CN108615059A - A kind of lake automatically selecting method and device based on Dynamic Multiscale cluster - Google Patents
A kind of lake automatically selecting method and device based on Dynamic Multiscale cluster Download PDFInfo
- Publication number
- CN108615059A CN108615059A CN201810443646.5A CN201810443646A CN108615059A CN 108615059 A CN108615059 A CN 108615059A CN 201810443646 A CN201810443646 A CN 201810443646A CN 108615059 A CN108615059 A CN 108615059A
- Authority
- CN
- China
- Prior art keywords
- lake
- cluster
- layer
- distance
- area
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to Map Generalization fields, and in particular to a kind of lake automatically selecting method and device based on Dynamic Multiscale cluster.The present invention sets area threshold according to target proportion ruler first, chooses the lake that all areas are more than threshold value；Then buffering area is built to all lakes, chooses important holding point of the lake of no-buffer overlapping relation as distribution characteristics；Next it mines massively to lake and divides region by distribution density difference with Dynamic Multiscale cluster；Different Selection Strategies finally are taken to different zones (i.e. small lakes group).The present invention effectively maintains lake mass selection under different target engineer's scale and front and back morphosis and density is taken to compare, and choosing result, there is good reasonability, the lake element that can be applied to topographic map under different target engineer's scale to choose integrated treatment automatically.
Description
Technical field
The present invention relates to Map Generalization fields, and in particular to a kind of lake side of selection automatically based on Dynamic Multiscale cluster
Method and device.
Background technology
The lake of big land distribution is thousands of, and area and amount of capacity great disparity are larger.See that lake is unevenly distributed on the whole,
They have plenty of the formal distribution with Hu Qunjuji, and are interconnected by river, have plenty of with scattered formal distribution.To system
For figure person, map is larger to the selection difficulty of lake group during reducing the staff, and the study of a large amount of knowledge and experiences is needed to accumulate, and
It takes longer.The current automatic selection research for lake element is relatively fewer, and mostly uses the form integrally chosen, it is difficult to simultaneous
Care for attributive character, distribution characteristics and the topological characteristic in lake.Therefore there is an urgent need to study one kind to close under different target engineer's scale
The Algorithms of Selecting that lake group is chosen on reason ground automatically.
Invention content
The object of the present invention is to provide it is a kind of based on Dynamic Multiscale cluster lake automatically selecting method and device, to
It solves existing lake choosing method and is difficult to take into account the attributive character in lake, distribution characteristics and topological characteristic to lead to lake in Map Generalization
Pool is automatic to choose inaccurate problem.
To achieve the above object, the present invention provides it is a kind of based on Dynamic Multiscale cluster lake automatically selecting method,
Include the following steps：
To pending lake group, determines the cluster starting point of lake cluster, establish corresponding cluster figure layer；
It is clustered at a distance from the cluster starting point according to lake, the lake of same cluster is moved into the cluster
Figure layer；
After current cluster, next cluster starting point is determined in remaining lake, corresponding cluster figure layer is established, opens
The cluster of beginning next round；
Finally obtain the corresponding cluster figure layer of each cluster, for each cluster figure layer, according to it includes lake quantity take
Corresponding choosing method chooses one or more than one lake, obtains the first lake figure layer.
Further, further include that area threshold is set according to target proportion ruler to lake group, area of lake is more than
The lake of the area threshold is chosen, and the second lake figure layer is obtained；To the lake in the group of the lake according to setpoint distance
Corresponding buffering area is established, does not have the lake of overlapping relation to choose the buffering area, obtains third lake figure layer, it is described
Buffering area has the lake of overlapping relation to constitute the pending lake group；By first lake figure layer, the second lake figure layer and
Third lake figure layer is merged, and is obtained final lake and is chosen result.
Further, the process of the cluster starting point of the determining lake cluster includes：
The central point in each lake is extracted, then establishing corresponding lake according to the radius set under target proportion ruler buffers
Area；
The quantity of the central point in each lake buffering area is calculated, it is most with the quantity including the central point
The corresponding lake of the lake buffering area as cluster starting point.
Further, include according to lake and the process clustered at a distance from starting point that clusters：
According to target proportion ruler setpoint distance threshold value, determines the nearly lake nearest apart from the cluster starting point, calculate institute
The distance between nearly lake and the cluster starting point are stated, if the distance is less than the distance threshold, and it is poly- less than current
The nearly lake is then moved into the cluster figure layer by the setting multiple of the group average distance in class figure layer between lake, and more
Group average distance in the new current cluster figure layer between lake, then determines next nearest apart from the cluster starting point
Nearly lake；Otherwise terminate to cluster.
