CN105913430B - Yellow River Main based on multi-spectral remote sensing image slips information synergism extracting method - Google Patents
Yellow River Main based on multi-spectral remote sensing image slips information synergism extracting method Download PDFInfo
- Publication number
- CN105913430B CN105913430B CN201610224322.3A CN201610224322A CN105913430B CN 105913430 B CN105913430 B CN 105913430B CN 201610224322 A CN201610224322 A CN 201610224322A CN 105913430 B CN105913430 B CN 105913430B
- Authority
- CN
- China
- Prior art keywords
- master
- river
- slips
- line
- degree
- 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
Classifications
-
- 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
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a kind of Yellow River Mains based on multi-spectral remote sensing image to slip information synergism extracting method, and the technical problem of information extracting method accuracy rate difference is slipped for solving existing Yellow River Main.Technical solution is that the method first by degree of bias analysis method and based on spectral similarity and spatial continuity obtains one group of master respectively and slips a position；Then the master that degree of bias analysis method obtains is slipped and is input in the extracting method based on spectral similarity and spatial continuity as prior information, slipped a little for obtaining the master after correction；Hereafter, master obtained is slipped to the input for being a little re-used as degree of bias analysis method, further correction master slips a little；Such iteration is stablized until the master obtained slips a position；It is a little attached finally, obtained master is slipped, forms master and slip line.The method of the present invention on the Landsat TM remote sensing images of the Yellow River middle and lower reaches the experimental results showed that, when verification and measurement ratio reaches 95%, false alarm rate is only 1%, and the master obtained slips line with good continuity.
Description
Technical field
The present invention relates to a kind of Yellow River Mains to slip information extracting method, more particularly to a kind of based on multi-spectral remote sensing image
Yellow River Main slips information synergism extracting method.
Background technique
Remote sensing images have been widely used for identifying water boy, river extraction, water quality detection, big flood detection, water area changes inspection
The fields such as survey, methods that these researchs use are substantially the characteristic shown on the image according to water body and use traditional
Some classification methods are detected.The extraction that Yellow River Main slips information is all traditionally by manually drawing, and this method is not only taken
When, it is laborious, and be easy to be influenced by natural conditions such as weather, it is often more important that, it is difficult to grasps to lead in time in flood season and slips variation
Situation.The continuous development of remote sensing technology and remote sensing image processing becomes so that carrying out Yellow River Main with remote sensing images and slipping line detection
It may.River is main to slip the problem of showing in the changeable siltation sand bed matter river of river regime protrusion, due in the world to heavily silt-carrying river
River river regime research is less.Document " Main-stream Feature Extraction of the Yellow River
Based on Regional Spectral Un-mixing,4th International Yellow River Forum,
2009, pp212-219 " are proposed the master mixed in one based on local spectrum solution and slip feature extracting method, using spectral vector in height
Geometric attribute in dimension space carries out solution and mixes, and seeks Yellow River Main and slips the end member in region, so that extracting typical master slips information.So
And this method fails to make full use of the spatial information of remote sensing images, and single method is difficult to obtain the robust property of many aspects,
It thus is difficult to extract the master that continuity is good, accuracy rate is high and slips information.
Summary of the invention
In order to overcome the shortcomings of that it is poor that existing Yellow River Main slips information extracting method accuracy rate, the present invention provides a kind of based on mostly light
The Yellow River Main of spectrum remote-sensing image slips information synergism extracting method.This method by degree of bias analysis method and is based on spectral similarity first
One group of master is obtained respectively with the method for spatial continuity slips a position；Then the master that degree of bias analysis method obtains is slipped into conduct first
Information input is tested into the extracting method based on spectral similarity and spatial continuity, is slipped a little for obtaining the master after correcting；This
Afterwards, master obtained is slipped to the input for being a little re-used as degree of bias analysis method, further correction master slips a little；Such iteration, until obtaining
The master obtained slips a position and stablizes；It is a little attached finally, obtained master is slipped, forms master and slip line.Method disclosed by the invention exists
It is on the Landsat TM remote sensing images of the Yellow River middle and lower reaches the experimental results showed that, when verification and measurement ratio reaches 95%, false alarm rate is only
1%, and the master obtained slips line with good continuity.
The technical solution adopted by the present invention to solve the technical problems is: a kind of Yellow River Main based on multi-spectral remote sensing image
Information synergism extracting method is slipped, its main feature is that the following steps are included:
(1) river is divided.
A width Landsat TM multispectral image is inputted, first progress river coarse segmentation.Using spectral classification and match party
Method carries out spectrum picture classification, and carries out post-classification comparison according to the features of shape of the Yellow River section, right respectively using two kinds of samples
Image is classified, then composograph.Due to there is the influence of bridge and beach in image.
(2) riverbank line drawing.
Edge is detected using Canny operator, and tracking connection is carried out to the edge detected, sorts out river south
The flowage line of northern bank.Tracking is attached to obtained edge line；According to neighborhood method, obtained using a border following algorithm
One group of continuous line segment removes some shorter interference line segments according to length, while according to the style characteristic of beach, removing beach
It applies, obtains one group of useful line segment.According to the distribution characteristics of Yellow River, the judgement of north and south bank is carried out to every section of line segment.Based on calculation
Method efficiency considers that the judgement that only need to carry out north and south bank to representational points certain in every section obtains belonging to this whole section of flowage line
The information of which bank.According to statistical property, the number that the point judged in every section belongs to a certain bank is greater than a certain threshold value and judges
This section is to belong to southern bank or northern bank, conversely, then belonging to an other bank, to obtain north and south bank image.As needed north and south two
The line segment of bank carries out orderly storing into matrix respectively, to obtain the complete flowage line in two sides, carries out master and slips line drawing.
(3) river is segmented.
The Yellow River is divided into Typical River and atypia section.Typical River be divided into it is straight it is micro-bend, be bent and divide branch of a river three classes.Benefit
With the bending degree and direction of bending coefficient and curvature representation section, determined whether by the river distribution between the flowage line of north and south
In the presence of dividing the branch of a river.It is defined as that Curved Continuous connects section and bend comes over and pledges allegiance to section to spatial relationship, using continuous between section
Variation arrangement is distinguished from space.Specific segmentation method is as follows:
A) the dam bank in space is transformed to by curvature domain, the maximum point of gained curvature sequence by the curvature estimation to open a window
It sets, where the position for representing the curved top of bend.
B) the minimum point position between two continuous threshold points represents the transition and linkage point position between two continuous river bends.
(4) extraction master is analyzed using the degree of bias slip initial position.
Extraction master is analyzed using the degree of bias and slips information, and initially master slips point Y for one group of acquisition(0).Degree of bias parser specific steps are such as
Under:
Step1 obtains p × n observing matrix data X=[X1,X2,…,Xn], wherein each column XiRepresent an observation sample
Vector, every a line represent an observation attribute；Two class sample data S1,S2；
Step2 seeks the average deviation form B of X: enablingThenThe center that reference axis is moved to former data, seeks S1,S2Sample average M1, M2；
Step3 seeks between class scatter matrix Gb, between class scatter matrix GbIt is the positive semidefinite matrix of p × p, is defined as Gb=(M1-
M2)(M1-M2)T；
Step4 seeks between class scatter matrix GbEach characteristic value evaliWith feature vector eigi, i ∈ [1, n]；
Step5 select from big to small by characteristic value and corresponding feature vector, composition transformation matrix
T=(eig1,eig2,…eigm), m≤n；
Step6 generates the data set H:H=T in new coordinate systemT·X.Degree of bias analysis is carried out to transformed first component:
Wherein, SD is standard deviation.Skewness=0 illustrates that distributional pattern is identical as the normal state degree of bias；Skewness > 0, just
Partially, peak value is on a left side；Skewness < 0 is negative bias, and peak value is on the right side.| Skewness | bigger, distributional pattern degrees of offset is bigger.?
In the histogram set of graphs of a certain section, take the position of coefficient of skewness maximum point as the position for slipping line main on current section.I.e. with 3
× 3 window calculation coefficients of skewness, and the ordinate of the maximum point of the recording image each column coefficient of skewness, then will be adjacent 3 in record
The ordinate of column is average, slips a position as master.
(5) master is extracted using spectral similarity and spatial continuity method slip initial position.
Extraction master is analyzed using the degree of bias and slips information, and initially master slips point Z for one group of acquisition(0).Spectral similarity and spatial continuity
Specific step is as follows for method:
First by artificial field exploring to master slip into a large amount of study of row, establish master and slip library of spectra.River is chosen to enter
Point most like in library of spectra is slipped with master at mouthful, slips sample as initial master.By calculating its direction line initially master slips a little,
And obtain its length and position, the specific steps are as follows:
Stepl define direction line be across center pel a series of line segments, their length is different, length by
Spectral similarity measurement and threshold value between adjacent picture elements determine.Similarity measure are as follows:
Wherein, x ' is xiAdjacent picture elements,Indicate mahalanobis distance.
Step2 direction line is extended according to specific rule from center pel towards both sides, while the new picture that will be selected into
Member is added result and combines.The condition of direction line extension is that the similarity measure of current pixel is less than threshold value th1.Here, th1For light
Similarity measurements threshold value is composed, it is related to the variation degree of pixel gray scale in the same shape region.Since water flow flowing has centainly
Directionality, the extension of actual direction line can not carry out towards all directions, but according to river direction, will be with river direction
The direction removal deviated from, then be extended.
Step3 traverses whole river, tracks to obtain the direction line of initial pixel according to Stepl and two step of Step2, obtain simultaneously
The pixel collection Z gradually formed when must form direction line(0)。
(6) iteration realizes that the main information synergism that slips extracts.
Master obtained in step (5) is slipped into point set Z(0)It is inputted in degree of bias analysis method as tutorial message, to its inside
Parameter is adjusted, if Z(0)With Y(0)Consistency L (Y(0),Z(0)) it is less than given threshold value th2, then terminate collaborative processes；Otherwise,
If Z(0)In pixel number be less than Y(0)In pixel number, degree of bias analysis window size reduce 1；Conversely, degree of bias analysis window size
Increase 1；By the degree of bias analysis method after adjusting parameter, the master for obtaining its output slips point set Y(1).Similarly, by Y(0)For instructing
The parameter of spectral similarity and spatial continuity method adjusts, if Z(0)With Y(0)Consistency L (Y(0),Z(0)) it is less than given threshold
Value, then terminate collaborative processes；Otherwise, if Z(0)In pixel number be less than Y(0)In pixel number, th1Increase by 2%；Conversely, th1Subtract
It is small by 2%；By the degree of bias analysis method after adjusting parameter, the master for obtaining its output slips point set Z(1)。
Above-mentioned collaborative processes are repeated, until algorithm is because of L (Y(t),Z(t)) it is less than given threshold value th2And terminate, t is iteration generation
Number.
The beneficial effects of the present invention are: this method first by degree of bias analysis method and based on spectral similarity and space it is continuous
The method of property obtains one group of master respectively and slips a position；Then the master that degree of bias analysis method obtains a conduct prior information is slipped to input
Into the extracting method based on spectral similarity and spatial continuity, slipped a little for obtaining the master after correcting；Hereafter, it will be obtained
Master slip the input for being a little re-used as degree of bias analysis method, further correction master slips a little；Such iteration, until the master obtained slips point
Set stabilization；It is a little attached finally, obtained master is slipped, forms master and slip line.Method disclosed by the invention is in the Yellow River middle and lower reaches
It is on Landsat TM remote sensing images the experimental results showed that, when verification and measurement ratio reaches 95%, false alarm rate is only 1%, and obtain
The main line that slips is with good continuity.
It elaborates With reference to embodiment to the present invention.
Specific embodiment
Slipping information synergism extracting method the present invention is based on the Yellow River Main of multi-spectral remote sensing image, specific step is as follows:
(7) river is divided.
A width Landsat TM multispectral image is inputted, first progress river coarse segmentation.Using spectral classification and matching skill
Art, such as: the methods of spectral vector matching, mahalanobis distance segmentation and Gauss Markov carry out spectrum picture classification, and according to Huang
The features of shape of river section carries out post-classification comparison, such as: region merging technique.Supervised classification method is used when classification, due to figure
As being influenced by factors such as weather, the spectrum of the upper reaches of the Yellow River and downstream water has larger difference in image, according to single sample
It carries out Spectral angle mapper classification and has very big error, therefore classified respectively to image using two kinds of samples, then composograph.By
There is the influence of bridge and beach in image, be not easy to the extraction of north and south two sides flowage line, requires to be suitble into one in edge definition
Morphological scale-space is carried out to image in the case where step research, i.e. image carries out expansion and etching operation, eliminates lesser beach
Painting and bridge.
(8) riverbank line drawing.
Edge is detected using Canny operator, and tracking connection is carried out to the edge detected, sorts out river south
The flowage line of northern bank is ready for the segmentation of next step river.It, need to be to obtained edge line for the ease of extracting north and south two sides
It is attached tracking；According to neighborhood method, it can arrive to obtain one group of continuous line segment using a border following algorithm, be gone according to length
Except some shorter interference line segments, while beach can be removed, to obtain one group of useful line segment according to the style characteristic of beach.
According to the distribution characteristics of Yellow River, observe it with certain transverse direction and vertical features, accordingly can algorithm for design to every section of line
The judgement of the north and south Duan Jinhang bank.Consider that only need to carry out north and south bank to representational points certain in every section sentences based on efficiency of algorithm
The disconnected information which bank belonging to this whole section of flowage line can be obtained.According to statistical property, the point judged in every section belongs to a certain
The number of bank, which is greater than a certain threshold value, can determine whether that this section is to belong to southern bank or northern bank, conversely, then belonging to an other bank, to obtain
North and south bank image.The line segment of north and south two sides is carried out orderly storage into matrix respectively in certain sequence as needed, thus
To the complete flowage line in two sides, so that the master of next step slips line drawing.
(9) river is segmented.
Line problem is slipped with regard to Yellow River Main, the Yellow River is divided into Typical River and atypia section.Typical River is divided into straight micro- again
It is curved, be bent and divide branch of a river three classes.The shape of every one kind section has very big difference, describes the major way of difference first is that using bending
Coefficient and curvature describe bending degree and the direction of section, second is that being distributed by the river between the flowage line of north and south to determine is
No presence divides the branch of a river.Space relationship between section and section is also that the Yellow River Main determined slips one of line foundation, therefore, for section
Between spatial relationship calculating be also very important, the present invention to spatial relationship be defined as Curved Continuous connect, it is curved
Equal sections are come over and pledged allegiance in road, are arranged using the consecutive variations between section from space and are distinguished.Specific segmentation method is as follows:
A) the dam bank in space is transformed to by curvature domain, the maximum point of gained curvature sequence by the curvature estimation to open a window
It sets, where the position for representing the curved top of bend.
B) the minimum point position between two continuous threshold points represents the transition and linkage point position between two continuous river bends.
(10) extraction master is analyzed using the degree of bias slip initial position.
Extraction master is analyzed using the degree of bias and slips information, and initially master slips point Y for one group of acquisition(0).Degree of bias parser specific steps are such as
Under:
Step1 obtains p × n observing matrix data X=[X1,X2,…,Xn], wherein each column XiRepresent an observation sample
Vector, every a line represent an observation attribute；Two class sample data S1,S2；
Step2 seeks the average deviation form B of X: enablingThenThe center that reference axis is moved to former data, seeks S1,S2Sample average M1, M2；
Step3 seeks between class scatter matrix Gb, it is the positive semidefinite matrix of p × p, is defined as Gb=(M1-M2)(M1-M2)T；
Step4 seeks between class scatter matrix GbEach characteristic value evaliWith feature vector eigi, i ∈ [1, n]；
Step5 select from big to small by characteristic value and corresponding feature vector, composition transformation matrix
T=(eig1,eig2,…eigm), m≤n；
Step6 generates the data set H:H=T in new coordinate systemT·X.Degree of bias analysis is carried out to transformed first component:
Wherein, SD is standard deviation.Skewness=0 illustrates that distributional pattern is identical as the normal state degree of bias；Skewness > 0, just
Partially, peak value is on a left side；Skewness < 0 is negative bias, and peak value is on the right side.| Skewness | bigger, distributional pattern degrees of offset is bigger.It is logical
Research discovery is crossed to be greater than containing the main region coefficient of skewness for slipping line without the main region coefficient of skewness for slipping line.Therefore, a certain disconnected
In the histogram set of graphs in face, take the position of coefficient of skewness maximum point as the position for slipping line main on current section.I.e. with 3 × 3 windows
Mouth calculates the coefficient of skewness, and the ordinate of the maximum point of the recording image each column coefficient of skewness, then by the vertical of 3 column adjacent in record
Coordinate is average, slips a position as master.
