CN104615999A - Landslide debris flow area detection method based on sparse representation classification - Google Patents
Landslide debris flow area detection method based on sparse representation classification Download PDFInfo
- Publication number
- CN104615999A CN104615999A CN201510081296.9A CN201510081296A CN104615999A CN 104615999 A CN104615999 A CN 104615999A CN 201510081296 A CN201510081296 A CN 201510081296A CN 104615999 A CN104615999 A CN 104615999A
- Authority
- CN
- China
- Prior art keywords
- landslide
- image block
- rarefaction representation
- doubtful
- testing image
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 32
- 238000000034 method Methods 0.000 claims abstract description 53
- 238000012360 testing method Methods 0.000 claims description 72
- 238000010606 normalization Methods 0.000 claims description 19
- 238000004891 communication Methods 0.000 claims description 3
- 230000004069 differentiation Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 239000004575 stone Substances 0.000 description 3
- 239000002131 composite material Substances 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 230000000116 mitigating effect Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- WURBVZBTWMNKQT-UHFFFAOYSA-N 1-(4-chlorophenoxy)-3,3-dimethyl-1-(1,2,4-triazol-1-yl)butan-2-one Chemical compound C1=NC=NN1C(C(=O)C(C)(C)C)OC1=CC=C(Cl)C=C1 WURBVZBTWMNKQT-UHFFFAOYSA-N 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Pit Excavations, Shoring, Fill Or Stabilisation Of Slopes (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a landslide debris flow area detection method based on sparse representation classification. The method includes: dividing a target area image to be detected into multiple image blocks to be detected with the same sizes, analyzing and judging the brightness of three channels of red, green and blue through a sparse representation classification method, obtaining a suspected landslide debris flow image block collection, presetting the mean value of the difference between the first threshold value and the sparse representation errors; obtaining the landslide debris flow severely damaged area and the area boundaries in the suspected landslide debris flow image block collection through a connected area detection method; obtaining the land type of the image of each point before a disaster according to the position of each point of the area boundaries, obtaining the expansion degree influence coefficient of each point corresponds to the disaster loss weight according to the preset land type, obtaining the extended distance of the landslide debris flow area according to the mean value of the difference between the preset first threshold value and the sparse representation errors, and obtaining the landslide debris flow general damaged area through the extension of the landslide debris flow seriously damaged area. The landslide debris flow detection method is high in detection efficiency and accuracy and low in false alarm rate.
Description
Technical field
The present invention relates to landslide region detection technical field, particularly relate to a kind of landslide method for detecting area based on rarefaction representation classification (sparse representation-based classifier is called for short SRC).
Background technology
Define according to the theory of calamity, landslide to refer on slope that certain a part of ground is under gravity (comprising the using dynamic and static pressures of the gravity of ground own and underground water) effect, effect and phenomenon along certain weak structural face (band) generation shear displacemant integrally to movement below slope; And rubble flow refers in the dangerously steep area of mountain area or other cheuch deep gullies, landform, because heavy rain, severe snow or other disasteies landslide of causing carry the special mighty torrent of a large amount of silt and stone.The global feature such as profile, shape that landslide and the Forming Mechanism of rubble flow make its region both ununified on remote sensing images, the local features such as uniform texture, gradient are not had yet, existing detection method is mostly applicable to landslide and generates position and its periphery morphologic characteristics and have the situation comparatively obviously distinguished or assist by information such as DEM and carry out region, scope interpretation, exist and require harsh to Data Source and lack the problems such as broad applicability, be not suitable for utilizing xx data to carry out the application requirement of the condition of a disaster assessment.
Given this, how efficient, high accuracy is carried out to landslide region and the detection of low false alarm rate becomes the current technical issues that need to address.
Summary of the invention
For defect of the prior art, the invention provides a kind of landslide method for detecting area based on rarefaction representation classification, can carry out efficient detection to the landslide region of target area Aerial Images, accuracy is high, and false alarm rate is low.
First aspect, the invention provides a kind of landslide method for detecting area based on rarefaction representation classification, comprising:
Target area testing image is divided into the identical M of a size testing image block, M be greater than 1 integer;
The method utilizing rarefaction representation to classify is analyzed the three-channel brightness of described testing image block red, green, blue and differentiates, to obtain the set of the doubtful landslide image block in testing image block, and obtain the average of the difference of preset first threshold value and rarefaction representation error;
Utilize connected region detection method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof;
According to the position of each point on described zone boundary, obtain the land type of each point before calamity in image;
According to the land type of described each point before calamity in image and disaster-stricken loss weight corresponding to default land type, obtain the influence coefficient of degree of expansion corresponding to each point;
According to the influence coefficient of the average of described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, obtain the extended range in landslide region, and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region.
Alternatively, the described method utilizing rarefaction representation to classify is to the three-channel brightness analysis of described testing image block red, green, blue and differentiation, to obtain the set of the doubtful landslide image block in testing image block, and obtain the average of the difference of preset first threshold value and rarefaction representation error, comprising:
By testing image block
carry out vectorization, normalization successively;
Complete dictionary A was built according to the typical sample of red, green, blue triple channel component
q(q=R, G, B), wherein, R, G, B are respectively red, green, blue triple channel component;
Obtain the testing image block after vectorization and normalization
rarefaction representation coefficient
According to described rarefaction representation coefficient
to described testing image block
be reconstructed, obtain rarefaction representation error
According to described rarefaction representation error
obtain preset first threshold value
with rarefaction representation error
difference
Will
compare with default Second Threshold,
during>=default Second Threshold, determine testing image block
red/green/blue channel component image block is doubtful landslide image block;
At testing image block
red, green, blue triple channel component image block when being doubtful landslide image block, determine that this testing image block is the doubtful image block of landslide, and then obtain the set of doubtful landslide image block;
Described preset first threshold value when acquisition testing image block is landslide doubtful image block
with rarefaction representation error
difference
average
Alternatively, the testing image block after described acquisition vectorization and normalization
rarefaction representation coefficient
comprise:
Testing image block after utilizing orthogonal matching pursuit OMP algorithm to obtain vectorization and normalization
rarefaction representation coefficient
Alternatively, described utilize orthogonal matching pursuit algorithm OMP algorithm to obtain vectorization and normalization after testing image block
rarefaction representation coefficient
comprise:
According to the testing image block after vectorization and normalization
with red, green, blue Three Channel Color component dictionary A
q(q=R, G, B), establishing target function:
Solve objective function, obtain the testing image block after vectorization and normalization
rarefaction representation coefficient
Alternatively, described according to described rarefaction representation coefficient
to described testing image block
be reconstructed, obtain rarefaction representation error
comprise:
According to described rarefaction representation coefficient
by the first formula to described testing image block
be reconstructed, obtain rarefaction representation error
Described first formula is:
Alternatively, described general
compare with default Second Threshold, also comprise:
?
when < presets Second Threshold, determine testing image block
red/green/blue channel component image block is non-landslide image block.
Alternatively, described default Second Threshold is 0.
Alternatively, describedly utilize connected region detection method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof, comprising:
Utilize the connected region detection method of eight neighborhood method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof.
Alternatively, the described connected region detection method utilizing eight neighborhood method, obtains landslide and seriously damages region and zone boundary thereof, comprising in the set of described doubtful landslide image block:
Using the central point of each doubtful landslide image block as a gauge point (i, j);
From each gauge point, each direction along eight neighborhood is searched for s time, if find other gauge points, then the party is upwards all designated as gauge point between original tally point and newfound gauge point;
Communication with detection is carried out to all gauge points obtained, in doubtful landslide image block, obtains landslide seriously damage region;
Wherein, the value of s depends on the size of gauge point position, target area image size and doubtful landslide image block.
Alternatively, the influence coefficient of the described average according to described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, obtain the extended range in landslide region, and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region, comprising:
According to the influence coefficient of the average of described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, extended range l (the i in landslide region is obtained by the 5th formula, j), and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region;
Wherein, described 5th formula is:
for the average of the difference of preset first threshold value and rarefaction representation error, θ
nfor n-th kind of land type corresponding to each point on described zone boundary is to the influence coefficient of degree of expansion.
As shown from the above technical solution, the landslide method for detecting area based on rarefaction representation classification of the present invention, can carry out efficient detection to the landslide region of target area Aerial Images, accuracy is high, and false alarm rate is low.
Accompanying drawing explanation
The schematic flow sheet of the landslide method for detecting area based on rarefaction representation classification that Fig. 1 provides for one embodiment of the invention;
The schematic diagram of the connected region detection algorithm of the eight neighborhood method that Fig. 2 provides for the embodiment of the present invention;
Aerial Images after area, Zhouqu County Debris-flow Hazard that Fig. 3 provides for the embodiment of the present invention;
Fig. 4 analyzes for using the method utilizing rarefaction representation to classify in method described in the present embodiment of the present invention to carry out the three-channel brightness of red, green, blue to the Aerial Images block after the Debris-flow Hazard of area, Zhouqu County and differentiates, obtains the triple channel Detection results composite diagram after the set of doubtful landslide image block;
Fig. 5 utilizes connected region detection method for using in method described in the embodiment of the present invention, obtain in the doubtful landslide image block in area, Zhouqu County landslide seriously damage region after design sketch;
Fig. 6 is for using the extended range of acquisition landslide range of influence the rapid evaluation figure in the area, Zhouqu County after expanding that described in the embodiment of the present invention, method provides;
Zhouqu County rubble flow, Remote Sensing For Landslides surveillance map that Fig. 7 provides for existing Ministry of Civil Affairs's mitigation center;
Reference numeral:
1, landslide region; 2, landslide seriously damages region; 3, landslide generally damages region.
Embodiment
For making the object of the embodiment of the present invention, technical scheme and advantage clearly, below in conjunction with the accompanying drawing in the embodiment of the present invention, clear, complete description is carried out to the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on embodiments of the invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
According to theory of calamity definition, on the one hand, in landslide region, necessarily there is silt and the stone of not determined number, image shows as similarity and the regularity of local; On the other hand, silt or stone etc. have certain continuity in the existence in landslide region.Thus, we propose the landslide method for detecting area based on rarefaction representation classification SRC, comprise based on rarefaction representation classification SRC doubtful rubble flow method for detecting area, seriously to damage method for detecting area based on the landslide of eight neighborhood and generally damage method for detecting area based on the landslide of information before rarefaction representation and calamity.
The method of rarefaction representation derives from human vision research.Research finds, in human visual system, different neuronal cell is for different stimulations, as color, texture, yardstick, direction etc. produce reaction, these neuronal cells form a complete dictionary of mistake thus represent various visual effect, and this expression is sparse.Meanwhile, if the image of certain type things (or signal) has similarity, so build dictionary with the typical image sample of the type things, linear expression can be carried out to testing image, the principle of Here it is rarefaction representation.If set up the sub-dictionary of dissimilar things, the rarefaction representation error of each sub-dictionary to testing image just can be utilized to classify to it, differentiate, this is the ultimate principle of classification (SRC) method based on rarefaction representation.
Fig. 1 shows the schematic flow sheet of the landslide method for detecting area based on rarefaction representation classification that one embodiment of the invention provides, and as shown in Figure 1, the landslide method for detecting area based on rarefaction representation classification of the present embodiment is as described below.
101, target area testing image is divided into the identical M of a size testing image block, M be greater than 1 integer.
Will be understood that, with the pane of pre-set dimension, testing image Y can be divided into M testing image block, obtain testing image set of blocks
wherein, R, G, B are respectively red R, green G, blue triple channel component.
102, the method utilizing rarefaction representation to classify is to the three-channel brightness analysis of described testing image block red, green, blue and differentiation, to obtain the set of the doubtful landslide image block in testing image block, and obtain the average of the difference of preset first threshold value and rarefaction representation error.
Will be understood that, observe landslide debris stream picture, what can find landslide region has certain unitarity in color, can the brightness of different color channels be analyzed, be differentiated thus, thus obtain the set of the doubtful landslide image block in testing image block, and obtain the average of the difference of preset first threshold value and rarefaction representation error.
103, utilize connected region detection method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof.
104, according to the position of each point on described zone boundary, the land type of each point before calamity in image is obtained.
105, according to the land type of described each point before calamity in image and disaster-stricken loss weight corresponding to default land type, the influence coefficient of degree of expansion corresponding to each point is obtained.
It should be noted that disaster-stricken loss weight corresponding to the described default land type of this step 105 is arranged according to expertise.
106, according to the influence coefficient of the average of described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, obtain the extended range in landslide region, and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region.
Will be understood that, according to the analysis to landslide region, the range of influence having certain limit around major disaster region can be found, and this scope is relevant with land type, thus, in conjunction with testing result after land use type before calamity and calamity, the extended range of landslide range of influence can be obtained.
The landslide method for detecting area based on rarefaction representation classification of the present embodiment, the method of being classified by rarefaction representation obtains the set of doubtful landslide image block, the landslide that connected region detection method obtains in testing image seriously damages region and zone boundary thereof, the land type of each point before calamity in image and the influence coefficient to degree of expansion of its correspondence is obtained according to various point locations on described zone boundary, the extended range in landslide region is obtained according to preset first threshold value and the average of the difference of rarefaction representation error and the influence coefficient of the corresponding degree of expansion of each point, and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region, efficient detection can be carried out to the landslide region in the Aerial Images to be measured of target area, accuracy is high, false alarm rate is low.
In a particular application, above-mentioned steps 102 can comprise not shown step 102a-102h:
102a, by testing image block
carry out vectorization, normalization successively.
102b, built complete dictionary A according to the typical sample of red, green, blue triple channel component
q(q=R, G, B), wherein, R, G, B are respectively red, green, blue triple channel component.
Will be understood that, by typical image sample vector, then that these are vectorial as atom one by one, carry out arrangement set, just constitute a matrix, this matrix is referred to as dictionary by us.Wherein matrix column number represents sample number, i.e. atom number.Suppose there is dictionary A ∈ R
mxnif m<n, then claim this dictionary to be complete,
y=Ax
Wherein, y is pending image block (or signal), and A is in advance according to the dictionary that typical sample establishes, and is complete, and x is linear coefficient vector.In x vector, the number of non-zero element is few as much as possible, namely has openness.When given dictionary, the object of rarefaction representation selects coefficient vector sparse as far as possible most, carrys out reconstructed image block y.
If set up the sub-dictionary of dissimilar things, the rarefaction representation error of each sub-dictionary to testing image just can be utilized to classify to it, differentiate, this is the ultimate principle of classification (SRC) method based on rarefaction representation.Because the embodiment of the present invention only needs to detect landslide region, and without the need to detecting other atural object classification, therefore, only set up the dictionary of this type of landslide.The embodiment of the present invention first chooses the landslide image block of some typicalness as sample, these image block vectorizations assembled a matrix, then matrix is carried out Gaussian number projection, namely obtain the dictionary that we want.
Testing image block after 102c, acquisition vectorization and normalization
rarefaction representation coefficient
In a particular application, for example, above-mentioned steps 102c can be preferably:
Orthogonal matching pursuit OMP (orthogonal matching pursuit is called for short OMP) algorithm is utilized to obtain the testing image block after vectorization and normalization
rarefaction representation coefficient
it specifically can comprise:
According to the testing image block after vectorization and normalization
with red, green, blue Three Channel Color component dictionary A
q(q=R, G, B), establishing target function:
Solve objective function, obtain the testing image block after vectorization and normalization
rarefaction representation coefficient
102d, according to described rarefaction representation coefficient
to described testing image block
be reconstructed, obtain rarefaction representation error
(namely reconstructing redundancy).
In a particular application, above-mentioned steps 102d can comprise:
According to described rarefaction representation coefficient
by the first formula to described testing image block
be reconstructed, obtain rarefaction representation error
Described first formula is:
102e, according to described rarefaction representation error
obtain preset first threshold value
with rarefaction representation error
difference
Wherein,
102f, general
compare with default Second Threshold,
during>=default Second Threshold, determine testing image block
red/green/blue channel component image block is doubtful landslide image block.
In a particular application, above-mentioned steps 102f can also comprise:
?
when < presets Second Threshold, determine testing image block
red/green/blue channel component image block is non-landslide image block.
For example, the default Second Threshold in this step 102f can be 0, can utilize pre-set identification and classification device
be, by testing image block
be divided into two classes:
Wherein, 1 represents image block
be judged as doubtful landslide image block, 0 represents image block
be judged as non-landslide image block.
102g, at testing image block
red, green, blue triple channel component image block when being doubtful landslide image block, determine that this testing image block is the doubtful image block of landslide, and then obtain the set of doubtful landslide image block.
Will be understood that, for example, if the default Second Threshold in above-mentioned steps 102f is 0, and utilize pre-set identification and classification device
time, by testing image block
be divided into two classes:
Wherein, 1 represents image block
be judged as doubtful landslide image block, 0 represents image block
be judged as non-landslide image block;
Then can at identity (Y in this step 102g
pduring)=1, determine that this testing image block is the doubtful image block of landslide, and then obtain the set { identity (Y of doubtful landslide image block
p)=1};
Wherein,
Described preset first threshold value when 102h, acquisition testing image block are landslide doubtful image block
with rarefaction representation error
difference
average
Wherein,
In a particular application, for example, above-mentioned steps 103 can be preferably:
Utilize the connected region detection method of eight neighborhood method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof, as shown in Figure 2, it specifically can comprise:
Using the central point of each doubtful landslide image block as a gauge point (i, j);
From each gauge point, each direction along eight neighborhood is searched for s time, if find other gauge points, then the party is upwards all designated as gauge point between original tally point and newfound gauge point;
Carry out communication with detection to all gauge points obtained, in doubtful landslide image block, the landslide obtained in testing image seriously damages region;
Wherein, the value of s depends on the size of gauge point position, target area image size and doubtful landslide image block.
In a particular application, this step 106 can comprise:
According to the influence coefficient of the average of described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, extended range l (the i in landslide region is obtained by the 5th formula, j), and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region;
Wherein, described 5th formula is:
for the average of the difference of preset first threshold value and rarefaction representation error, θ
nfor n-th kind of land type corresponding to each point on described zone boundary is to the influence coefficient of degree of expansion.
Through to choose Zhouqu County area Debris-flow Hazard after Aerial Images (i.e. Fig. 3) use method described in the present embodiment to detect, obtain design sketch as Figure 4-Figure 6, Fig. 4 shows and uses the method utilizing rarefaction representation to classify in method described in the present embodiment of the present invention to carry out red to the Aerial Images block after the Debris-flow Hazard of area, Zhouqu County, green, blue three-channel brightness is carried out analyzing and differentiating, obtain the triple channel Detection results composite diagram after the set of doubtful landslide image block, Fig. 5 shows to use and utilizes connected region detection method in method described in the embodiment of the present invention, Zhouqu County area doubtful landslide image block in obtain landslide seriously damage region after design sketch, Fig. 6 shows and uses the extended range of acquisition landslide range of influence the rapid evaluation figure in the area, Zhouqu County after expanding that described in the embodiment of the present invention, method provides, Fig. 7 shows the Zhouqu County rubble flow that existing Ministry of Civil Affairs's mitigation center provides, Remote Sensing For Landslides surveillance map, Fig. 6 and Fig. 7 is compared, through statistics, the detection accuracy that Fig. 6 detects the landslide region, area, Zhouqu County obtained is 83.35%, false alarm rate is 2.43%, be 6240 × 6431 in experiment test image size, when win7 64, I52.9GHz 4G memory machines is tested, the whole flow process used time is about 55min.
The landslide method for detecting area based on rarefaction representation classification of the present embodiment, can carry out efficient detection to the landslide region in the Aerial Images to be measured of target area, accuracy is high, and false alarm rate is low.
One of ordinary skill in the art will appreciate that: all or part of step realizing above-mentioned each embodiment of the method can have been come by the hardware that programmed instruction is relevant.Aforesaid program can be stored in a computer read/write memory medium.This program, when performing, performs the step comprising above-mentioned each embodiment of the method; And aforesaid storage medium comprises: ROM, RAM, magnetic disc or CD etc. various can be program code stored medium.
Last it is noted that above each embodiment is only in order to illustrate technical scheme of the present invention, be not intended to limit; Although with reference to foregoing embodiments to invention has been detailed description, those of ordinary skill in the art is to be understood that: it still can be modified to the technical scheme described in foregoing embodiments, or carries out equivalent replacement to wherein some or all of technical characteristic; And these amendments or replacement, do not make the essence of appropriate technical solution depart from the scope of the claims in the present invention.
Claims (10)
1., based on a landslide method for detecting area for rarefaction representation classification, it is characterized in that, comprising:
Target area testing image is divided into the identical M of a size testing image block, M be greater than 1 integer;
The method utilizing rarefaction representation to classify is analyzed the three-channel brightness of described testing image block red, green, blue and differentiates, to obtain the set of the doubtful landslide image block in testing image block, and obtain the average of the difference of preset first threshold value and rarefaction representation error;
Utilize connected region detection method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof;
According to the position of each point on described zone boundary, obtain the land type of each point before calamity in image;
According to the land type of described each point before calamity in image and disaster-stricken loss weight corresponding to default land type, obtain the influence coefficient of degree of expansion corresponding to each point;
According to the influence coefficient of the average of described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, obtain the extended range in landslide region, and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region.
2. method according to claim 1, it is characterized in that, the described method utilizing rarefaction representation to classify is to the three-channel brightness analysis of described testing image block red, green, blue and differentiation, to obtain the set of the doubtful landslide image block in testing image block, and obtain the average of the difference of preset first threshold value and rarefaction representation error, comprising:
By testing image block
carry out vectorization, normalization successively;
Complete dictionary A was built according to the typical sample of red, green, blue triple channel component
q(q=R, G, B), wherein, R, G, B are respectively red, green, blue triple channel component;
Obtain the testing image block after vectorization and normalization
rarefaction representation coefficient
According to described rarefaction representation coefficient
to described testing image block
be reconstructed, obtain rarefaction representation error
According to described rarefaction representation error
obtain preset first threshold value
with rarefaction representation error
difference
Will
compare with default Second Threshold,
when presetting Second Threshold, determine testing image block
red/green/blue channel component image block is doubtful landslide image block;
At testing image block
red, green, blue triple channel component image block when being doubtful landslide image block, determine that this testing image block is the doubtful image block of landslide, and then obtain the set of doubtful landslide image block;
Described preset first threshold value when acquisition testing image block is landslide doubtful image block
with rarefaction representation error
difference
average
3. method according to claim 2, is characterized in that, the testing image block after described acquisition vectorization and normalization
rarefaction representation coefficient
comprise:
Testing image block after utilizing orthogonal matching pursuit OMP algorithm to obtain vectorization and normalization
rarefaction representation coefficient
4. method according to claim 3, is characterized in that, described utilize orthogonal matching pursuit algorithm OMP algorithm to obtain vectorization and normalization after testing image block
rarefaction representation coefficient
comprise:
According to the testing image block after vectorization and normalization
with red, green, blue Three Channel Color component dictionary A
q(q=R, G, B), establishing target function:
Solve objective function, obtain the testing image block after vectorization and normalization
rarefaction representation coefficient
5. method according to claim 2, is characterized in that, described according to described rarefaction representation coefficient
to described testing image block
be reconstructed, obtain rarefaction representation error
comprise:
According to described rarefaction representation coefficient
by the first formula to described testing image block
be reconstructed, obtain rarefaction representation error
Described first formula is:
6. method according to claim 2, is characterized in that, described general
compare with default Second Threshold, also comprise:
?
time, determine testing image block
red/green/blue channel component image block is non-landslide image block.
7. the method according to claim 5 or 6, is characterized in that, described default Second Threshold is 0.
8. method according to claim 1, is characterized in that, describedly utilizes connected region detection method, obtains landslide and seriously damages region and zone boundary thereof, comprising in the set of described doubtful landslide image block:
Utilize the connected region detection method of eight neighborhood method, in the set of described doubtful landslide image block, obtain landslide seriously damage region and zone boundary thereof.
9. method according to claim 8, is characterized in that, the described connected region detection method utilizing eight neighborhood method, obtains landslide and seriously damages region and zone boundary thereof, comprising in the set of described doubtful landslide image block:
Using the central point of each doubtful landslide image block as a gauge point (i, j);
From each gauge point, each direction along eight neighborhood is searched for s time, if find other gauge points, then the party is upwards all designated as gauge point between original tally point and newfound gauge point;
Communication with detection is carried out to all gauge points obtained, in doubtful landslide image block, obtains landslide seriously damage region;
Wherein, the value of s depends on the size of gauge point position, target area image size and doubtful landslide image block.
10. method according to claim 9, it is characterized in that, the influence coefficient of the described average according to described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, obtain the extended range in landslide region, and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region, comprising:
According to the influence coefficient of the average of described preset first threshold value and the difference of rarefaction representation error and degree of expansion corresponding to each point, extended range l (the i in landslide region is obtained by the 5th formula, j), and utilize this extended range seriously to damage region to landslide to expand, obtain landslide and generally damage region;
Wherein, described 5th formula is:
for the average of the difference of preset first threshold value and rarefaction representation error, θ
nfor n-th kind of land type corresponding to each point on described zone boundary is to the influence coefficient of degree of expansion.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510081296.9A CN104615999B (en) | 2015-02-15 | 2015-02-15 | The landslide method for detecting area classified based on rarefaction representation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510081296.9A CN104615999B (en) | 2015-02-15 | 2015-02-15 | The landslide method for detecting area classified based on rarefaction representation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104615999A true CN104615999A (en) | 2015-05-13 |
CN104615999B CN104615999B (en) | 2017-11-07 |
Family
ID=53150436
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510081296.9A Expired - Fee Related CN104615999B (en) | 2015-02-15 | 2015-02-15 | The landslide method for detecting area classified based on rarefaction representation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104615999B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105469058A (en) * | 2015-11-30 | 2016-04-06 | 成都理工大学 | Sequence spectrum surface feature-based landslide identification method |
CN106205059A (en) * | 2016-06-23 | 2016-12-07 | 西北大学 | The early warning system of a kind of slide creep monitoring and method |
CN107871115A (en) * | 2016-11-01 | 2018-04-03 | 中国科学院沈阳自动化研究所 | A kind of recognition methods of the submarine hydrothermal solution spout based on image |
CN107944492A (en) * | 2017-11-27 | 2018-04-20 | 深圳市唯特视科技有限公司 | A kind of land use cover classification method based on rarefaction representation |
CN108920754A (en) * | 2018-05-25 | 2018-11-30 | 重庆交通大学 | The structural plane for shearing constitutive model and change in displacement based on strain softening controls slope stability dynamic evaluation method |
CN109558881A (en) * | 2018-11-22 | 2019-04-02 | 重庆广睿达科技有限公司 | A kind of crag avalanche monitoring method based on computer vision |
CN110807399A (en) * | 2019-10-29 | 2020-02-18 | 北京师范大学 | Single-category support vector machine-based collapse and slide hidden danger point detection method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102542295A (en) * | 2012-01-08 | 2012-07-04 | 西北工业大学 | Method for detecting landslip from remotely sensed image by adopting image classification technology |
CN102622583A (en) * | 2012-02-23 | 2012-08-01 | 北京师范大学 | Multi-angle type number recognition method and system based on model and sparse representations |
CN103049760A (en) * | 2012-12-27 | 2013-04-17 | 北京师范大学 | Image block and position weight-based sparse representation target recognition method |
CN103093472A (en) * | 2013-01-24 | 2013-05-08 | 西安电子科技大学 | Optical remote sensing image change detection method based on double dictionary cross sparse representation |
-
2015
- 2015-02-15 CN CN201510081296.9A patent/CN104615999B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102542295A (en) * | 2012-01-08 | 2012-07-04 | 西北工业大学 | Method for detecting landslip from remotely sensed image by adopting image classification technology |
CN102622583A (en) * | 2012-02-23 | 2012-08-01 | 北京师范大学 | Multi-angle type number recognition method and system based on model and sparse representations |
CN103049760A (en) * | 2012-12-27 | 2013-04-17 | 北京师范大学 | Image block and position weight-based sparse representation target recognition method |
CN103093472A (en) * | 2013-01-24 | 2013-05-08 | 西安电子科技大学 | Optical remote sensing image change detection method based on double dictionary cross sparse representation |
Non-Patent Citations (6)
Title |
---|
GRACIELA METTERNICHT等: "Remote sensing of landslides: An analysis of the potential contribution to geo-spatial systems for hazard assessment in mountainous environments", 《REMOTE SENSING OF ENVIRONMENT》 * |
LI ZHANG等: "Kernel Sparse Representation-Based Classifier", 《IEEE TRANSACTIONS ON SIGNAL PROCESSING》 * |
MENG YANG等: "Fisher Discrimination Dictionary Learning for Sparse Representation", 《2011 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION》 * |
宋琳等: "基于稀疏表示模型和自回归模型的高光谱分类", 《光学学报》 * |
李宏宏: "面向对象的高分辨率遥感影像分类及泥石流信息提取研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
王治华等: "基于数字滑坡技术的暴雨滑坡、泥石流预警、监测模型研究", 《地学前缘(中国地质大学(北京);北京大学)》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105469058A (en) * | 2015-11-30 | 2016-04-06 | 成都理工大学 | Sequence spectrum surface feature-based landslide identification method |
CN105469058B (en) * | 2015-11-30 | 2018-12-14 | 成都理工大学 | A kind of landslide identification method based on series spectral curved surface features |
CN106205059A (en) * | 2016-06-23 | 2016-12-07 | 西北大学 | The early warning system of a kind of slide creep monitoring and method |
CN107871115A (en) * | 2016-11-01 | 2018-04-03 | 中国科学院沈阳自动化研究所 | A kind of recognition methods of the submarine hydrothermal solution spout based on image |
CN107871115B (en) * | 2016-11-01 | 2021-05-04 | 中国科学院沈阳自动化研究所 | Image-based submarine hydrothermal vent identification method |
CN107944492A (en) * | 2017-11-27 | 2018-04-20 | 深圳市唯特视科技有限公司 | A kind of land use cover classification method based on rarefaction representation |
CN108920754A (en) * | 2018-05-25 | 2018-11-30 | 重庆交通大学 | The structural plane for shearing constitutive model and change in displacement based on strain softening controls slope stability dynamic evaluation method |
CN108920754B (en) * | 2018-05-25 | 2022-07-01 | 重庆交通大学 | Structural plane control slope stability dynamic evaluation method based on strain softening shearing constitutive model and displacement change |
CN109558881A (en) * | 2018-11-22 | 2019-04-02 | 重庆广睿达科技有限公司 | A kind of crag avalanche monitoring method based on computer vision |
CN110807399A (en) * | 2019-10-29 | 2020-02-18 | 北京师范大学 | Single-category support vector machine-based collapse and slide hidden danger point detection method |
Also Published As
Publication number | Publication date |
---|---|
CN104615999B (en) | 2017-11-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104615999A (en) | Landslide debris flow area detection method based on sparse representation classification | |
Akagic et al. | Pothole detection: An efficient vision based method using rgb color space image segmentation | |
US20150254499A1 (en) | Multi-view 3d object recognition from a point cloud and change detection | |
CN103364410B (en) | Crack detection method of hydraulic concrete structure underwater surface based on template search | |
KR101507333B1 (en) | Slope Monitoring Method, System Using The Method and Recording Medium For Performing The Method | |
CN102043958B (en) | High-definition remote sensing image multi-class target detection and identification method | |
CN103456020B (en) | Based on the method for detecting change of remote sensing image of treelet Fusion Features | |
CN102087790B (en) | Method and system for low-altitude ground vehicle detection and motion analysis | |
CN104408711A (en) | Multi-scale region fusion-based salient region detection method | |
KR102660837B1 (en) | Method for landslide crack detection based deep learning and Method for landslide monitoring therewith and Apparatus thereof | |
CN111462193A (en) | Object detection method and device for monitoring multiple cameras or surrounding views | |
CN103729653A (en) | Supervision change detection method of high-resolution remote sensing images | |
CN103955913A (en) | SAR image segmentation method based on line segment co-occurrence matrix characteristics and regional maps | |
Abou Chacra et al. | Fully automated road defect detection using street view images | |
CN115439750A (en) | Road disease detection method and device, electronic equipment and storage medium | |
Buza et al. | Unsupervised method for detection of high severity distresses on asphalt pavements | |
CN104036277A (en) | Method and equipment for extracting road characteristics | |
Dhiman et al. | A multi-frame stereo vision-based road profiling technique for distress analysis | |
CN104715231A (en) | Method and device for monitoring dangerous goods transportation vehicles at traffic intersection | |
CN105528787A (en) | Polarimetric SAR image bridge detection method and device based on level set segmentation | |
CN108876760A (en) | A kind of remote sensing image variation detection method based on history interpretation knowledge | |
Gecen et al. | Road detection from high and low resolution satellite images | |
CN103927523A (en) | Fog level detection method based on longitudinal gray features | |
Moughal et al. | An automatic unsupervised method based on context-sensitive spectral angle mapper for change detection of remote sensing images | |
Walicka et al. | Individual rocks segmentation in terrestrial laser scanning point cloud using iterative DBSCAN algorithm |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171107 |