Further, for each cluster figure layer, according to it includes lake quantity take corresponding choosing method to choose one
The process in a or more than one lake includes：
The cluster figure layer for only including a lake is all chosen；
If the cluster figure layer includes two lakes, the most short face distance between two lakes is judged, if described
Most short face distance is more than the distance on the spot set under target proportion ruler, then all chooses；Otherwise according to area of lake and lake
It connect the weighted value that quantity calculates each lake with river, the big lake of weighted value is chosen；
If the cluster figure layer includes at least three lakes, it is divided into two kinds of situations：First, if the cluster figure layer
Middle lake quantity is more than setting quantity, then iterates to calculate the weighted value in lake using Principal Component Analysis according to importance factor,
The importance factor includes area of lake, river connection number, domain of influence area, density, centrad and average adjacency；Often
It takes turns iterative process and deletes weighted value ranking inverse setting digit and non-interfering lake, choose quantitative index until meeting and delete
Except quantitative index, described do not interfere with each other refers to not being overlapped identical adjacent lake；Second is that lake quantity in the cluster figure layer
Less than setting quantity, then connect with river according to area of lake and lake quantity calculate each lake weighted value and from greatly to
Small sequence, the lake that digit is set to ranking positive number are chosen.
The present invention also provides a kind of automatic selecting devices in lake based on Dynamic Multiscale cluster, including processor and deposit
Reservoir, the processor are stored with the instruction for realizing following method：
To pending lake group, determines the cluster starting point of lake cluster, establish corresponding cluster figure layer；
It is clustered at a distance from the cluster starting point according to lake, the lake of same cluster is moved into the cluster
Figure layer；
After current cluster, next cluster starting point is determined in remaining lake, corresponding cluster figure layer is established, opens
The cluster of beginning next round；
Finally obtain the corresponding cluster figure layer of each cluster, for each cluster figure layer, according to it includes lake quantity take
Corresponding choosing method chooses one or more than one lake, obtains the first lake figure layer.
Further, further include that area threshold is set according to target proportion ruler to lake group, area of lake is more than
The lake of the area threshold is chosen, and the second lake figure layer is obtained；To the lake in the group of the lake according to setpoint distance
Corresponding buffering area is established, does not have the lake of overlapping relation to choose the buffering area, obtains third lake figure layer, it is described
Buffering area has the lake of overlapping relation to constitute the pending lake group；By first lake figure layer, the second lake figure layer and
Third lake figure layer is merged, and is obtained final lake and is chosen result.
Further, the process of the cluster starting point of the determining lake cluster includes：
The central point in each lake is extracted, then establishing corresponding lake according to the radius set under target proportion ruler buffers
Area；
The quantity of the central point in each lake buffering area is calculated, it is most with the quantity including the central point
The corresponding lake of the lake buffering area as cluster starting point.
Further, include according to lake and the process clustered at a distance from starting point that clusters：
According to target proportion ruler setpoint distance threshold value, determines the nearly lake nearest apart from the cluster starting point, calculate institute
The distance between nearly lake and the cluster starting point are stated, if the distance is less than the distance threshold, and it is poly- less than current
The nearly lake is then moved into the cluster figure layer by the setting multiple of the group average distance in class figure layer between lake, and more
Group average distance in the new current cluster figure layer between lake, then determines next nearest apart from the cluster starting point
Nearly lake；Otherwise terminate to cluster.
Further, for each cluster figure layer, according to it includes lake quantity take corresponding choosing method to choose one
The process in a or more than one lake includes：
The cluster figure layer for only including a lake is all chosen；
If the cluster figure layer includes two lakes, the most short face distance between two lakes is judged, if described
Most short face distance is more than the distance on the spot set under target proportion ruler, then all chooses；Otherwise according to area of lake and lake
It connect the weighted value that quantity calculates each lake with river, the big lake of weighted value is chosen；
If the cluster figure layer includes at least three lakes, it is divided into two kinds of situations：First, if the cluster figure layer
Middle lake quantity is more than setting quantity, then iterates to calculate the weighted value in lake using Principal Component Analysis according to importance factor,
The importance factor includes area of lake, river connection number, domain of influence area, density, centrad and average adjacency；Often
It takes turns iterative process and deletes weighted value ranking inverse setting digit and non-interfering lake, choose quantitative index until meeting and delete
Except quantitative index, described do not interfere with each other refers to not being overlapped identical adjacent lake；Second is that lake quantity in the cluster figure layer
Less than setting quantity, then connect with river according to area of lake and lake quantity calculate each lake weighted value and from greatly to
Small sequence, the lake that digit is set to ranking positive number are chosen.
The beneficial effects of the invention are as follows：Area threshold is set according to target proportion ruler first, all areas is chosen and is more than threshold
The lake of value；Then buffering area is built to all lakes, chooses weight of the lake of no-buffer overlapping relation as distribution characteristics
Want holding point；Next it mines massively to lake and divides region by distribution density difference with Dynamic Multiscale cluster；Finally to not same district
(i.e. small lakes group) takes different Selection Strategies in domain.The present invention effectively maintain lake mass selection under different target engineer's scale take it is front and back
Morphosis and density comparison, choose result have good reasonability, can be applied to landform under different target engineer's scale
The lake element of figure chooses integrated treatment automatically.
Description of the drawings
Fig. 1 is the flow chart of the method for the present invention；
Fig. 2 is the flow chart of Dynamic Multiscale clustering method in the method for the present invention；
Cluster test result when Fig. 3 is present invention cluster under different parameters；
Fig. 4 is original lake group image in embodiment；
Fig. 5 is that the image after the extraction of lake is carried out to original lake group；
Fig. 6 is that area chooses result schematic diagram；
Fig. 7 is that buffering area establishes schematic diagram；
Fig. 8 is that buffering area chooses schematic diagram；
Fig. 9 is Clustering Effect figure；
Figure 10 is the image before the selection of 4 lake of class in cluster；
Figure 11 is the image after the selection of 4 lake of class in cluster；
Figure 12 is the image before the selection of 5 lake of class in cluster；
Figure 13 is the image after the selection of 5 lake of class in cluster；
Figure 14 is the result figure of the choosing method based on area；
Figure 15 is the enlarged drawing of a-quadrant in Figure 14；
Figure 16 is the enlarged drawing of B area in Figure 14；
Figure 17 is the selection result figure of the method for the present invention；
Figure 18 is the enlarged drawing in the regions C in Figure 17；
Figure 19 is the enlarged drawing in the regions D in Figure 17；
Figure 20 is the enlarged drawing in the regions E in Figure 17.
Specific implementation mode
The present invention will be further described in detail below in conjunction with the accompanying drawings.
The purpose of the present invention is selecting large area lake by area selecting step, chosen by buffering area selecting step
" isolated " lake for going out to help to maintain morphosis passes through the region of Dynamic Multiscale clustering difference lake distribution density
Lake is calculated with Principal Component Analysis scientific quantitative analysis and integrates importance overall merit, and the two, which is used in combination, " divides and select to lake group
It ", to solve the automatic On The Choice of lake group；Method is：
Area is chosen, and large area lake is chosen；
Buffering area is chosen, and " isolated " lake is chosen；
It is clustered using Dynamic Multiscale and carries out subregion；
It mines massively to different small lakes and takes different Selection Strategies.
Specific steps are as shown in Figure 1, include following processing step.
(1) area is chosen.Area threshold is set as 1mm on target proportion ruler figure below2The area on the spot represented, traversal are all
Lake, the lake that threshold value is more than to area are chosen.
(2) buffering area is chosen.Buffering area is established to all lakes, buffering area radius is set as 3mm under target proportion ruler and represents
Distance on the spot, there is no the lake of overlapping relation to choose buffering area.
(3) to being chosen except buffering area in addition to all lakes use Dynamic Multiscale clustering method, be divided into multiple small lakes
Group.
(4) it mines massively to different small lakes and takes different Selection Strategies, wherein for including a fairly large number of small lakes in lake
Group, iteration carry out the calculating of importance overall merit using Principal Component Analysis, and the lake that ranks behind is deleted after sequence, until full
Foot chooses quantitative index.
Not comprising the lake chosen by buffering area it can be regarded as to other lakes because they are relatively isolated when cluster
Influence of the lake for not having an impact, and being chosen by area to the lake especially small lakes of surrounding is apparent, thus buffering area
All must include them when choosing and clustering.
The lake image that three kinds of selection modes obtain can finally be merged, the lake repeated is only chosen once i.e.
Can, finally obtain the selection result to original lake group.
Invention herein is using Dynamic Multiscale clustering method by entire lake group according to the different demarcation of lake distribution density
For numerous small lakes groups, and then classification uses different Selection Strategies in part, is maintained step by step from entirety to part
Lake group's morphosis chooses front and back consistency.
Based on analysis above, the basic thought of the Dynamic Multiscale clustering algorithm of step (3) is：According to different target ratio
By map range threshold value d setting, distance threshold D corresponds to longitudinal multiple dimensioned clustered demand to example ruler on the spot, with the average lake of class away from
Identify that distribution density difference corresponds to laterally multiple dimensioned clustered demand from (hereinafter referred group average distance).Cluster is to divide every time
The maximum lake of cloth density by center dynamic expanding around, judges the distance in such nearest lake as starting point successively
Whether it is less than distance threshold D and less than n times of group average distance, if then gathering in nearest lake for one kind, and recalculates class
Average distance；If not then terminating this time to cluster, start cluster next time.D and n is obtained by the effect of Dynamic Multiscale cluster experiment
Best value.Whether D values are more than n times of group average distance, first by whether less than D primarily determining whether cluster, then by small
N times in group average distance finally determines whether to cluster.
Its step is as shown in Fig. 2, include：
(1) it extracts lake center point and establishes buffering area；
(2) the central point quantity in each buffering area face is calculated, as the quantization index value of distribution density, with distribution
The maximum lake of density creates cluster figure layer as cluster starting point；
(3) it is clustered at a distance from all lakes of former lake group's figure layer by calculating cluster all lakes of figure layer nearest
Lake if such is less than the distance threshold D set according to target proportion ruler with a distance from nearest lake, and is less than group average distance
N times when starting point (cluster do not judge the condition), then the nearest lake in the group's figure layer of former lake is moved into the cluster layer, and
Recalculate such group average distance；Otherwise such cluster terminates, return to step 2.
Invention herein uses different Selection Strategies to different classes, and is divided in portion and is chosen according to square root regularity
Quantitative index.After Dynamic Multiscale clustering, lake realm can be divided into 3 kinds according to the quantity in lake in class：Only
There are one " the single-point classes " in lake；Containing there are two " the two point classes " in lake；Containing there are three " the multiple spot classes " in lake or more.It is each
The Selection Strategy of class is as follows：
One, for single-point class, the important holding point of lake group's distribution characteristics is equally regarded as, is all chosen.
Two, for two point class, judge the most short face distance between two lakes, represented if more than 3mm under target proportion ruler
Distance on the spot is then all chosen；Otherwise it connect the weight calculation importance that number respectively accounts for 0.5 by area of lake and with river, deletes
A wherein relatively unessential lake, such as I=0.5*SArea+0.5*NRiver, I expression lake importance, SArea、NRiverPoint
Area of lake connects the value of number with river after Biao Shi not normalizing.
Three, it for multiple spot class, is divided into as two kinds of situations：First, lake quantity is more in class (being greater than 7), then select
Area of lake, river connection number, domain of influence area, density, centrad and average 6 indexs of adjacency as importance because
Son, lake importance overall merit is iterated to calculate using Principal Component Analysis, after often taking turns iterative process deletion importance ranking
30% and non-interfering lake (not interfereing with each other finger without being overlapped identical adjacent lake), until meet choose quantitative index and
Quantity is deleted to touch the mark；Second is that lake quantity is relatively fewer in class (being, for example, less than equal to 7), by area of lake and and river
The weight calculation importance ranking that connection number respectively accounts for 0.5 is chosen.
A specific embodiment is given below.To determine the best value and the multiple dimensioned dynamic of verification of clustering parameter d and n
The validity of cluster carries out cluster experiment to certain lake group, and Fig. 3 is the experimental result under different clustering parameters, and wherein m is lake
The quantity of clustering class is gathered and is indicated with same color for a kind of small lakes group.D=4mm in Fig. 3 (a), n=2, m=13；Fig. 3
(b) d=4mm in, n=2.5, m=11；D=5mm in Fig. 3 (c), n=2, m=8；D=5mm in Fig. 3 (d), n=2.5, m=6；
D=5mm in Fig. 3 (e), n=1.9, m=15；D=6mm in Fig. 3 (f), n=2.5, m=4；Following knot can be obtained by result
By：When n is less than 2, cluster is excessively scattered, as shown in Fig. 3 (e)；When d is more than 5mm, cluster is then excessively concentrated, such as Fig. 3 (f)
It is shown.Experiments verify that the cluster for more meeting human eye vision when d values 4mm-5mm, n value 2-2.5 judges, therefore d values are taken
5mm, n value take 2.5.
With the 1 of somewhere：1000000 water system data instances, are illustrated in figure 4 raw experimental data, and Fig. 5 is the lake after extraction
Element distribution is moored, comprehensive target proportion ruler is set as 1：4000000, the choosing method of the present invention is tested.
Setting area threshold value is 16km2(1：4000000 times 1mm2The area on the spot represented), area selection is carried out, shares 29
Lake is more than threshold value, is all chosen to these lakes, sees that (black indicates that lake is more than area threshold to Fig. 6, and grey represents less than threshold
Value).
Buffering area is built to lake, radius is set as 12km (1：The distance on the spot that 4000000 times 3mm are represented), as shown in fig. 7,
There is 1 lake (in circle) no-buffer intersection situation, therefore this lake is chosen, as shown in figure 8, it is possible thereby to keeping whole
The more apparent distribution characteristics in body lake group southeast orientation.
It is clustered, distance threshold is set as 20km (1：The distance on the spot that 4000000 times 5mm are represented), it obtains result to gather being 5
Class, wherein two point class 2, multiple spot class 3, including lake quantity is respectively 2,2,4,31,86, group average distance is respectively
4514.4,19582.2,10456.5,4736.4,10412.4, specific Clustering Effect is shown in Fig. 9.It can be seen that cluster result meets
The vision cluster impression of human eye, and the difference of all kinds of group average distances can effectively reflect the lake point in each small lakes group region
Cloth density variation.
Each small lakes realm is determined and chooses quantitative index, and Selection Strategy proposed by the invention is respectively adopted.
According to root modelIt is 1 that target proportion ruler, which can be calculated,：4000000 it is all kinds of in answer
The lake quantity of selection, as shown in table 1.
1 all kinds of selection quantitative indexes of table
Lake distance calculating is carried out to class 1, the class 2 of two point class to judge whether to be more than threshold value (1：3mm under 4000000 engineer's scales
The distance on the spot represented), two lakes distance is less than threshold value in class 1, therefore connect with river the weight progress that number respectively accounts for 0.5 with area
Importance calculates, and deletes one of lake；The two lakes distance of class 2 is more than threshold value, therefore is all chosen to two lakes.
To the class 3 of multiple spot class, it is 4 it includes lake quantity, is less than 7, principal component analysis can not be carried out, with area and river
The weight that connection number respectively accounts for 0.5 carries out importance calculating, chooses lake successively by importance ranking and refers to until reaching and choosing quantity
Mark.
White space skeleton line is extracted to the class 4 of multiple spot class, class 5, builds mesh, obtain area of lake, river connection number,
6 domain of influence area, density, centrad and average adjacency attributes, iteration obtain lake importance using principal component analysis
Overall merit is often taken turns iterative process and is only deleted 30% and non-interfering lake after sequence, up to lake quantity as shown in table 2
Reach selection quantitative index.
2 lake importance principal component analysis few examples of table
By 7 wheel iteration, class 4 is completed to choose.By 21 wheel iteration, class 5 is completed to choose.Class 4, the selection effect of class 5 are such as
Shown in Figure 10 to Figure 13, Figure 10 is before class 4 is chosen, and Figure 11 is after class 4 is chosen, and Figure 12 is before class 5 is chosen, and Figure 13 chooses for class 5
Afterwards.
Comprehensive all kinds of selection is as a result, can obtain finally choosing result as shown in figure 17.To prove the method for the present invention
Reasonability, here by using based on area choosing method (as shown in figure 14) and the method for the present invention (as shown in figure 17) carry out pair
Than, circle indicates that the two chooses the difference of result in two width figures, and dashed rectangle A, B, C, D and E indicate local magnification region,
A, the enlarged drawing of B, C, D, E are respectively as shown in Figure 15, Figure 16, Figure 18, Figure 19, Figure 20, and black indicates to choose in partial enlarged view,
Grey indicates to delete.It can be obtained by comparison：
In choosing method of the present invention, in the case where area is not much different, river connection number is more, importance overall merit
Higher (being shown in Table 1), thus selection result preferably remains some positions important (connection river number is more) but area is relatively
Small lake, it is relatively large to delete some areas, but the unessential lake in position, such as Figure 20.On the contrary, the choosing based on area
Method is taken only solely to consider area attribute, when selection does not take the incidence relation with river element into account.This is because of the invention
Method calculates the importance overall merit in lake by Principal Component Analysis, considered area and with river connection relation
The importance factor of two attributes.
Choosing method result of the present invention preferably ensure that lake mass selection takes front and back distribution characteristics that can be maintained, packet
Include the distribution density otherness of whole morphosis similitude and local different zones.In lake, distribution density is more intensive
The degree in region, choice is relatively large, but still maintains the characteristics of region is with respect to dense distribution, such as Figure 19；In lake density
More sparse region, then the degree accepted or rejected is relatively small, but region selection result still seems that distribution is relatively sparse, such as schemes
18.On the contrary, the selection result based on area is then distributed very uneven, the distribution of the lakes of certain areas is excessively intensive, such as Figure 16, and
Some areas are then excessively sparse, such as Figure 15, keep poor to lake group's distribution characteristics.This is because choosing method of the present invention passes through
Buffering area chooses the distribution characteristics that " isolated " lake maintains the more apparent specific position of lake group；It is poly- by Dynamic Multiscale
The region of different distributions density is identified in class, and is divided into each small lakes group, maintains whole distribution characteristics；Each
The inside of a small lakes group, the importance factor for taking reflection distribution characteristics and topological characteristic into account are iterated selection and maintain office
Distribution characteristics in portion region.
Through analysis and summary, the characteristics of lake choosing method of the present invention, is as follows：
1, due to taking Dynamic Multiscale clustering method, by entire lake group according to the different demarcation of lake distribution density
For numerous small lakes groups, and then classification uses different Selection Strategies in part, is maintained step by step from entirety to part
Lake group's morphosis chooses front and back consistency, in addition the selection in " isolated " lake, to keep the distribution of entire lake group special
Good holding is obtained.
2, overall merit is carried out to lake importance as a result of Principal Component Analysis, it is contemplated that can fully reflect attribute
The importance factor of feature, topological characteristic and distribution characteristics makes importance overall merit based on the selection of lake more fully
Science, thus selection result can reflect truth well.
These characteristics so that choosing method of the present invention has followed lake selection principle, considers the importance in lake comprehensively
The factor carries out overall merit, takes front and back morphosis and density to compare lake mass selection and keeps more effective.From experimental result
It sees, the automatic selection in lake can be effectively performed in the present invention.
Specific implementation mode of the present invention is presented above, but the present invention is not limited to described embodiment,
Such as change the concrete numerical value arrived involved in above-mentioned implementation process, or the sequence of different selecting steps is adjusted, it is formed in this way
Technical solution is finely adjusted above-described embodiment to be formed, and this technical solution is still fallen in protection scope of the present invention.
Claims (10)
1. a kind of lake automatically selecting method based on Dynamic Multiscale cluster, which is characterized in that include the following steps：
To pending lake group, determines the cluster starting point of lake cluster, establish corresponding cluster figure layer；
It is clustered at a distance from the cluster starting point according to lake, the lake of same cluster is moved into the dendrogram
Layer；
After current cluster, next cluster starting point is determined in remaining lake, corresponding cluster figure layer is established, under starting
The cluster of one wheel；
Finally obtain the corresponding cluster figure layer of each cluster, for each cluster figure layer, according to it includes lake quantity take correspondence
Choosing method choose one or more than one lake, obtain the first lake figure layer.
2. a kind of lake automatically selecting method based on Dynamic Multiscale cluster according to claim 1, it is characterised in that：
Further include that area threshold is set according to target proportion ruler to lake group, area of lake is more than to the lake of the area threshold
It is chosen, obtains the second lake figure layer；Corresponding buffering area is established according to setpoint distance to the lake in the group of the lake, it will
The buffering area does not have the lake of overlapping relation to be chosen, and obtains third lake figure layer, and the buffering area has overlapping relation
Lake constitutes the pending lake group；First lake figure layer, the second lake figure layer and third lake figure layer are melted
It closes, obtains final lake and choose result.
3. a kind of lake automatically selecting method based on Dynamic Multiscale cluster according to claim 1 or 2, feature exist
In：The process of the cluster starting point of the determining lake cluster includes：
The central point in each lake is extracted, corresponding lake buffering area is then established according to the radius set under target proportion ruler；
The quantity for calculating the central point in each lake buffering area, with the most institute of the quantity including the central point
The corresponding lake of lake buffering area is stated as cluster starting point.
4. a kind of lake automatically selecting method based on Dynamic Multiscale cluster according to claim 3, it is characterised in that：
Include according to lake and the process clustered at a distance from starting point that clusters：
According to target proportion ruler setpoint distance threshold value, determines the nearly lake nearest apart from the cluster starting point, calculate described close
The distance between lake and the cluster starting point if the distance is less than the distance threshold, and are less than current dendrogram
The setting multiple of group average distance in layer between lake, then move into the cluster figure layer, and update institute by the nearly lake
The group average distance between lake in current cluster figure layer is stated, is then determined next apart from nearest close of the cluster starting point
Lake；Otherwise terminate to cluster.
5. a kind of lake automatically selecting method based on Dynamic Multiscale cluster according to claim 4, it is characterised in that：
For each cluster figure layer, according to it includes lake quantity take corresponding choosing method to choose one or more than one lake
The process of pool includes：
The cluster figure layer for only including a lake is all chosen；
If the cluster figure layer includes two lakes, the most short face distance between two lakes is judged, if described most short
Identity distance set under more than target proportion ruler on the spot with a distance from, then all choose；Otherwise according to area of lake and lake and river
Stream connection quantity calculates the weighted value in each lake, and the big lake of weighted value is chosen；
If the cluster figure layer includes at least three lakes, it is divided into two kinds of situations：First, if lake in the cluster figure layer
It moors quantity and is more than setting quantity, then iterate to calculate the weighted value in lake using Principal Component Analysis according to importance factor, it is described
Importance factor includes area of lake, river connection number, domain of influence area, density, centrad and average adjacency；Often wheel changes
Weighted value ranking inverse setting digit and non-interfering lake are deleted for process, until meeting selection quantitative index and deleting number
Figureofmerit, described do not interfere with each other refers to not being overlapped identical adjacent lake；Second is that lake quantity is less than in the cluster figure layer
Quantity is set, then quantity is connect with river according to area of lake and lake and calculates the weighted value in each lake and arrange from big to small
Sequence, the lake that digit is set to ranking positive number are chosen.
6. a kind of automatic selecting device in lake based on Dynamic Multiscale cluster, including processor and memory, which is characterized in that
The processor is stored with the instruction for realizing following method：
To pending lake group, determines the cluster starting point of lake cluster, establish corresponding cluster figure layer；
It is clustered at a distance from the cluster starting point according to lake, the lake of same cluster is moved into the dendrogram
Layer；
After current cluster, next cluster starting point is determined in remaining lake, corresponding cluster figure layer is established, under starting
The cluster of one wheel；
Finally obtain the corresponding cluster figure layer of each cluster, for each cluster figure layer, according to it includes lake quantity take correspondence
Choosing method choose one or more than one lake, obtain the first lake figure layer.
7. a kind of automatic selecting device in lake based on Dynamic Multiscale cluster according to claim 6, it is characterised in that：
Further include that area threshold is set according to target proportion ruler to lake group, area of lake is more than to the lake of the area threshold
It is chosen, obtains the second lake figure layer；Corresponding buffering area is established according to setpoint distance to the lake in the group of the lake, it will
The buffering area does not have the lake of overlapping relation to be chosen, and obtains third lake figure layer, and the buffering area has overlapping relation
Lake constitutes the pending lake group；First lake figure layer, the second lake figure layer and third lake figure layer are melted
It closes, obtains final lake and choose result.
8. a kind of automatic selecting device in lake based on Dynamic Multiscale cluster described according to claim 6 or 7, feature exist
In：The process of the cluster starting point of the determining lake cluster includes：
The central point in each lake is extracted, corresponding lake buffering area is then established according to the radius set under target proportion ruler；
The quantity for calculating the central point in each lake buffering area, with the most institute of the quantity including the central point
The corresponding lake of lake buffering area is stated as cluster starting point.
9. a kind of automatic selecting device in lake based on Dynamic Multiscale cluster according to claim 8, it is characterised in that：
Include according to lake and the process clustered at a distance from starting point that clusters：
According to target proportion ruler setpoint distance threshold value, determines the nearly lake nearest apart from the cluster starting point, calculate described close
The distance between lake and the cluster starting point if the distance is less than the distance threshold, and are less than current dendrogram
The setting multiple of group average distance in layer between lake, then move into the cluster figure layer, and update institute by the nearly lake
The group average distance between lake in current cluster figure layer is stated, is then determined next apart from nearest close of the cluster starting point
Lake；Otherwise terminate to cluster.
10. a kind of automatic selecting device in lake based on Dynamic Multiscale cluster according to claim 9, feature exist
In：For each cluster figure layer, according to it includes lake quantity take corresponding choosing method choose one or more than one
The process in lake include：
The cluster figure layer for only including a lake is all chosen；
If the cluster figure layer includes two lakes, the most short face distance between two lakes is judged, if described most short
Identity distance set under more than target proportion ruler on the spot with a distance from, then all choose；Otherwise according to area of lake and lake and river
Stream connection quantity calculates the weighted value in each lake, and the big lake of weighted value is chosen；
If the cluster figure layer includes at least three lakes, it is divided into two kinds of situations：First, if lake in the cluster figure layer
It moors quantity and is more than setting quantity, then iterate to calculate the weighted value in lake using Principal Component Analysis according to importance factor, it is described
Importance factor includes area of lake, river connection number, domain of influence area, density, centrad and average adjacency；Often wheel changes
Weighted value ranking inverse setting digit and non-interfering lake are deleted for process, until meeting selection quantitative index and deleting number
Figureofmerit, described do not interfere with each other refers to not being overlapped identical adjacent lake；Second is that lake quantity is less than in the cluster figure layer
Quantity is set, then quantity is connect with river according to area of lake and lake and calculates the weighted value in each lake and arrange from big to small
Sequence, the lake that digit is set to ranking positive number are chosen.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810443646.5A CN108615059B (en) | 2018-05-10 | 2018-05-10 | Lake automatic selection method and device based on dynamic multi-scale clustering |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810443646.5A CN108615059B (en) | 2018-05-10 | 2018-05-10 | Lake automatic selection method and device based on dynamic multi-scale clustering |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108615059A true CN108615059A (en) | 2018-10-02 |
| CN108615059B CN108615059B (en) | 2020-08-07 |
Family
ID=63662948
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810443646.5A Active CN108615059B (en) | 2018-05-10 | 2018-05-10 | Lake automatic selection method and device based on dynamic multi-scale clustering |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108615059B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111651501A (en) * | 2020-06-01 | 2020-09-11 | 中南大学 | Spatial aggregation scale selection method for geographic big data |
| CN112802069A (en) * | 2020-12-30 | 2021-05-14 | 北京师范大学 | Method for judging dam bank of shallow lake water-blocking dam |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102063833A (en) * | 2010-12-17 | 2011-05-18 | 中国科学院计算技术研究所 | Method for drawing synchronously displayed symbols and marks of dot map layers of map |
| US20110137903A1 (en) * | 2009-12-04 | 2011-06-09 | University Of South Carolina | Optimization and visual controls for regionalization |
| US20130181993A1 (en) * | 2012-01-13 | 2013-07-18 | Fiona Elizabeth HERRING | Place heat geometries |
| CN103699654A (en) * | 2013-12-26 | 2014-04-02 | 武汉大学 | Cross-measuring-scale vector map water network same name target matching method |
| US20150286667A1 (en) * | 2013-06-27 | 2015-10-08 | Google Inc. | Systems and Methods for Using Listing Proximities to Improve Entity Listings |
| KR101721114B1 (en) * | 2016-06-27 | 2017-03-30 | 서울대학교산학협력단 | Method for Determining the Size of Grid for Clustering on Multi-Scale Web Map Services using Location-Based Point Data |
| CN108009391A (en) * | 2017-05-29 | 2018-05-08 | 兰州交通大学 | A kind of multiple dimensioned lower Grouped point object similarity calculating method |
-
2018
- 2018-05-10 CN CN201810443646.5A patent/CN108615059B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110137903A1 (en) * | 2009-12-04 | 2011-06-09 | University Of South Carolina | Optimization and visual controls for regionalization |
| CN102063833A (en) * | 2010-12-17 | 2011-05-18 | 中国科学院计算技术研究所 | Method for drawing synchronously displayed symbols and marks of dot map layers of map |
| US20130181993A1 (en) * | 2012-01-13 | 2013-07-18 | Fiona Elizabeth HERRING | Place heat geometries |
| US20150286667A1 (en) * | 2013-06-27 | 2015-10-08 | Google Inc. | Systems and Methods for Using Listing Proximities to Improve Entity Listings |
| CN103699654A (en) * | 2013-12-26 | 2014-04-02 | 武汉大学 | Cross-measuring-scale vector map water network same name target matching method |
| KR101721114B1 (en) * | 2016-06-27 | 2017-03-30 | 서울대학교산학협력단 | Method for Determining the Size of Grid for Clustering on Multi-Scale Web Map Services using Location-Based Point Data |
| CN108009391A (en) * | 2017-05-29 | 2018-05-08 | 兰州交通大学 | A kind of multiple dimensioned lower Grouped point object similarity calculating method |
Non-Patent Citations (6)
| Title |
|---|
| GOKGOZ, T等: "A New Algorithm for Cartographic Simplification of Streams and Lakes Using Deviation Angles and Error Bands", 《ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION 》 * |
| SHEN, YL等: "A New Approach to Line Simplification Based on Image Processing: A Case Study of Water Area Boundaries", 《ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION》 * |
| 张志勇: "地图自动综合的研究与分析", 《中国勘察设计》 * |
| 郭沛沛: "信息技术条件下自然要素自动综合及实现方法的研究", 《中国优秀硕士学位论文全文数据库基础科学辑》 * |
| 高朝: "顾及地理特征的复杂形态湖泊地图综合研究", 《中国优秀硕士学位论文全文数据库基础科学辑》 * |
| 黄琦: "湖泊自动综合的研究与实现", 《中国优秀硕士学位论文全文数据库基础科学辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111651501A (en) * | 2020-06-01 | 2020-09-11 | 中南大学 | Spatial aggregation scale selection method for geographic big data |
| CN111651501B (en) * | 2020-06-01 | 2021-05-14 | 中南大学 | Spatial aggregation scale selection method for geographic big data |
| CN112802069A (en) * | 2020-12-30 | 2021-05-14 | 北京师范大学 | Method for judging dam bank of shallow lake water-blocking dam |
| CN112802069B (en) * | 2020-12-30 | 2023-07-04 | 北京师范大学 | Determination method for water-blocking dike cofferdam of shallow lake |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108615059B (en) | 2020-08-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fisher | First experiments in viewshed uncertainty: simulating fuzzy viewsheds | |
| CN103426169B (en) | A kind of dividing method of medical image | |
| CN105894587B (en) | A kind of ridge line and valley route filter method of rule-based constraint | |
| CN104978763B (en) | A kind of simultaneous synthesis map emulation mode of network of waterways key element and DEM based on three-dimensional Douglas Peucker algorithms | |
| CN103236086A (en) | Multiscale DEM (Digital Elevation Model) modeling method giving consideration to contents of surface hydrology | |
| CN108615059A (en) | A kind of lake automatically selecting method and device based on Dynamic Multiscale cluster | |
| CN109636870A (en) | A kind of long and narrow figure spot piecemeal melts method and device | |
| CN112001415A (en) | Location difference privacy protection method based on countermeasure network | |
| CN107240087A (en) | Object detection system and method | |
| CN110415346A (en) | The method for carrying out soil erosion simulation using the three-dimensional cellular automaton of object-oriented | |
| CN109886128A (en) | A kind of method for detecting human face under low resolution | |
| CN102147921A (en) | Graph theory-based Chinese medicinal tongue nature and tongue coat separation algorithm | |
| CN109712161A (en) | A kind of image partition method, device, equipment and storage medium | |
| CN109816749A (en) | Figure spot map symbol fill method, device, computer equipment and storage medium | |
| CN110335352A (en) | A kind of biradical first multiresolution level filtering method of airborne laser radar point cloud | |
| CN109671154A (en) | The curved surface non-iterative gridding method again that triangle gridding indicates | |
| CN110910462B (en) | Point cloud light weight method based on feature calculation and storage medium | |
| CN110738637A (en) | Automatic classification method and system for breast cancer pathological sections | |
| CN111626321A (en) | Image data clustering method and device | |
| CN107945191A (en) | A kind of random Grouped point object Outside contour extraction and target simplify method | |
| CN108962382A (en) | A kind of layering important feature selection method based on breast cancer clinic high dimensional data | |
| CN104715034A (en) | Weighed graph overlapping community discovery method based on central persons | |
| Erwin et al. | Relationship between laminar topology and retinotopy in the rhesus lateral geniculate nucleus: results from a functional atlas | |
| CN104766367B (en) | Three-dimensional gridding topology structure chart construction method in computing three-dimensional model processing | |
| CN106373192A (en) | Non-topology-consistency three-dimensional grid block tracking algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |