CN103971370A - Intelligent ocean oil spill detection method for remote sensing large image - Google Patents
Intelligent ocean oil spill detection method for remote sensing large image Download PDFInfo
- Publication number
- CN103971370A CN103971370A CN201410204690.2A CN201410204690A CN103971370A CN 103971370 A CN103971370 A CN 103971370A CN 201410204690 A CN201410204690 A CN 201410204690A CN 103971370 A CN103971370 A CN 103971370A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- large image
- image
- oil spill
- weibull
- 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.)
- Pending
Links
Landscapes
- Image Processing (AREA)
Abstract
The invention relates to an intelligent ocean oil spill detection method for a remote sensing large image. The method includes the following steps of (1) remote sensing large image input and processing; (2) AOI detection of suspected oil spill areas; (3) oil spill area extraction based on CFAR. The intelligent ocean oil spill detection method for the remote sensing large image has the advantages of greatly improving detection efficiency and detection accuracy of ocean oil spill of the remote sensing large image.
Description
Technical field
The present invention relates to technical field of remote sensing image processing, particularly relate to a kind of marine oil spill intelligent detecting method for the large image of remote sensing.
Background technology
Along with the utilization of continually developing of petroleum resources, following one by one ocean water body oil pollution problem is on the rise, and in various marine pollutions, no matter oil pollution is at occurrence frequency, distribution range, or in the extent of injury, all rank first, people's productive life is caused to serious harm.Therefore, how scientificlly and effectively solving offshore spilled oil pollution becomes when the extremely urgent key subjects of forward swing in face of ours, and how offshore spilled oil being carried out identifying accurately and is fast the prerequisite of solution oil spill.
Pollute owing to utilizing satellite remote sensing technology to monitor in time, accurately, all sidedly marine oil spill, particularly utilize SAR image, because it possesses the round-the-clock monitoring of round-the-clock, become the emphasis data source that oil spilling detects.Countries in the world were carried out the also expansion in succession of research of automatic oil spilling algorithm in recent years to SAR image, as combining, Canada Center for Remote Sensing (CCRS), fishery and ocean portion (DFO), Canadian Coast Guard (CCG), Ministry of National Defence (DFD) and Canadian space office (CSA) successfully develop RADARSAT marine surveillance workstation (Ocean Monitoring Workstation, be called for short OMW), comprise ship detection module, Oil spill detection module, ocean wave spectrum module, oceanic winds module and sea situation analysis module.The SARTool software of BOOST Technologies company of France exploitation, it utilizes the algorithm of rim detection that oil film region is split, and has provided the parameters such as edge gradient simultaneously.MaST (Automated Maritime Surveillance Tool) software is to detect software systems by the SAR image marine features of the DERA subordinate's of national defence research institution of Britain Defense Research Laboratories QinetiQ exploitation.Up to the present, the states such as Norway, Germany, Russia, Britain, France, Japan, Brazil, India and Singapore have carried out the research work that utilizes SAR monitoring marine oil spill in succession, have proposed multiple automatic oil film detection algorithm and have obtained good result.Ocean office of China has also developed oil spill for Bohai Sea detection system, this system mainly adopts the ENVISAT of European Space Agency satellite and the Canadian Radarsat-1 satellite SAR image document of real-time reception, through flow processs such as data pre-service, graphical analysis, oil identification and area measurings, finally form spilled oil monitoring report.And aspect remote sensing oil spilling detection method, Katrine Weisteen Bjerde proposes the method detecting based on the adaptive threshold of Moving Window, first judge in window whether be homogeneous region, bimodal if the histogram in this region presents, this window is carried out to threshold test.Mauro Barni and Lena Chang all propose to utilize half-tone information or statistical property that image is carried out to pre-segmentation, and the method merging again afterwards detects oil spilling.R.T.S.Araujo utilizes region growing to carry out information extraction, obtains certain effect.In at present conventional several method, the oil spilling information that the method that adaptive threshold detects detects is imperfect, if be that oil spilling information cannot detect entirely in window.And method based on cutting apart merging, its difficult point is choosing of judgment criterion, needs to expend many time in merging process simultaneously, is not suitable for the processing to large image.And the quality of region growing depends on choosing of Seed Points, application is to utilize histogrammic peak point as Seed Points more widely at present, but the feature of SAR side-looking imaging more makes image, bright one side is dark on one side, histogram cannot well reflect local features, is difficult to obtain good testing result.
In existing SAR oil spilling image detection, seldom relate to the processing that Hai Lu is cut apart, or only by SAR image is mated with existing shore line information (Global Sea Surface water front general picture), reach the object of removing land information.Shore line changes, all the time not identical, and the variation of morning and evening tides also makes some islands not be marked on map, if with bringing error in the fixing shore line data land of deshielding.
The current existing oil spilling detection algorithm based on SAR data, is all the artificial decipher of some or all of dependence to the identification of oil film, and work efficiency is lower, and false alarm rate is higher.Along with satellite SAR drops into businessization operation, the sharply increase of SAR picture number, China territorial waters is wide simultaneously, has the seas under its jurisdiction of nearly 3,000,000 square kilometres, marine oil spill problem is serious, and traditional artificial decipher process can not meet the demand of practical application far away.Simultaneously, because main detection algorithm major parts more both domestic and external are all to detect for entire image, seldom consider side-looking imaging characteristics and the image size of SAR, therefore detection efficiency is lower, and for large image oil spilling context of detection, because the complicacy of entire image is higher, this has brought difficulty also to the precision detecting.
Summary of the invention
The object of this invention is to provide a kind of marine oil spill intelligent detecting method for the large image of remote sensing, to overcome above shortcomings in currently available technology.
The object of the invention is to be achieved through the following technical solutions:
For a marine oil spill intelligent detecting method for the large image of remote sensing, comprise the following steps:
(1) the large image input of remote sensing and processing:
Obtain the SAR view data of input computing machine and it is carried out to LEE and MAP Gamma filtering processing and geometric correction;
(2) the doubtful district AOI of oil spilling surveys:
The imagery exploitation Ratio edge detection algorithm of finishing dealing with in step (1) is carried out to AOI detection, and described Ratio edge detection algorithm is as follows:
u 1 with
u 2for the not overlapping region along central pixel point rectilinear direction
r 1with
r 2the gray average of interior N point;
(3) the oil spilling extracted region based on CFAR:
After step (2) AOI is extracted, the target detection region threshold of the large image of remote sensing is carried out CFAR oil spilling extracted region and is obtained in several AOI regions that comprise oil film object of gained, and described CFAR algorithm is as follows:
Can obtain threshold value
, wherein
for false-alarm probability,
p(
x) by the clutter Weibull distribution probability Density functional calculations as background taking ocean:
Wherein parameter
bfor scale parameter,
cfor form parameter.
When
c=2 o'clock, Weibull distributed and degenerates into Rayleigh distribution, when
c=1 o'clock, Weibull distributed and degenerates into exponential distribution;
The computing formula of the first order and second order moments that the scale parameter that Weibull distributes and form parameter are distributed by Weibull draws:
Wherein
.
Further, between described step (1) and (2), increase Hai Lu and cut apart, shield land step:
By the calculating to image pixel, to determine in image non-sea area and shield these regions, computing formula is as follows:
wherein,
，
。
Beneficial effect of the present invention is: this method has significantly improved detection efficiency and the accuracy of detection of the large image marine oil spill of remote sensing.
Brief description of the drawings
With reference to the accompanying drawings the present invention is described in further detail below.
Fig. 1 is the marine oil spill intelligent detecting method FB(flow block) for the large image of remote sensing described in the embodiment of the present invention;
Fig. 2 is the oil spilling panorama sketch in the region, the Envisat Bohai Sea described in the embodiment of the present invention;
Fig. 3 is that Fig. 2 uses overall CFAR to detect the oil spilling result figure of gained;
Fig. 4 is that Fig. 2 uses the inventive method to detect the oil spilling result figure of gained.
Embodiment
As shown in Figure 1, the SAR image of input is carried out to geometric correction and noise filtering, then carry out Hai Lu and cut apart, the impact of shielding land; Then for whole scape image, utilize ratio rim detection (ROA) to monitor out the approximate location of oil film, and be labeled as the doubtful district of oil spilling (AOI), then carry out adaptive partition detection for the CFAR detection algorithm after these AOI application enhancements, detect final oil spilling region, and extract relevant information.The method can better adapt to SAR image Sea background complexity and the strong situation of locality, obtains high-precision testing result.Specifically comprise the following steps:
Step 1: carry out the accurate processing of whole scape image: comprise SAR image LEE and MAP Gamma filtering, geometric correction etc.;
Step 2: Hai Lu is cut apart, shielding land.Application Markov Random Field Theory is carried out Hai Lu and is cut apart, and surveying may shore line and isolated island region, and shields these regions.Computing formula is as follows:
Known according to SAR marine oil spill characteristics of image, land is generally high than the gray scale of seawater and oil film, therefore can, by an initial threshold is set, carry out initial segmentation to SAR image.Now
value have two kinds, correspond respectively to land and marine site (comprising seawater and oil film), can be expressed as by the form of mathematics:
Conditional probability density function
provide the strength information of image pixel, structural information
can, by Markov random field (MRF) model modeling, suppose
in each pixel independent same distribution, have:
Supposing the intensity Gaussian distributed of pixel, is following formula:
The structural information of image
can, by providing with the Gibbs distribution energy function of Markov random field model equivalence, consider the speckle noise of SAR image, structural information can be expressed as:
Wherein:
.
Therefore, final classification is estimated to be expressed as:
Step 3: the doubtful district of oil spilling (AOI) surveys.Use Ratio edge detection algorithm, whole image is carried out to AOI detection, algorithm is as follows:
Ratio edge detection algorithm is that a kind of average of calculating adjacent two regions recently determines whether object pixel is the algorithm of marginal point.In slip detection window, getting center pixel is measuring point to be checked, along the direction of crossing this some straight line, calculates not overlapping region, both sides
with
in
the gray average of individual point
with
, ask the ratio of two averages:
Consider the different orientation at edge, retain
be worth minimum result.Can adjust window size according to actual conditions.If window is arranged in the homogeneous area of image,
with
it is close,
be tending towards 1; Otherwise, in the time that window center is positioned at the intersection of zones of different, due to the statistical property difference in two regions,
to be less than 1.
less, declare area difference is larger, and the possibility that window is positioned at edge is higher.
In actual conditions, the contrast of seawater and oil spilling is little, and oil spilling image oils and is not very clear to have certain ambiguity with the border of water.The pixel adjacent with object pixel is made as to mask pixels, do not participate in the calculating of average, the number of mask pixels is depending on actual conditions, and we just can obtain fuzzy edge like this.
Step 4: based on the oil spilling extracted region of improving CFAR.After extracting through the target area of previous step, large-sized image is divided into the AOI region that comprises one by one oil film object, just carries out oil spilling extraction for these AOI regions afterwards.The main oil spilling that adopts improvement CFAR algorithm to carry out this step detects.Algorithm is as follows:
The key of CFAR detection technique is to determine adaptive threshold, supposes
for the probability density function of radar clutter distributed model, order
, it is visible,
be increasing function, pass through solving equation
Can obtain threshold value
, wherein
for false-alarm probability.
Distributed model to clutter background is estimated, is the basis that CFAR detects, and is also to reduce false-alarm, the not key of lose objects.It is generally acknowledged, conventionally obey K taking ocean as the clutter of background and distribute and Weibull distributed model.The probability density function that Wei Buer distributes is expressed as
Wherein parameter b is scale parameter, and c is form parameter.In the time of c=2, Weibull distributes and degenerates into Rayleigh distribution, and in the time of c=1, Weibull distributes and degenerates into exponential distribution.
Utilize above-mentioned Clutter Model, for a local window of each AOI definition, each like this AOI is just divided into protection part and background parts, and backdrop window is added up for background clutter, thereby calculates target detection threshold value.According to the data of backdrop window, the character of utilizing Weibull to distribute, the mutual relationship of the first order and second order moments that can distribute by Weibull is tried to achieve scale parameter and the form parameter that Weibull distributes:
The first order and second order moments that Weibull distributes is respectively:
Definition
A best polynomial curve fitting of above formula is
, wherein parameter
.
Thereby the scale parameter that Weibull distributes
, according to b, the value of c, can obtain threshold value.
In actual applications, as shown in Figure 2, the oil spilling image in region, the Envisat Bohai Sea, size is 4560*4128.From figure, can see that sea is subject to the impact of wind-force or other factors, on image, form fish scale-shaped ripple, and due to the impact of side-looking imaging, the gray scale of entire image presents uneven distribution.
As shown in Figure 3, after overall CFAR detects, in the time that false alarm rate is 0.03, testing result is relatively desirable, but be global threshold due to what adopt, make the partially dark wave of image central authorities' gray scale also be detected as oil spilling, the false-alarm of monitoring out is many, and the linear oil spilling of the part that identifies out with square frame in original image does not have complete being detected on the contrary.
As shown in Figure 4, detect through the large image marine oil spill of remote sensing of the present invention detection method the result drawing, can find out, can effectively detect for the most of oil spilling in image, also effectively suppress the negative effect that wave of the sea brings for detection simultaneously.
By adopting many scapes of zones of different SAR image to carry out detection validation, prove the large image marine oil spill of remote sensing of the present invention detection method, can obtain in actual applications more accurate marine oil spill testing result.
The present invention is not limited to above-mentioned preferred forms; anyone can draw other various forms of methods under enlightenment of the present invention; though but do any variation at it in form, every have identical with a application or akin technical scheme, within all dropping on protection scope of the present invention.
Claims (4)
1. for a marine oil spill intelligent detecting method for the large image of remote sensing, it is characterized in that: comprise the following steps:
(1) the large image input of remote sensing and processing:
Obtain the SAR view data of input computing machine and it is carried out to LEE and MAP Gamma filtering processing and geometric correction;
(2) the doubtful district AOI of oil spilling surveys:
The imagery exploitation Ratio edge detection algorithm of finishing dealing with in step (1) is carried out to AOI detection, and described Ratio edge detection algorithm is as follows:
U
1and u
2for the not overlapping region R along central pixel point rectilinear direction
1and R
2the gray average of interior N point;
(3) the oil spilling extracted region based on change CFAR:
After step (2) AOI is extracted, the target detection region threshold of the large image of remote sensing is carried out CFAR oil spilling extracted region and is obtained in several AOI regions that comprise oil film object of gained, and described CFAR algorithm is as follows:
Can obtain threshold value
, wherein
for false-alarm probability
, P(
x) by the clutter Weibull distribution probability Density functional calculations as background taking ocean:
Wherein parameter
bfor scale parameter,
cfor form parameter.
2. a kind of marine oil spill intelligent detecting method for the large image of remote sensing according to claim 1, is characterized in that: in step (3), when
c=2 o'clock, Weibull distributed and degenerates into Rayleigh distribution, when
c=1 o'clock, Weibull distributed and degenerates into exponential distribution.
3. a kind of marine oil spill intelligent detecting method for the large image of remote sensing according to claim 2, is characterized in that: the computing formula of the first order and second order moments that the scale parameter that Weibull distributes and form parameter are distributed by Weibull draws:
Wherein
.
4. a kind of marine oil spill intelligent detecting method for the large image of remote sensing according to claim 3, is characterized in that: between described step (1) and (2), increase Hai Lu and cut apart, shield land step:
Cross the calculating to image pixel, determine in image non-sea area and shield these regions, computing formula is as follows:
Wherein,
,
。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410204690.2A CN103971370A (en) | 2014-05-15 | 2014-05-15 | Intelligent ocean oil spill detection method for remote sensing large image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410204690.2A CN103971370A (en) | 2014-05-15 | 2014-05-15 | Intelligent ocean oil spill detection method for remote sensing large image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103971370A true CN103971370A (en) | 2014-08-06 |
Family
ID=51240819
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410204690.2A Pending CN103971370A (en) | 2014-05-15 | 2014-05-15 | Intelligent ocean oil spill detection method for remote sensing large image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103971370A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104637063A (en) * | 2015-02-16 | 2015-05-20 | 天津大学 | Method for detecting oil film edge in synthetic aperture radar ocean oil overflow image |
| CN105046706A (en) * | 2015-07-13 | 2015-11-11 | 北京化工大学 | Rational polynomial function fitting sea clutter based SAR image ship detection method |
| CN106022288A (en) * | 2016-05-30 | 2016-10-12 | 电子科技大学 | Marine oil spill information identification and extraction method based on SAR image |
| WO2018000252A1 (en) * | 2016-06-29 | 2018-01-04 | 深圳大学 | Oceanic background modelling and restraining method and system for high-resolution remote sensing oceanic image |
| CN108564054A (en) * | 2018-04-24 | 2018-09-21 | 电子科技大学 | A kind of accurate oil spilling detection method based on CFAR |
| CN108596065A (en) * | 2018-04-13 | 2018-09-28 | 深圳职业技术学院 | One kind is based on deep semantic segmentation marine oil spill detecting system and method |
| CN109829858A (en) * | 2019-01-29 | 2019-05-31 | 大连海事大学 | A kind of shipborne radar image spilled oil monitoring method based on local auto-adaptive threshold value |
| CN111025291A (en) * | 2019-11-06 | 2020-04-17 | 中国石油大学(华东) | Ocean oil spill detection method based on new characteristics of fully-polarized SAR |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103778627A (en) * | 2014-01-02 | 2014-05-07 | 北京理工大学 | Sea oil spill detection method based on SAR image |
-
2014
- 2014-05-15 CN CN201410204690.2A patent/CN103971370A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103778627A (en) * | 2014-01-02 | 2014-05-07 | 北京理工大学 | Sea oil spill detection method based on SAR image |
Non-Patent Citations (5)
| Title |
|---|
| 张翠 等: "基于MRF场的SAR图像分割方法", 《遥感技术与应用》 * |
| 曹永锋 等: "基于MPM准则的无监督SAR图像分割", 《武汉大学学报 信息科学版》 * |
| 邹亚荣 等: "基于SAR后向散射的海上溢油检测研究", 《遥感信息》 * |
| 马腾波 等: "基于边缘分析的海面溢油检测", 《遥感学报》 * |
| 马腾波: "基于SAR图像的海面溢油检测的研究", 《中国科学院电子学研究所》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104637063A (en) * | 2015-02-16 | 2015-05-20 | 天津大学 | Method for detecting oil film edge in synthetic aperture radar ocean oil overflow image |
| CN105046706B (en) * | 2015-07-13 | 2019-01-29 | 北京化工大学 | SAR image ship detection method based on rational polynominal Function Fitting sea clutter |
| CN105046706A (en) * | 2015-07-13 | 2015-11-11 | 北京化工大学 | Rational polynomial function fitting sea clutter based SAR image ship detection method |
| CN106022288A (en) * | 2016-05-30 | 2016-10-12 | 电子科技大学 | Marine oil spill information identification and extraction method based on SAR image |
| CN106022288B (en) * | 2016-05-30 | 2019-07-12 | 电子科技大学 | The identification of marine oil spill information and extracting method based on SAR image |
| WO2018000252A1 (en) * | 2016-06-29 | 2018-01-04 | 深圳大学 | Oceanic background modelling and restraining method and system for high-resolution remote sensing oceanic image |
| US10497128B2 (en) * | 2016-06-29 | 2019-12-03 | Shenzhen University | Method and system for sea background modeling and suppression on high-resolution remote sensing sea images |
| CN108596065A (en) * | 2018-04-13 | 2018-09-28 | 深圳职业技术学院 | One kind is based on deep semantic segmentation marine oil spill detecting system and method |
| CN108564054A (en) * | 2018-04-24 | 2018-09-21 | 电子科技大学 | A kind of accurate oil spilling detection method based on CFAR |
| CN108564054B (en) * | 2018-04-24 | 2020-11-10 | 电子科技大学 | Accurate oil spill detection method based on CFAR |
| CN109829858A (en) * | 2019-01-29 | 2019-05-31 | 大连海事大学 | A kind of shipborne radar image spilled oil monitoring method based on local auto-adaptive threshold value |
| CN109829858B (en) * | 2019-01-29 | 2022-10-18 | 大连海事大学 | Ship-borne radar image oil spill monitoring method based on local adaptive threshold |
| CN111025291A (en) * | 2019-11-06 | 2020-04-17 | 中国石油大学(华东) | Ocean oil spill detection method based on new characteristics of fully-polarized SAR |
| CN111025291B (en) * | 2019-11-06 | 2021-07-16 | 中国石油大学(华东) | Ocean oil spill detection method based on new characteristics of fully-polarized SAR |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103971370A (en) | Intelligent ocean oil spill detection method for remote sensing large image | |
| Chen et al. | Ship detection from coastal surveillance videos via an ensemble Canny-Gaussian-morphology framework | |
| CN106022288B (en) | The identification of marine oil spill information and extracting method based on SAR image | |
| CN109190538A (en) | A kind of fully-loaded stream river deposition coast band evolution analysis method based on remote sensing technology | |
| CN102096070B (en) | Sea surface oil spilling detection method of complete polarization synthetic aperture radar | |
| CN102609701B (en) | Remote sensing detection method based on optimal scale for high-resolution SAR (synthetic aperture radar) | |
| Hu et al. | Monitoring coastline variations in the Pearl River Estuary from 1978 to 2018 by integrating Canny edge detection and Otsu methods using long time series Landsat dataset | |
| CN104077777A (en) | Sea surface vessel target detection method | |
| CN102842130B (en) | Method for detecting buildings and extracting number information from synthetic aperture radar image | |
| CN103236063A (en) | Multi-scale spectral clustering and decision fusion-based oil spillage detection method for synthetic aperture radar (SAR) images | |
| CN103984945A (en) | Optical remote sensing image ship detection method | |
| CN112487912B (en) | Arbitrary direction ship detection method based on improved YOLOv3 | |
| CN103778627A (en) | Sea oil spill detection method based on SAR image | |
| Bouchahma et al. | Automatic measurement of shoreline change on Djerba Island of Tunisia | |
| Hong et al. | Automatic sub-pixel coastline extraction based on spectral mixture analysis using EO-1 Hyperion data | |
| Ding et al. | Coastline detection in SAR images using multiscale normalized cut segmentation | |
| Xu-kai et al. | Automated detection of coastline using Landsat TM based on water index and edge detection methods | |
| CN109034083A (en) | A kind of natural Extracting costline method and apparatus based on more scape remote sensing images | |
| CN106872976A (en) | A kind of marine oil spill information extracting system based on radar data | |
| CN102034233B (en) | Method for detecting SAR (stop and reveres) image wave group parameters based on contourlet conversion | |
| Kumar et al. | A deep learning framework to map riverbed sand mining budgets in large tropical deltas | |
| Wang et al. | An automatic detection of green tide using multi-windows with their adaptive threshold from Landsat TM/ETM plus image | |
| CN106291555A (en) | A kind of offshore spilled oil detection method based on C-band polarization SAR image | |
| Xu et al. | Extraction and spatiotemporal evolution analysis of tidal flats in the Bohai Rim during 1984–2019 based on remote sensing | |
| Yu et al. | Automatic extraction of green tide using dual polarization Chinese GF-3 SAR images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20140806 |