(11) master is extracted using spectral similarity and spatial continuity method slip initial position.
Extraction master is analyzed using the degree of bias and slips information, and initially master slips point Z for one group of acquisition(0).Spectral similarity and spatial continuity
Specific step is as follows for method:
First by artificial field exploring to master slip into a large amount of study of row, establish master and slip library of spectra.River is chosen to enter
Point most like in library of spectra is slipped with master at mouthful, slips sample as initial master.By calculating its direction line initially master slips a little,
And obtain its length and position, the specific steps are as follows:
Stepl define direction line be across center pel a series of line segments, their length is different, length by
Spectral similarity measurement and threshold value between adjacent picture elements determine.Similarity measure are as follows:
Wherein, x ' is xiAdjacent picture elements,Indicate mahalanobis distance.
Step2 direction line is extended according to specific rule from center pel towards both sides, while the new picture that will be selected into
Member is added result and combines.The condition of direction line extension is that the similarity measure of current pixel is less than threshold value th1.Here, th1For light
Similarity measurements threshold value is composed, it is related to the variation degree of pixel gray scale in the same shape region.Since water flow flowing has centainly
Directionality, the extension of actual direction line can not carry out towards all directions, but according to river direction, will be with river direction
The direction removal deviated from, then be extended.
Step3 traverses whole river, can track to obtain the direction line of initial pixel according to Stepl and two step of Step2, together
When the pixel collection Z that gradually forms when obtaining to form direction line(0)。
(12) iteration realizes that the main information synergism that slips extracts.
Master obtained in step (5) is slipped into point set Z(0)It is inputted in degree of bias analysis method as tutorial message, to its inside
Parameter is adjusted, if Z(0)With Y(0)Consistency L (Y(0),Z(0)) it is less than given threshold value th2, then terminate collaborative processes；Otherwise,
If Z(0)In pixel number be less than Y(0)In pixel number, degree of bias analysis window size reduce 1；Conversely, degree of bias analysis window size
Increase 1；By the degree of bias analysis method after adjusting parameter, the master for obtaining its output slips point set Y(1).Similarly, by Y(0)For instructing
The parameter of spectral similarity and spatial continuity method adjusts, if Z(0)With Y(0)Consistency L (Y(0),Z(0)) it is less than given threshold
Value, then terminate collaborative processes；Otherwise, if Z(0)In pixel number be less than Y(0)In pixel number, th1Increase by 2%；Conversely, th1Subtract
It is small by 2%；By the degree of bias analysis method after adjusting parameter, the master for obtaining its output slips point set Z(1)。
Above-mentioned collaborative processes are repeated, until algorithm is because of L (Y(t),Z(t)) it is less than given threshold value th2And terminate (t be iteration generation
Number), the testing result obtained at this time is that two kinds of algorithms tend to consistent as a result, having higher reliability after cooperateing with.
Claims (1)
1. a kind of Yellow River Main based on multi-spectral remote sensing image slips information synergism extracting method, it is characterised in that including following step
It is rapid:
(1) river is divided；
A width Landsat TM multispectral image is inputted, first progress river coarse segmentation；Using spectral classification and matching process into
The classification of row spectrum picture, and post-classification comparison is carried out according to the features of shape of the Yellow River section, using two kinds of samples respectively to image
Classify, then composograph；Due to there is the influence of bridge and beach in image, need to expand image and corrode behaviour
Make；
(2) riverbank line drawing；
Edge is detected using Canny operator, and tracking connection is carried out to the edge detected, sorts out river north and south bank
Flowage line；Tracking is attached to obtained edge line；According to neighborhood method, one group is obtained using a border following algorithm
Continuous line segment removes some shorter interference line segments according to length, while according to the style characteristic of beach, removing beach, obtaining
To one group of useful line segment；According to the distribution characteristics of Yellow River, the judgement of north and south bank is carried out to every section of line segment；Based on efficiency of algorithm
Consider the judgement that only need to carry out north and south bank to representational points certain in every section obtains which bank belonging to every section of flowage line
Information；According to statistical property, the number that the point judged in every section belongs to a certain bank is greater than a certain threshold value and judges that this section is to belong to
Yu Nanan or northern bank, conversely, then belonging to an other bank, to obtain north and south bank image；As needed the line segment of north and south two sides
Orderly storage, to obtain the complete flowage line in two sides, carries out master and slips line drawing into matrix respectively；
(3) river is segmented；
The Yellow River is divided into Typical River and atypia section；Typical River be divided into it is straight it is micro-bend, be bent and divide branch of a river three classes；Using curved
The bending degree and direction of bowed pastern number and curvature representation section are determined whether there is by the river distribution between the flowage line of north and south
Divide the branch of a river；Spatial relationship between section is defined as to Curved Continuous connects section and bend comes over and pledges allegiance to section, using between section
Consecutive variations arrangement distinguished from space；Specific segmentation method is as follows:
A) the dam bank in space is transformed to by curvature domain by the curvature estimation to open a window, the maximum point position of gained curvature sequence,
Where the position for representing the curved top of bend；
B) the minimum point position between two continuous threshold points represents the transition and linkage point position between two continuous river bends；
(4) extraction master is analyzed using the degree of bias slip initial position；
Extraction master is analyzed using the degree of bias and slips information, and initially master slips point Y for one group of acquisition(0)；Specific step is as follows for degree of bias parser:
Step1 obtains p × n observing matrix data X=[X1,X2,…,Xn], wherein each column XiAn observation sample vector is represented,
Every a line represents an observation attribute；Two class sample data S1,S2；
Step2 seeks the average deviation form B of X: enablingThenThe center that reference axis is moved to former data, seeks S1,S2Sample average M1, M2；
Step3 seeks between class scatter matrix Gb, between class scatter matrix GbIt is the positive semidefinite matrix of p × p, is defined as Gb=(M1-M2)
(M1-M2)T；
Step4 seeks between class scatter matrix GbEach characteristic value evaliWith feature vector eigi, i ∈ [1, p]；
Sequence arrangement of the Step5 by characteristic value from big to small, and corresponding feature vector is selected, constitute transformation matrix T=
(eig1,eig2,…eigm), m≤n；
Step6 generates the data set H:H=T in new coordinate systemT·X；Degree of bias analysis is carried out to transformed first component:
Wherein, SD is standard deviation；Skewness=0 illustrates that distributional pattern is identical as the normal state degree of bias；Skewness > 0, positively biased, peak
Value is on a left side；Skewness < 0 is negative bias, and peak value is on the right side；| Skewness | bigger, distributional pattern degrees of offset is bigger；A certain disconnected
In the histogram set of graphs in face, take the position of coefficient of skewness maximum point as the position for slipping line main on current section；I.e. with 3 × 3 windows
Mouth calculates the coefficient of skewness, and the ordinate of the maximum point of the recording image each column coefficient of skewness, then by the vertical of 3 column adjacent in record
Coordinate is average, slips a position as master；
(5) master is extracted using spectral similarity and spatial continuity method slip initial position；
Slipping initial position using spectral similarity and spatial continuity method extraction master, specific step is as follows:
First by artificial field exploring to master slip into a large amount of study of row, establish master and slip library of spectra；It chooses at bayou
Point most like in library of spectra is slipped with master, slips sample as initial master；By initially master slips a little, calculating its direction line, and obtain
To its length and position, the specific steps are as follows:
It is a series of line segments across center pel that Step a, which defines direction line, their length is different, and length is by phase
Spectral similarity measurement and threshold value between adjacent pixel determine；Similarity measure are as follows:
Wherein, x ' is xiAdjacent picture elements,Indicate mahalanobis distance；
Step b direction line is extended according to specific rule from center pel towards both sides, while the new pixel being selected into being added
Enter result combination；The condition of direction line extension is that the similarity measure of current pixel is less than threshold value th1；Here, th1For spectrum phase
Like property degree threshold value, it is related to the variation degree of pixel gray scale in the same shape region；Since water flow flowing has certain side
Tropism, actual direction line extension can not be carried out towards all directions, but according to river direction, it will deviate from river direction
Direction removal, then be extended；
Step c traverses whole river, tracks to obtain the direction line of initial pixel according to two step of Step a and Step b, obtain simultaneously
The pixel collection Z gradually formed when must form direction line(0)；
(6) iteration realizes that the main information synergism that slips extracts；
Master obtained in step (5) is slipped into point set Z(0)It is inputted in degree of bias analysis method as tutorial message, to its inner parameter
It is adjusted, if Z(0)With Y(0)Consistency L (Y(0),Z(0)) it is less than given threshold value th2, then terminate collaborative processes；Otherwise, if Z(0)In pixel number be less than Y(0)In pixel number, degree of bias analysis window size reduce 1；Conversely, degree of bias analysis window size increases
1；By the degree of bias analysis method after adjusting parameter, the master for obtaining its output slips point set Y(1)；Similarly, by Y(0)For instructing spectrum
The parameter of similitude and spatial continuity method adjusts, if Z(0)With Y(0)Consistency L (Y(0),Z(0)) it is less than given threshold value th2,
Then terminate collaborative processes；Otherwise, if Z(0)In pixel number be less than Y(0)In pixel number, th1Increase by 2%；Conversely, th1Reduce
2%；By the spectral similarity and spatial continuity method after adjusting parameter, the master for obtaining its output slips point set Z(1)；
Above-mentioned collaborative processes are repeated, until algorithm is because of L (Y(t),Z(t)) it is less than given threshold value th2And terminate, t is iterative algebra.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610224322.3A CN105913430B (en) | 2016-04-12 | 2016-04-12 | Yellow River Main based on multi-spectral remote sensing image slips information synergism extracting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610224322.3A CN105913430B (en) | 2016-04-12 | 2016-04-12 | Yellow River Main based on multi-spectral remote sensing image slips information synergism extracting method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105913430A CN105913430A (en) | 2016-08-31 |
| CN105913430B true CN105913430B (en) | 2019-01-15 |
Family
ID=56745874
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610224322.3A Active CN105913430B (en) | 2016-04-12 | 2016-04-12 | Yellow River Main based on multi-spectral remote sensing image slips information synergism extracting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105913430B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070014435A1 (en) * | 2005-07-13 | 2007-01-18 | Schlumberger Technology Corporation | Computer-based generation and validation of training images for multipoint geostatistical analysis |
| CN101980295A (en) * | 2010-09-25 | 2011-02-23 | 西北工业大学 | Method for detecting mainstream line of Yellow River based on skewness analysis |
| CN101980296A (en) * | 2010-09-25 | 2011-02-23 | 西北工业大学 | Spectral de-aliasing-based Yellow River mainstream line detection method |
| CN101980294A (en) * | 2010-09-25 | 2011-02-23 | 西北工业大学 | Remote sensing image-based method for detecting ice flood of Yellow River |
| CN102013097A (en) * | 2010-09-25 | 2011-04-13 | 西北工业大学 | Yellow River main humping line detection method based on spectrum similarity and space continuity |
-
2016
- 2016-04-12 CN CN201610224322.3A patent/CN105913430B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070014435A1 (en) * | 2005-07-13 | 2007-01-18 | Schlumberger Technology Corporation | Computer-based generation and validation of training images for multipoint geostatistical analysis |
| CN101980295A (en) * | 2010-09-25 | 2011-02-23 | 西北工业大学 | Method for detecting mainstream line of Yellow River based on skewness analysis |
| CN101980296A (en) * | 2010-09-25 | 2011-02-23 | 西北工业大学 | Spectral de-aliasing-based Yellow River mainstream line detection method |
| CN101980294A (en) * | 2010-09-25 | 2011-02-23 | 西北工业大学 | Remote sensing image-based method for detecting ice flood of Yellow River |
| CN102013097A (en) * | 2010-09-25 | 2011-04-13 | 西北工业大学 | Yellow River main humping line detection method based on spectrum similarity and space continuity |
Non-Patent Citations (8)
| Title |
|---|
| A Study on Yellow River Mainstream Line Feature Extraction;Zhang Haichao 等;《Proceedings of the 3rd International Yellow River Forum》;20071231;68-72 |
| Detection the river regime variation of the Yellow River based on remote sensing imagery;Hongwei She 等;《2009 Fifth International Conference on Image and Graphics》;20091231;744-748 |
| Interpretation of River Main-flow from Remote Sensing Images：Studying on dynamic transmission cross-correlation method;Han Lin 等;《2008 IEEE International Joint Conference on Neural Networks》;20081231;2281-2289 |
| Main-stream Feature Extraction of the Yellow River Based on Regional Spectral Un-mixing;Zhang Yanning 等;《Proceedings of the 4th International Yellow River Forum》;20091231;212-219 |
| SAR图像中的黄河主溜线提取;董光行 等;《计算机工程与应用》;20101231;第46卷(第14期);228-230 |
| 基于小波分析的河道主溜线遥感检测研究;韩琳 等;《遥感技术与应用》;20121031;第27卷(第5期);790-796 |
| 线形特征检测及其在黄河主溜线上的应用;王志印 等;《计算机应用》;20080331;第28卷(第3期);669-670、673 |
| 融合光谱和空间信息的黄河主溜特征提取;赵娜 等;《微电子学与计算机》;20100131;第27卷(第1期);169-172 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105913430A (en) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wei et al. | Toward automatic building footprint delineation from aerial images using CNN and regularization | |
| Monegaglia et al. | Automated extraction of meandering river morphodynamics from multitemporal remotely sensed data | |
| Huang et al. | A multidirectional and multiscale morphological index for automatic building extraction from multispectral GeoEye-1 imagery | |
| CN106228544B (en) | A kind of conspicuousness detection method propagated based on rarefaction representation and label | |
| US7400770B2 (en) | Method and apparatus for automatically extracting geospatial features from multispectral imagery suitable for fast and robust extraction of landmarks | |
| CN105469094B (en) | A kind of edge vectors line drawing method of road surface bianry image | |
| CN105139379B (en) | Based on the progressive extracting method of classified and layered airborne Lidar points cloud building top surface | |
| CN107424161B (en) | Coarse-to-fine indoor scene image layout estimation method | |
| CN109523566A (en) | A kind of automatic division method of Sandstone Slice micro-image | |
| CN105931241B (en) | A kind of automatic marking method of natural scene image | |
| CN101901343A (en) | Remote sensing image road extracting method based on stereo constraint | |
| US20230334627A1 (en) | Method for segmenting and denoising triangle mesh | |
| Gamba et al. | Urban climate zone detection and discrimination using object-based analysis of VHR scenes | |
| CN110598748B (en) | Heterogeneous image change detection method and device based on convolutional neural network fusion | |
| Gleason et al. | A fusion approach for tree crown delineation from lidar data. | |
| CN110084107A (en) | A kind of high-resolution remote sensing image method for extracting roads and device based on improvement MRF | |
| CN111476723B (en) | Remote sensing image lost pixel recovery method for failure of Landsat-7 scanning line corrector | |
| CN105809673A (en) | SURF (Speeded-Up Robust Features) algorithm and maximal similarity region merging based video foreground segmentation method | |
| Shao et al. | A kind of pavement crack detection method based on digital image processing | |
| CN101980295A (en) | Method for detecting mainstream line of Yellow River based on skewness analysis | |
| Peng et al. | Incorporating generic and specific prior knowledge in a multiscale phase field model for road extraction from VHR images | |
| Mu et al. | McGET: A rapid image-based method to determine the morphological characteristics of gravels on the Gobi desert surface | |
| Zhang | A method for continuous extraction of multispectrally classified urban rivers | |
| CN104820992B (en) | A kind of remote sensing images Semantic Similarity measure and device based on hypergraph model | |
| Bouaziz et al. | Remote gully erosion mapping using aster data and geomorphologic analysis in the Main Ethiopian Rift |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |