normalizing the obtained wave number-power spectrum P, i.e. obtaining
Where u and v are frequency components, x and y are spatial image variables, and u is 0,1,2, M-1, v is 0,1,2, N-1M, N are the height and width of the image, respectively.
The wavelength of the ocean internal waves is generally hundreds of meters to kilometers, and the normalized energy in the corresponding wave number range is estimated as the ocean internal wave spectrum characteristic according to the wavelength range of the ocean internal waves.
S7: ocean internal wave coarse detection based on spectral features
The difference analysis between the ocean background power spectrum and the internal wave power spectrum is as follows:
the wave number-power spectrum of the ocean background area and the wave number-power spectrum of the block containing the internal wave have obvious difference (figure 6), and in the normalized power spectrum, the normalized energy with the wave number of 0.0025-0.0105, namely the corresponding wavelength of 600-2500 meters, is integrated, so that the difference between the ocean internal wave and the ocean background can be roughly detected. Is provided withSetting a threshold value T to perform primary processing on the image block set M2, and obtaining a first coarse selection inner wave image sub-block set { D3 } when the wave number is 0.0025-0.0105 and the integral value is greater than Tk}；
In this embodiment, the threshold value is selected to be 0.52 by performing statistics on normalized power spectrums of the ocean internal wave region and the ocean background region.
S8: local feature line extraction
Sequentially comparing the screened SAR image subblock region sets { D3kExtracting local characteristic lines, detecting linear characteristics in a window by using improved Radon transformation, and estimating direction and position (figure 7).
The steps of improving Radon transformation to detect straight line characteristics and estimate direction are as follows:
s81: and carrying out Radon transformation on the image block.
S82: in order to remove the diagonal effect, the average value of the original image is calculated to form a uniform image with the same size as the original image and the gray value as the average value, and the image is also subjected to Radon transformation.
S83: and performing point division on the Radon coefficient value obtained in the step S81 by using the Radon domain coefficient in the step S82 to obtain a normalized Radon coefficient value.
S84: if a straight line exists in the Radon transform domain, the projection integral value in the normal direction of the straight line is larger, a threshold value T is set, the position of a peak point which is larger than the threshold value in the Radon domain is found, a linear equation of the peak point is recorded, and the inclination angle and intercept information of the straight line are stored in a two-dimensional array.
S85: according to the linear equation of step S84, a straight line is drawn in the original image, and a threshold T2 is set, and truncation is performed when the gray scale at both ends of the straight line is less than T2.
S9: false alarm removal processing
Grouping the local characteristic line two-dimensional arrays obtained by calculation according to S8, calculating the angles and mutual distances of the local characteristic line sets in a grouping set, only reserving the local characteristic line sets with the inclination deviation smaller than 20 degrees and the distance between intercepts ranging from 500 meters to 2500 meters, removing other characteristic line sets, calculating to obtain the number of the final reserved characteristic line sets, judging that an image block has internal waves if the number is larger than 2, and judging that no internal waves exist if the number is not larger than 2;
through the processing of S7, ocean internal wave blocks are roughly detected by ocean internal wave wavelengths, but the interference of other ocean phenomenon textures or ocean waves in the blocks cannot be excluded. S8 further extracts the straight features from the block selected in S7, and the false alarm needs to be further removed by these features. The ocean internal wave is generally represented by more than 2 groups of nearly parallel textures with the wavelength of 500-2000 m and close to parallel in the image. According to this feature, the false alarm is further removed.
The false alarm removing steps are as follows:
s91: grouping the two-dimensional arrays of the local characteristic lines calculated according to the step S8, wherein the two-dimensional arrays of the local characteristic lines are grouped into a group, the inclination deviation is less than 10 degrees, the intercept distance difference is less than 100 meters, the group is regarded as a straight line, and a local characteristic line grouping set { R is obtainediI represents the final grouping number and also represents how many lines are in the image block.
S92: computing local feature line grouping set { RiLocal characteristic line set with inclination deviation less than 20 degrees and distance between intercepts of 500-2500 m, local characteristic line grouping set { R }iIn the method, the texture of the natural phenomenon of the ocean is generally formed by too large inclination deviation, the inclination deviation is small (the characteristic lines are basically parallel), but the distance difference is less than 500 meters and is often caused by sea waves, so that the local characteristic line grouping set { RiGet rid of other natural phenomena and wave false alarms, and organize the set of characteristic lines in part { R }iOnly partial characteristic line sets with inclination deviation smaller than 20 degrees and distance between intercepts of 500-2500 m are reserved, other characteristic line sets are removed, and { R } is obtainedi′}。
S93: the local feature line set { R } satisfying the tilt deviation and the distance deviation obtained in S92 is calculatedi' } if the number of the groups is greater than 2, the image block is considered to contain the internal wave, otherwise, the block is judged not to contain the internal wave.
S10: image block set { D3kS8-S9 are sequentially processed, and then the image block set { D3 is outputkAccording to the image block set { D3 }, and according to the image block setkDiagram ofAnd (3) partitioning the image blocks, outputting the position of the internal wave-containing region, and showing the detection result of the internal wave region of the SAR image of the east of the south island of the 7-month sea in Radarsat 22010 years in figure 8.
S11: and (5) carrying out prototype verification on the SAR image ocean internal wave automatic detection algorithm.
In order to fully verify the SAR image marine internal wave automatic detection algorithm prototype and demonstrate the algorithm function, the feasibility of the marine internal wave automatic detection algorithm scheme is verified by the algorithm scheme on simulation data, ENVISAT data and high score number 3 data, and the verification result shows that the algorithm scheme meets the index requirement.
1) Simulation data verification
The research content of the internal wave SAR detection simulation is arranged in the background pre-research of the underwater operation environment detection satellite, and the SAR detection simulation is carried out on the ocean internal wave under the support of the project.
The use of on-board processing to detect internal waves requires that the internal wave region with large amplitude (the general larger the internal wave amplitude, the stronger the internal wave signal on the image) can be indicated.
The ocean internal wave appears as a modulated signal with a certain spatial distribution on the SAR image, and the intensity of the internal wave on the SAR image is represented by contrast signal intensity. As shown in fig. 9 below. And (3) taking the gray level section of the SAR image along the propagation direction of the internal wave, wherein the difference (image gray level is expressed by dB) between the maximum value and the minimum value of the internal wave modulation signal is the intensity of the internal wave contrast signal.
The SAR images of the two groups of weak and strong internal waves under different sea conditions, wave bands and incident angles are simulated, and the contrast signal intensity is shown in the table below. The results of the analysis of two groups of marine internal wave microwave contrast signals with different amplitudes under the conditions of L/C wave band, VV polarization, different incidence angles and different sea conditions are given in Table 1.
Table 1: simulation analysis of marine internal wave microwave contrast signal intensity
Fig. 10 is a diagram of a result of detecting a simulated marine internal wave SAR image under typical conditions.
2) Verification of measured data
The verification of the measured data of the ocean internal wave automatic detection algorithm adopts ENVISAT satellite data and high-resolution No. 3 satellite data, and complex ocean phenomena such as ocean waves, ocean bottom topography, targets, trail ocean fronts and the like in a scene are found. The results of the internal wave detection of the measured data are shown in fig. 11.
The results of statistical findings obtained by verifying and analyzing the SAR internal wave simulation data and the measured data and comparing the results with the visual interpretation results are shown in the following table 2.
Table 2: ocean internal wave automatic detection result statistical table
From the table 2, it can be found that the detection rate of the strong internal wave of the simulation data is 85.6%, the detection rate of the strong internal wave of the ENVISAT data is 76.5%, and the detection rate of the strong internal wave of the high score No. 3 data is 77.3%, which are all better than 75%.
As can be seen from the test of the simulation and actual measurement data by S11, the method can quickly identify the existence region of the ocean internal wave and eliminate false alarms such as ocean front, sea waves and the like, the detection rate of the internal wave region is higher than 75%, strong internal wave existence region information can be quickly provided for the ground, and the method is a basic algorithm for detecting the internal wave region of the SAR satellite onboard processing system and can also provide reference for internal wave ground processing.
Claims (7)
1. An automatic detection method for an internal wave region of a satellite SAR image is characterized by comprising the following steps:
s1, removing land areas and areas with the water depth less than 30m of the SAR images of the satellites to be detected, and further processing the SAR images of the ocean areas with the water depth more than 30 m;
s2, smoothing the SAR image processed in the step S1 by using a Gaussian window;
s3: filtering the SAR image processed in the step S2 by adopting low-frequency filtering to carry out filtering processing on smaller-scale random sea waves and SAR image noise;
s4: dividing the SAR image processed in the step S3 into a plurality of sub-blocks, wherein the sub-blocks are overlapped; taking the sub-blocks starting from the edge, forming a to-be-processed image block set { D1 }k}；
S6, sequentially processing the image blocks of the image block set to be processed, calculating the SAR image power spectrum of the blocks by utilizing Fourier transform, and calculating the wave number-power spectrum; normalized energy in the wavelength range of the marine internal wave is used as marine internal spectrum characteristics;
s7, according to the extracted marine internal wave spectrum characteristics, carrying out coarse detection on the region possibly containing internal wave image subblocks to obtain a coarse selection internal wave image subblock set { D3 }k}；
S8 for image subblock set { D3kCoarse detection is carried out on the screened SAR image subblock areas to extract local characteristic lines, improved Radon transformation is utilized to detect linear characteristics in a window, and the direction and the position are determined;
s9, according to the local characteristic line two-dimensional array obtained by calculation of S8, grouping the local characteristic line two-dimensional array, then calculating the angle and the mutual distance of the local characteristic line sets in a grouping set, only reserving the local characteristic line sets with the inclination deviation less than 20 degrees and the distance between the intercepts ranging from 500 meters to 2500 meters, removing other characteristic line sets, and calculating to obtain the number of the final reserved characteristic line sets, if the number is more than 2, judging that the image block has internal waves, otherwise, judging that no internal waves exist;
s10: image block set { D3kS8-S9 are sequentially processed, and then the image block set { D3 is outputkAccording to the image block set { D3 }, and according to the image block setkAnd (4) dividing the image blocks into blocks, and outputting the positions of the areas containing the internal waves.
2. The method of claim 1, wherein the size of the gaussian window in S2 is set to 5 km.
3. The method of claim 1, wherein the cut-off spatial wavelength of the low-frequency filtering in S3 is 200 m.
4. The method of claim 1, wherein the { D1 } in S4kThe image data information of each block is contained, and the image data information also contains the block position information and the image resolution information of the block.
5. The method of claim 1, wherein a step S5 of removing a block having a large area of invalid values is added between the steps S4 and S6.
6. The method as claimed in claim 1, wherein the internal wave wavelength in S6 is defined as 600-2500 m, corresponding to a wave number of 0.0025-0.0105.
7. The method of claim 1, wherein the modified Radon transform of S8, comprises the steps of:
s81: the image block is subjected to a Radon transform,
s82: in order to remove diagonal effect, calculating the average value of the original image, forming a uniform image with the same size as the original image and the mean value of gray value, and performing Radon transformation on the image;
s83: performing point division on the Radon coefficient value obtained in the step S81 by using the Radon domain coefficient in the step S82 to obtain a normalized Radon coefficient value;
s84: setting a threshold T, finding out the position of a peak point which is greater than the threshold in a Radon domain, recording a linear equation of the peak point, and storing the inclination angle and intercept information of a straight line into a two-dimensional array;
s85: according to the linear equation of step S84, a straight line is drawn in the original image, and a threshold T2 is set, and truncation is performed when the gray scale at both ends of the straight line is less than T2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110329744.8A CN112991309A (en) | 2021-03-29 | 2021-03-29 | Automatic detection method for internal wave region of satellite SAR image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110329744.8A CN112991309A (en) | 2021-03-29 | 2021-03-29 | Automatic detection method for internal wave region of satellite SAR image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112991309A true CN112991309A (en) | 2021-06-18 |
Family
ID=76334026
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110329744.8A Withdrawn CN112991309A (en) | 2021-03-29 | 2021-03-29 | Automatic detection method for internal wave region of satellite SAR image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112991309A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113689449A (en) * | 2021-08-25 | 2021-11-23 | 中国人民解放军海军潜艇学院 | Inversion method and system for characteristic parameters of mesoscale vortices |
| CN113689450A (en) * | 2021-08-25 | 2021-11-23 | 中国人民解放军海军潜艇学院 | Ocean front characteristic parameter inversion method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542277A (en) * | 2012-01-06 | 2012-07-04 | 中国人民解放军第二炮兵装备研究院第三研究所 | Method for detecting ship trail of ocean synthetic aperture radar image |
| CN106951854A (en) * | 2017-03-16 | 2017-07-14 | 中国人民解放军国防科学技术大学 | A kind of SAR image internal wave of ocean parameter extracting method |
| CN110147716A (en) * | 2019-04-02 | 2019-08-20 | 北京理工雷科电子信息技术有限公司 | Wave method for detecting area in a kind of SAR image combined based on frequency domain with airspace |
| CN110232389A (en) * | 2019-06-13 | 2019-09-13 | 内蒙古大学 | A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance |
-
2021
- 2021-03-29 CN CN202110329744.8A patent/CN112991309A/en not_active Withdrawn
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102542277A (en) * | 2012-01-06 | 2012-07-04 | 中国人民解放军第二炮兵装备研究院第三研究所 | Method for detecting ship trail of ocean synthetic aperture radar image |
| CN106951854A (en) * | 2017-03-16 | 2017-07-14 | 中国人民解放军国防科学技术大学 | A kind of SAR image internal wave of ocean parameter extracting method |
| CN110147716A (en) * | 2019-04-02 | 2019-08-20 | 北京理工雷科电子信息技术有限公司 | Wave method for detecting area in a kind of SAR image combined based on frequency domain with airspace |
| CN110232389A (en) * | 2019-06-13 | 2019-09-13 | 内蒙古大学 | A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance |
Non-Patent Citations (1)
| Title |
|---|
| 陈捷等: "一种新的SAR图像内波检测与参数估计方法", 《中国图象图形学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113689449A (en) * | 2021-08-25 | 2021-11-23 | 中国人民解放军海军潜艇学院 | Inversion method and system for characteristic parameters of mesoscale vortices |
| CN113689450A (en) * | 2021-08-25 | 2021-11-23 | 中国人民解放军海军潜艇学院 | Ocean front characteristic parameter inversion method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gautam et al. | Assessment of surface water Dynamicsin Bangalore using WRI, NDWI, MNDWI, supervised classification and KT transformation | |
| CN112991309A (en) | Automatic detection method for internal wave region of satellite SAR image | |
| Kurtz et al. | A sea-ice lead detection algorithm for use with high-resolution airborne visible imagery | |
| Trujillo et al. | Scaling properties and spatial organization of snow depth fields in sub‐alpine forest and alpine tundra | |
| Key | Cloud cover analysis with Arctic advanced very high resolution radiometer data: 2. Classification with spectral and textural measures | |
| CN110147716B (en) | SAR image internal wave region detection method based on combination of frequency domain and airspace | |
| CN105184804A (en) | Sea surface small target detection method based on airborne infrared camera aerially-photographed image | |
| Song et al. | Efficient water area classification using Radarsat-1 SAR imagery in a high relief mountainous environment | |
| CN107678025B (en) | Sea wave height calculation method and device, storage medium and processor | |
| Leng et al. | A novel bathymetry signal photon extraction algorithm for photon-counting LiDAR based on adaptive elliptical neighborhood | |
| KR101221755B1 (en) | Method for identifying reflectivity cells associated with severe weather | |
| CN115422981A (en) | Land and water classification method and system for single-frequency airborne laser sounding data and application | |
| CN112946643B (en) | Offshore wind power extraction method and system based on time sequence radar remote sensing | |
| Hu et al. | A sample enhancement method based on simple linear iterative clustering superpixel segmentation applied to multibeam seabed classification | |
| Maaβ et al. | Improving passive microwave sea ice concentration algorithms for coastal areas: applications to the Baltic Sea | |
| CN104616253A (en) | Light cloud removing method of optical remote sensing image utilizing independent component analysis technology | |
| Zhao et al. | An automatic SAR-based change detection method for generating large-scale flood data records: the UK as a test case | |
| CN113552563B (en) | Method for analyzing correspondence between vertical measurement information and high-frequency ground wave radar clutter information | |
| Sun et al. | A Wave Texture Difference Method for Rainfall Detection Using X‐Band Marine Radar | |
| Grünthal et al. | Monitoring of coastal processes by using airborne laser scanning data | |
| Zahn et al. | Investigation of past and future polar low frequency in the North Atlantic | |
| Wang et al. | A methodological framework for specular return removal from photon-counting LiDAR data | |
| Lian et al. | Denoising algorithm based on local distance weighted statistics for photon counting LiDAR point data | |
| CN111144226A (en) | Sea surface infrared small target detection method based on sea clutter curvature characteristic | |
| Arai | Ground control point generation from simulated sar image derived from digital terrain model and its application to texture feature extraction |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210618 |
|
| WW01 | Invention patent application withdrawn after publication |