CN106548141A - A kind of object-oriented farmland information extraction method based on the triangulation network - Google Patents
A kind of object-oriented farmland information extraction method based on the triangulation network Download PDFInfo
- Publication number
- CN106548141A CN106548141A CN201610934973.1A CN201610934973A CN106548141A CN 106548141 A CN106548141 A CN 106548141A CN 201610934973 A CN201610934973 A CN 201610934973A CN 106548141 A CN106548141 A CN 106548141A
- Authority
- CN
- China
- Prior art keywords
- triangulation network
- cluster
- central point
- oriented
- farmland information
- 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
- 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
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Abstract
The present invention relates to a kind of object-oriented farmland information extraction method based on the triangulation network, its step is：High spatial resolution image is split using multi-scale division method；Reject strip cutting object（Road, irrigation canals and ditches etc.）；Extract the central point of surplus division object；The triangulation network is built using central point；Skinning operations are carried out to the triangulation network；The triangulation network is built using AUTOCLUST clustering algorithms, and is clustered；Using V constraint diagrams, optimize cluster result, it is to avoid excessively cluster and owe cluster；Constrained using maximum variance, reject remaining fragmentary forest land；Obtain plant extraction result and carry out accuracy evaluation.Instant invention overcomes high spatial resolution remote sense image data volume is big, process difficult problem, the semantic information for making full use of the object after segmentation to provide, the object for interfering is extracted to farmland information by rejecting settlement place and road etc., so as to efficiently carry out automatically extracting for farmland information, and ensure the globality of plant extraction.
Description
Technical field
The present invention relates to a kind of farmland information extraction method, more particularly to a kind of OO efficiently non-supervisory
Farmland information extracting method.
Background technology
Automatically extracting for farmland information has vital work for China rural holding resource management and the reallocation of land
With.Currently, the related art method that farmland information charts is mainly based on remote sensing technology.On different space scales,
There are many researchs extracted with regard to farmland information, be broadly divided into local, area and Global Regional.The research of regional area is mainly related to
And single area or irrigated area basin or specific region；Geographic coverage mainly includes large river basins and continental region；And it is global
Yardstick then relates generally to the farmland information drawing of global range.Under different spaces yardstick, in order to reach optimal farmland information
Extraction effect, different sensing datas are needed using different drawing modes, are summarized according to the achievement in research of many scholars,
The farmland information of regional area is extracted using Landsat TM/ETM+, SPOT, LISS, ASTER, CBERS data can reach compared with
Good cost performance, information extracting method relate generally to photo reading, image digital computing, image classification, segmentation, visual fusion.
Geographic coverage mainly adopts Landsat TM/ETM+, MODIS, MERIS, AVHRR, SPOT VGT data, by time serieses point
Analysis method, supervision or non-supervised classification, masking method identification farmland information.Global Scale mainly adopts MODIS, MERIS,
AVHRR, SPOT VGT data, information extracting method relates generally to Non-surveillance clustering, for the machine learning of time series data
Algorithm, and using other assistance datas (including statistical data, ground real time data etc.).
The spatial resolution for studying the remote sensing image for using above is relatively low, when identification small range and vital area arable land it is past
Past relatively difficult, this is accomplished by the image data for being aided with higher resolution.Especially, China starts from the high-resolution of Eleventh Five-Year Plan
The special enforcement of earth observation systems so that high spatial resolution image data (<10m) acquiring technology is further developed.
However, as high spatial resolution image spectral information more enriches, foreign body is serious with spectrum phenomenon, it is difficult to based on pixel
Sorting technique extracts arable land.In addition, the difference of the different phenologys in arable land, for example, in unmanned plane high score image, nonirrigated farmland and paddy field
Color differs greatly, and nonirrigated farmland is yellow, and paddy field is green, it is difficult to both are extracted simultaneously.Therefore, the arable land in high score image is special
The extraction difficulty of topic information is bigger.
For the arable land Extracting Thematic Information of high score image, existing many scholars have been correlational study work, Liu etc. 2008
Year writes articles " High qual ity prime farmland extract ion in " GIS and architectural environment joint conference "
Pattern based on object-oriented image analys is ", using aviation image data (3m), by side
Edge detection segmentation image, subsequently calculates characteristics of objects value, finally builds farmland information using C4.5 and extracts decision tree, realizes high-quality
The detection in amount standard farmland.Lu etc. 2007 exists《Internat ional Journal of Remote Sens ing》Write articles
“Comparison between several feature extract ion/classification methods for
mapping compl icated agricultural land use patches us ing airborne
Hyperspectral data ", extract complicated agricultural land overlay area using Airborne Hyperspectral data (2m).Duro etc. 2012
Nian《Remote Sens ing of Environment》Write articles " A comparison of pixel-based and
object-based image analys is with selected machine learning algorithms for
The class ificat ion of agricultural landscapes using SPOT-5HRG imagery ", use
SPOT images, using multi-scale division technology, are utilized respectively the supervised classification methods such as random gloomy, support vector machine, decision tree, real
The extraction of existing farmland information.It is such based on the object after segmentation, the method for extracting farmland information using measure of supervision is also
A lot, but high score image provides more abundant spectrum and texture feature information, above method mostly is secret operation, not
The characteristics of making full use of ground block message itself in arable land exclusive, causes extraction effect bad.
At present, it is specifically designed for the non-supervisory farmland information extracting method of high score image development seldom, automaticity is not
Height, utilizes and insufficient to the semantic information of the geographic object after segmentation.For example, Sun and Xu 2009 are in " Transact
Ions of the CSAE " write articles " Comer extraction algorithm for high-resolut ion imagery
Of agricultural land ", using for Quickbird panchromatic wave-band images, first split to image data, with reference to
The shape information (arable land profile rule, general with the corner characteristics point of 4 or more than 4) provided by corner characteristics point extracts segmentation
The farming land information of the regular shape such as rear arable land and swag.The research is that the extraction of high resolution spatial panchromatic image farmland information is opened up
One new thinking, to a certain extent using the geometry and textural characteristics in plot of ploughing in high resolution image data, and
The peculiar property in arable land plot enters line algorithm design, so as to extraction farmland information rapidly and efficiently.But which is merely with single
Shape facility, the extraction effect to complex environment is simultaneously bad.Have in view of farmland information and significantly concentrate feature in flakes, and
Due to the arable land object after high score Image Segmentation it is larger, and for concentrating settlement place region, due to internal atural object it is abundant, including room
Room, forest land, road, concrete floor etc., the cutting object for coming in every shape, relative arable land region, the cut zone look ratio
Relatively crush.Based on this feature, it is contemplated that the atural objects such as broken settlement place, forest land are rejected by the method for cluster, so as to carry automatically
Take farmland information.
The content of the invention
The technical problem to be solved in the present invention is：Overcome the above-mentioned deficiency of prior art, there is provided a kind of based on the triangulation network
Object-oriented farmland information extraction method, it is poly- which combines object-oriented classification, triangulation network analysis and AUTOCLUST
Class algorithm, based on high score remote sensing image, automatically, efficiently and accurately carries out the extraction of farmland information.
In order to solve above technical problem, a kind of object-oriented farmland information based on the triangulation network that the present invention is provided is automatic
Extracting method, comprises the following steps：
Step 1, high spatial resolution remote sense image is split using multi-scale division method；
Step 2, rejecting strip cutting object；
Step 3, the central point for extracting surplus division object；
Step 4, using central point build Delaunay triangulation network；
Step 5, the Delaunay triangulation network to step 4 acquisition carry out skinning operations, peel off the outer of Delaunay triangulation network
Triangle；
Step 6, the Delaunay triangulation network after peeling in step 5 is clustered, obtain triangulation network Node distribution intensive
Two kinds of several cluster results sparse with distribution；
Step 7, cluster result is modified using Vornoio constraint diagrams, modification method is as follows：Using Delaunay tri-
Angle net node builds Voronoi diagram, calculates Voronoi area of a polygon A in Voronoi diagram, and Voronoi area of a polygon is less than
During lower threshold, primarily determine that the corresponding central point of Voronoi polygons belongs to broken region, Voronoi area of a polygon
During more than upper limit threshold, the corresponding central point of the polygon is primarily determined that for arable land region；The center in broken region will be judged to
Point is left out or is included in the intensive cluster of Node distribution, and the central point in the region that is judged to plough is included into sparse poly- of Node distribution
Apoplexy due to endogenous wind；
Central point corresponding to the densely distributed cluster of step 8, deletion of node；
If the maximum spectral variance Max.diff of cutting object is obtained after the segmentation of step 9, the step 1 more than default threshold
Value, then judge all centers that the cutting object, as forest land, rejects the central point corresponding to the cutting object, finally remains
The corresponding cutting object of point is arable land, obtains plant extraction result.
The present invention also has feature further below：
1st, in step 1, the segmentation carried out to high spatial resolution image repeatedly using image analysis software is attempted, and is observed
The distribution characteristicss of cutting object, select to carry out Image Segmentation to plot segmentation preferably segmentation yardstick of ploughing.
2nd, cutting object of the rectangle fitting degree character numerical value less than 0.5 is judged to strip cutting object, and rejects.
3rd, the area of a polygon of surplus division object in step 3, is calculated, and according to the coordinate of areal calculation central point, institute
Central point is stated for the polygonal barycenter of cutting object.
4th, in step 5, chosen distance parameter d deletes side of the Delaunay triangulation network intermediate cam shape length of side more than d, and d's takes
Value scope is [100m, 150m].
5th, in step 7, lower threshold isUpper limit threshold isWherein,
Voronio_mean is the meansigma methodss of the area of the Thiessen polygon for generating,WithFor coefficient, and
6、
7th, before step 7 or step 8 are performed, make the following judgment, if j-th cluster meets
The cluster is included into into the sparse cluster of Node distribution then, whereinRepresent the gore of j-th cluster
Long-pending average,The area mean of mean of the triangle sets of all clusters is represented,
Represent the variance of the triangle area set of j-th cluster.
8th, predetermined threshold value is the maximum spectrum of the maximum spectral variance average with all cutting objects of 2 times of all cutting objects
Variance criterion difference sum.
9th, clustered using AUTOCLUST methods in step 6.
It is right that the present invention is realized with reference to AUTOCLUST clustering algorithms and various optimization constrained procedures based on Delaunay triangulation network
Farmland information is automatically extracted.The triangulation network and clustering algorithm can be tied on the basis of object-oriented classification by the present invention
Close, automatically and accurately therefrom choose reliable farmland information, meanwhile, in simplifying farmland information extraction process, man-machine interaction is strong
Degree, it is ensured that the globality of plant extraction.
Image division technology used in the step 1 of the present invention, using high spatial resolution image, is attempted by segmentation repeatedly
And observe the distribution characteristicss of cutting object, it is found that farmland information has, the arable land block after segmentation is all
Than larger, and due to the high-resolution feature of low latitude unmanned plane image, for settlement place region is concentrated, its internal atural object enriches,
Including house, forest land, road, concrete floor etc., the cutting object for coming in every shape after segmentation, is formed, relative arable land region, the segmentation
Region looks that comparison is crushed.On the other hand, for Chinese vast rural area, settlement place region is distributed in checkerboard in arable land,
Usually, centralized residence settlement place and road area are the main interference factors that farmland information is extracted, and road and irrigation canals and ditches etc. are right
As with obvious strip feature, not being inconsistent with farmland information, therefore this class object is directly rejected in step 2.
Further, for the atural object of non-strip, the present invention analyzes the interference factor to bare place based on the triangulation network and enters
Row is progressively rejected, and is introduced into AUTOCLUST clusters, Voronio constraint diagrams and Variance Constraints in this process, and various clusters are about
Beam algorithm improves the precision of plant extraction step by step.
It can be seen that, the present invention, based on the triangulation network, is main extraction algorithm using AUTOCLUST clusters, using long minor face
Cluster realizes the transmission of clustering information, specifically, characterizes long minor face information using local and global variance, deletes in the triangulation network
Minor face set and long line set further expanded scope.
Compared to existing technology, the present invention fully combines the characteristic information after the Remote Sensing Image Segmentation of arable land and the triangulation network point
Analysis, defines a kind of automatic, efficient, accurate farmland information extracting method.Concrete innovative point and have the beneficial effect that：
First, the present invention can make full use of the variable density that can adapt to automatically zones of different point of Delaunay triangulation network
Feature, with reference to the AUTOCLUST algorithms based on graph theory, realize the transmission of clustering information using long minor face cluster, specifically, adopt
Long minor face information is characterized with local and global variance such that it is able to the cluster of the triangulation network is realized using the method.
Second, the present invention on the basis of AUTOCLUST algorithms constrains further process using Thiessen polygon and is ignored
The feature of triangulation network intermediate cam shape area, obtains its dual graph Voronoi diagram according to Delaunay triangulation network, further by meter
The method for calculating Thiessen polygon area, characterizes the distribution density of point, and owing for occurring when solving using AUTOCLUST clustering algorithms is poly-
The phenomenon of class；Using cluster triangle web area average value constraint, the mistake cluster for solving the larger triangulation network in AUTOCLUST clusters is asked
Topic；Constrained using maximum variance, further optimized algorithm, reject the fragmentary forest land in plains region.
To sum up, the present invention fully excavates the semantic information of high score remote sensing image, it is proposed that one kind is tied based on the triangulation network
The plant extraction method of AUTOCLUST clusters is closed, the extraction of farmland information automatically, is efficiently and accurately carried out.Whole method, from
Dynamicization degree is higher, precision is higher, and stability is higher.The farmland information extraction efficiency of high score remote sensing image is not only increased, together
When, it is ensured that the globality of plant extraction.
Description of the drawings
The present invention is further illustrated below in conjunction with the accompanying drawings.
Fig. 1 is the inventive method flow chart.
Fig. 2 a are that unmanned plane tests raw video.
Fig. 2 b are the segmentation result that unmanned plane tests that raw video segmentation yardstick is 50.
Fig. 2 c are the segmentation result that unmanned plane tests that raw video segmentation yardstick is 100.
Fig. 2 d are the segmentation result that unmanned plane tests that raw video segmentation yardstick is 180.
Fig. 3 a are Delaunay triangulation network schematic diagram.
Fig. 3 b are the outer triangle exemplary plot for peeling off Delaunay triangulation network.
Fig. 4 a are the triangulation network that Fig. 2 d segmentation results build.
Fig. 4 b are AUTOCLUST cluster results.
Fig. 4 c are the dendrogram of Voronio figures lower threshold constraint.
Fig. 4 d are the dendrogram of Voronio figures upper limit threshold constraint.
Fig. 5 a are the plant extraction result using the method for proposing.
Fig. 5 b are to carry out the result that plant extraction is obtained using image analysis software eCognition.
Specific embodiment
The present invention is described in detail below according to accompanying drawing, the purpose of the present invention and effect will be apparent from.
It is illustrated in figure 1 flow process of the embodiment of the present invention based on the object-oriented farmland information extraction method of the triangulation network
Figure, present implementation are comprised the following steps：
Step 1, high spatial resolution image is split using multi-scale division method.In the present embodiment, use
ECognit ion carry out segmentation repeatedly and attempt to high spatial resolution image, and observe the distribution characteristicss of cutting object, choosing
Selecting carries out Image Segmentation to plot segmentation preferably segmentation yardstick of ploughing.It is proposed that the span of segmentation yardstick is
[150,200], to project the difference of the settlement place in high score image and the corresponding cutting object in arable land.
Fig. 2 is that three scape unmanned planes test image.In this step, carried out using image analysis software eCognition multiple dimensioned
Segmentation, partitioning parameters Shape are 0.5 for 0.4, Smoothness.Respectively implement three yardsticks segmentation (include yardstick 50,
100th, 180), by the segmentation result of observation different scale, select to carry out subsequent analysis to the preferable yardstick of plot segmentation of ploughing.
Easily find, with the increase of segmentation yardstick, the area of cutting object gradually increases.Opposite segment yardstick 50 and 100, works as segmentation
Yardstick at 180, the arable land plot that more preferably can will have same pixel feature in the unmanned plane image data of 0.2 meter of resolution
Split.
Fig. 2 a are raw video, and Fig. 2 b are to split the segmentation result that yardstick is 50, and Fig. 2 c are to split the segmentation that yardstick is 100
As a result, Fig. 2 d are to split the segmentation result that yardstick is 180.
Step 2, the cutting object (road, irrigation canals and ditches etc.) for rejecting elongate in shape.In the present embodiment, by rectangle fitting degree feature
Cutting object of (Rectangular fit) numerical value less than 0.5 is rejected, and rectangle fitting degree feature (Rectangular fit) can
For rejecting the strip cutting object of road, irrigation canals and ditches etc. and arable land ground block feature notable difference.
After successfully splitting to unmanned plane image data, broken settlement place region will be rejected using triangulation network cluster, with
Extract farmland information.Central point (step 3), the central point number and the region segmentation pair of extraction are extracted first with cutting object
As number it is identical.Although the distribution of Fig. 2 d arable lands is relatively concentrated, it is mingled with the atural objects such as road, irrigation canals and ditches, ridge unavoidably, in high-resolution
In the image of rate, these atural objects can be divided out, and they belong to elongate in shape atural object, have larger with the rectangle in arable land
Difference.In the central point that step 3 is extracted, the central point of these atural objects more, they can affect method precision, be raising method
Reliability, before the triangulation network is built, needs to reject these strip atural objects.So, first by space characteristics rectangle fitting
Degree (Rectangular fit) threshold value rejects strip atural object (Rectangular fit<0.5) corresponding central point.
Step 3, the central point for extracting surplus division object.In this example, the polygon facet of surplus division object is calculated first
Product, and according to the coordinate of its central point of areal calculation.
To calculate the polygonal central point (center of gravity or barycenter) of cutting object, polygonal area is calculated first, if certain
The cut-off rule of individual cutting object has N number of node (xi,yi) composition, wherein 0≤i≤N-1.Assume last node and first
Node is identical, then polygon Shape closed.So, polygonal areal calculation formula is expressed as：
Wherein xN=x0, then it represents that polygon Shape closed.After area is calculated, polygon central point is (here actually
Represent center of gravity or barycenter) coordinate (Cx, Cy) can be calculated by following formula：
Step 4, the central point structure Delaunay triangulation network extracted using step 3, Delaunay triangulation network such as Fig. 3 a institutes
Show.
Delaunay triangulation network is the triangle sets for not overlapping each other and adjoining each other, and need to typically meet two criterions：
There are no other points in the circumscribed circle of each triangle in Delaunay triangulation network.In all possible triangulation network,
The minimum angle of each triangle in Delaunay triangulation network is typically maximum.According to features above, many researchs and proposes different three
Angle net construction method, popular method include：Divide and conquer, scan-line algorithm, delta algorithm, Fast incremental construction algorithm, convex closure
Algorithm.Emphasis of the present invention does not lie in triangulation network developing algorithm, and the present embodiment is using the ripe algorithm (Divide-and- that divides and rules
Conquer Algorithm) Delaunay triangulation network is built, first, point set is divided as far as possible to sufficiently small subset, to different sons
Collection builds the sub- triangulation network, subsequently integrates each sub- triangulation network, finally, ensures that the triangulation network for building is using local optimum method
Delaunay triangulation network, the advantage for so building the triangulation network is that time efficiency is higher, has the disadvantage that the recursive operation amount of needs is big, right
The requirement of memory headroom is higher.
Step 5, the Delaunay triangulation network to step 4 acquisition carry out skinning operations, peel off the outer of Delaunay triangulation network
Triangle.In this example, using the Delaunay triangulation network obtained in step 4, chosen distance parameter d, distance parameter d should be not less than
Longest edge inside the triangulation network, it is proposed that the span of d is [100m, 150m], deletes side of the triangle edges length more than d,
And using the side for only associating with a triangle as point group border.It is peel off Delaunay triangulation network outer three as shown in Figure 3 b
The result at angle.
Point group distribution belongs to probabilistic problem.In the research of GIS, it is commonly used for substituting point group distribution.
But, when the distribution of point group it is slim and sexy, then may be in convex hull comprising the recess areas much not existed a little, so formation
Convex hull can not really represent the distribution of point group.Usually, based on the sorting procedure of Delaunay triangulation network it is：①
The triangulation network, such as Fig. 3 a are built using point group；2. a distance parameter d is selected, " peeling " to the triangulation network is realized by parameter d,
Delete side of the triangle edges length more than d；3. the border as point group is extracted on the side associated with a triangle.Fig. 3 b show,
Operated by " peeling ", delete the very big triangle of the length of side so that the triangulation network after " peeling " is closer to real point group point
Cloth.
Step 6, AUTOCLUST clusters are carried out to the Delaunay triangulation network after peeling in step 5, obtain triangulation network node
Two kinds of several cluster results densely distributed and that distribution is sparse.AUTOCLUST clustering algorithms are existing ripe algorithm,
See Estivill-Castro and Lee to be published in 2002《Computers,Environment and urban
systems》On document " Argument free clustering for large spat ial point data sets
via boundary extract ion from Delaunay Diagram”.As in Fig. 4 a, tetra- pieces of regions of A, B, C, D, E, F are
High density clusters position.
In addition to using AUTOCLUST clustering methods, it is also possible to using K-means, density-based spatial clustering etc.
Method is clustered to center point, obtains sparse two kinds of several clusters of densely distributed and distribution.
Step 7, AUTOCLUST cluster results are modified using Vornoio constraint diagrams, it is to avoid excessively cluster and owe poly-
Class, as shown in Figure 4 b, represents the residential area region not being clustered in black tetragon and triangle in figure, that is, owes cluster.This reality
Apply in example, Voronoi diagram is built using Delaunay triangulation network node (central point for namely obtaining in the 4th step), calculate
Voronoi area of a polygon A in Voronoi diagram, when Voronoi area of a polygon is more than upper limit threshold, primarily determines that this is polygon
The corresponding central point of shape is arable land, when Voronoi area of a polygon is less than lower threshold, to primarily determine that the Voronoi polygons
Corresponding central point belongs to broken region (bare place), leaves out or be included into intensive poly- of Node distribution by the central point in broken region
Apoplexy due to endogenous wind.Complete the amendment to cluster result, it is to avoid excessively cluster and owe cluster phenomenon.
As illustrated in fig. 4 c, it is the result of entering row constraint using lower threshold, the thicker Voronoi polygons of lines are in figure
Area is primarily determined that as bare place less than the polygon of lower limit.As shown in figure 4d, it is knot that practical upper limit threshold value enters row constraint
Really, in figure, the thicker Voronoi polygons of lines are polygon of the area more than lower limit, are primarily determined that as arable land.
Voronoi diagram is built based on Delaunay triangulation network, area of a polygon is calculated, is expressed as A.So, Voronoi is more
The dot density of side shape can be characterized by 1/A.Usually, the value is less, and point population density is less, and the value is bigger, represents that point group is close
Degree is bigger.The area-constrained strategy of Thiessen polygon：Usually, when AUTOCLUST is clustered, many regions occur owing cluster
Phenomenon, then when row constraint is entered using Thiessen polygon, SC service ceiling threshold value and lower threshold comprehensive constraint, usually,
When Thiessen polygon area is more than upper limit threshold, the corresponding central point of the polygon must be arable land, when Thiessen polygon face
When product is less than lower threshold, the corresponding central point of the polygon is necessarily belonging to broken region.Here bound threshold value is tried one's best
Arrange loose.High density point region and low-density point region are filtered with below equation：
Here Voronio_mean is the meansigma methodss of the area of the Thiessen polygon for generating,WithFor coefficient, andIt is apparent that coefficient here determines cluster degree a little, by experiment, in order to avoid excessively clustering and owing cluster,
ArrangeAs general value, AUTOCLUST cluster results are modified.
During AUTOCLUST is clustered, some clusters are main also to include the larger triangulation network, and its area relatively surrounding
Triangle web area is also big, and this is possibly due to the corresponding central point of two or three of only a few arable land cutting objects apart from close, and
Apart from other adjacent point farther out, this point is frequently not the central point of resident's region segmentation object.Triangle in this cluster
Net number is often few, and the meansigma methodss of triangle area are bigger very than needing the meansigma methodss of the triangle web area in the broken region of identification
It is many.
So being further introduced into clustering the concept of triangle web area average removing this kind of mistake in AUTOCLUST cluster process
Cluster by mistake, equation is as follows：
WhereinRepresent the average of the triangle area of j-th cluster, TiFor i-th triangle
Area, nj are the triangle number of j-th cluster,Represent all clusters triangle area average it is average
Value,The variance of the triangle area set of j-th cluster is represented, m represents the number of cluster, and i is represented
Triangle sequence number in j-th cluster.
If j-th cluster meetsThen should
Cluster is included in the sparse cluster of distribution.
Central point corresponding to the densely distributed cluster of step 8, deletion of node, so far completes the center in bare place region
The rejecting of point.
Step 9, constrained using maximum variance, reject remaining fragmentary forest land, if being split after the step 1 segmentation
The maximum spectral variance Max.diff of object is more than predetermined threshold value, then judge that the cutting object, as forest land, rejects the cutting object
Corresponding central point, the cutting object corresponding to all central points for finally remaining are arable land, obtain plant extraction knot
Fruit simultaneously carries out accuracy evaluation.In the present embodiment, the predetermined threshold value is 2 times of maximum spectral variance averages and maximum spectral variance mark
Quasi- difference sum.
Although the method that the present embodiment is proposed can cluster settlement place aggregation zone, and be rejected, for Plain
, also sporadicly there are many forest lands in area, as the textural characteristics in forest land are more or less the same, corresponding cutting object is often larger, very
Hardly possible triangulation network clustering method is rejected.And the coarse texture in forest land, variance is larger, ploughs relatively smooth, and variance is less, institute
To propose that the method for Variance Constraints is pruned to extract result to farmland information here.The present embodiment is using 2 times of maximum spectrum
Mean variance is identified as threshold value with maximum spectral variance standard deviation sum.Such as following formula：
Max.diff>2×mean_Max.diff+Standard Deviation_Max.diff (9)
Wherein mean_Max.diff represents the average of the maximum spectral variance feature of all objects in segmentation figure layer；Feature
The standard deviation of Max.diff is expressed as Standard Deviation_Max.diff.
Fig. 5 a be using based on the triangulation network object-oriented farmland information extraction method extract farmland information, Fig. 5 a
In, white represents the arable land extracted, and black represents the bare place of identification, and Fig. 5 b are represented and supervised using eCognit ion softwares
The farmland information that classification is extracted is superintended and directed, white is arable land in figure, and black is bare place.Relatively find：Extracting method of the present invention is being protected
On the premise of card required precision, the concentration characteristic in flakes ploughed can not only be kept on the whole, moreover it is possible to keep away to a certain extent
Exempt from due to sowing type difference (phenology difference), caused fragmentary wrong point of phenomenon.Simultaneously, it is to avoid supervised classification selects training sample
Process, improve automatic business processing degree.
In addition to the implementation, the present invention can also have other embodiment.All employing cutting object central point clusters etc.
The farmland information extractive technique scheme of form, all falls within the protection domain of application claims.
Claims (10)
1. a kind of object-oriented farmland information extraction method based on the triangulation network, comprises the following steps：
Step 1, high spatial resolution image is split using multi-scale division method；
Step 2, rejecting strip cutting object；
Step 3, the central point for extracting surplus division object；
Step 4, using central point build Delaunay triangulation network；
Step 5, the Delaunay triangulation network to step 4 acquisition carry out skinning operations, peel off the outer triangle of Delaunay triangulation network；
Step 6, the Delaunay triangulation network after peeling in step 5 is clustered, obtain triangulation network Node distribution it is intensive and point
Sparse two kinds of several cluster results of cloth；
Step 7, cluster result is modified using Vornoio constraint diagrams, modification method is as follows：Using Delaunay triangulation network
Node builds Voronoi diagram, calculates Voronoi area of a polygon in Voronoi diagram, and Voronoi area of a polygon is less than lower limit
During threshold value, primarily determine that the corresponding central point of Voronoi polygons belongs to broken region, Voronoi area of a polygon is more than
During upper limit threshold, the corresponding central point of the polygon is primarily determined that for arable land region；The central point for being judged to broken region is deleted
Go or be included in the intensive cluster of Node distribution, the central point for being judged to arable land region is included into into the sparse cluster of Node distribution
In；
Central point corresponding to the densely distributed cluster of step 8, deletion of node；
If the maximum spectral variance Max.diff of cutting object is obtained after the segmentation of step 9, the step 1 more than predetermined threshold value,
Judge that the cutting object, as forest land, rejects the central point corresponding to the cutting object, all central point institutes for finally remaining
Corresponding cutting object is arable land, obtains plant extraction result.
2. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
In step 1, the segmentation carried out to high spatial resolution image repeatedly using image analysis software is attempted, and observes cutting object
Distribution characteristicss, select to carry out Image Segmentation to plot segmentation preferably segmentation yardstick of ploughing.
3. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
Cutting object of the rectangle fitting degree character numerical value less than 0.5 is judged to strip cutting object, and rejects.
4. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
The area of a polygon of surplus division object in step 3, is calculated, and according to the coordinate of areal calculation central point, the central point is
The polygonal barycenter of cutting object.
5. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
In step 5, chosen distance parameter d deletes side of the Delaunay triangulation network intermediate cam shape length of side more than d, and the span of d is
[100m,150m]。
6. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
In step 7, lower threshold isUpper limit threshold isWherein, Voronio_
Mean is the meansigma methodss of the area of the Thiessen polygon for generating,WithFor coefficient, and
7. the object-oriented farmland information extraction method based on the triangulation network according to claim 6, it is characterised in that：
8. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
Before step 7 or step 8 are performed, make the following judgment, if j-th cluster meets
The cluster is included into into the sparse cluster of Node distribution then, whereinRepresent the triangle area of j-th cluster
Average,The area mean of mean of the triangle sets of all clusters is represented,
Represent the variance of the triangle area set of j-th cluster.
9. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, it is characterised in that：
Predetermined threshold value is the maximum spectral variance standard deviation of the maximum spectral variance average with all cutting objects of 2 times of all cutting objects
Sum.
10. the object-oriented farmland information extraction method based on the triangulation network according to claim 1, its feature exist
In：Automatic cluster is carried out using AUTOCLUST methods in step 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610934973.1A CN106548141B (en) | 2016-11-01 | 2016-11-01 | A kind of object-oriented farmland information extraction method based on the triangulation network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610934973.1A CN106548141B (en) | 2016-11-01 | 2016-11-01 | A kind of object-oriented farmland information extraction method based on the triangulation network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106548141A true CN106548141A (en) | 2017-03-29 |
| CN106548141B CN106548141B (en) | 2019-07-26 |
Family
ID=58392223
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610934973.1A Active CN106548141B (en) | 2016-11-01 | 2016-11-01 | A kind of object-oriented farmland information extraction method based on the triangulation network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106548141B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107221001A (en) * | 2017-05-18 | 2017-09-29 | 苏州恩巨网络有限公司 | A kind of area information acquisition methods and device |
| CN107564077A (en) * | 2017-09-04 | 2018-01-09 | 济南浪潮高新科技投资发展有限公司 | A kind of method and device for drawing road network |
| CN107705002A (en) * | 2017-09-21 | 2018-02-16 | 中国矿业大学(北京) | The determination method of mining soil content of beary metal sampled point exception high level coverage |
| CN107766496A (en) * | 2017-10-19 | 2018-03-06 | 大连大学 | Based on MapReduce and the extensive NN Query system for arranging Thiessen polygon |
| CN107766495A (en) * | 2017-10-19 | 2018-03-06 | 大连大学 | Distributed temporal index method based on Voronoi diagram |
| CN107958087A (en) * | 2017-10-19 | 2018-04-24 | 大连大学 | The representative point acquisition methods of the distributed temporal index of the row's of falling Thiessen polygon of portable medical |
| CN108021690A (en) * | 2017-10-19 | 2018-05-11 | 大连大学 | Arrange Voronoi indexes |
| CN108961403A (en) * | 2018-06-22 | 2018-12-07 | 武汉大学 | A method of major trunk roads are extracted with open source street map |
| CN111768505A (en) * | 2020-07-01 | 2020-10-13 | 江西师范大学 | Plane point set shape reconstruction method and device and electronic equipment |
| CN112330700A (en) * | 2020-11-16 | 2021-02-05 | 四川航天神坤科技有限公司 | Cultivated land plot extraction method of satellite image |
| CN112598724A (en) * | 2021-03-01 | 2021-04-02 | 武大吉奥信息技术有限公司 | Improved TIN-based planar element center line extraction method |
| CN112926029A (en) * | 2021-01-29 | 2021-06-08 | 四川省环境政策研究与规划院 | Residential area identification and division method for rural domestic sewage treatment |
| CN114354464A (en) * | 2021-12-22 | 2022-04-15 | 南京大学 | Method for quantitatively analyzing similarity between hyperspectral libraries of different metal nanoparticles |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090315882A1 (en) * | 2008-06-19 | 2009-12-24 | Xerox Corporation | Reducing the number of vertices in a clustered delaunay mesh |
| CN102073981A (en) * | 2011-02-23 | 2011-05-25 | 中国测绘科学研究院 | Point group geographic entity selection method under the restriction of correlated elements |
| CN104036024A (en) * | 2014-06-27 | 2014-09-10 | 浙江大学 | Spatial clustering method based on GACUC (greedy agglomerate category utility clustering) and Delaunay triangulation network |
| JP2016062170A (en) * | 2014-09-16 | 2016-04-25 | 三菱製紙株式会社 | Light transmissive conductive material |
| CN105608474A (en) * | 2015-12-30 | 2016-05-25 | 二十一世纪空间技术应用股份有限公司 | High-resolution-image-based regional adaptive cultivated land extraction method |
| CN105719014A (en) * | 2016-01-18 | 2016-06-29 | 中国农业大学 | Provincial cultivated land back-up resource exploiting and zoning method and system |
-
2016
- 2016-11-01 CN CN201610934973.1A patent/CN106548141B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090315882A1 (en) * | 2008-06-19 | 2009-12-24 | Xerox Corporation | Reducing the number of vertices in a clustered delaunay mesh |
| CN102073981A (en) * | 2011-02-23 | 2011-05-25 | 中国测绘科学研究院 | Point group geographic entity selection method under the restriction of correlated elements |
| CN104036024A (en) * | 2014-06-27 | 2014-09-10 | 浙江大学 | Spatial clustering method based on GACUC (greedy agglomerate category utility clustering) and Delaunay triangulation network |
| JP2016062170A (en) * | 2014-09-16 | 2016-04-25 | 三菱製紙株式会社 | Light transmissive conductive material |
| CN105608474A (en) * | 2015-12-30 | 2016-05-25 | 二十一世纪空间技术应用股份有限公司 | High-resolution-image-based regional adaptive cultivated land extraction method |
| CN105719014A (en) * | 2016-01-18 | 2016-06-29 | 中国农业大学 | Provincial cultivated land back-up resource exploiting and zoning method and system |
Non-Patent Citations (3)
| Title |
|---|
| JINFEI LIU等: "Effective Privacy Preserved Clustering Based on Voronoi Diagram", 《2011 EIGHTH INTERNATIONAL SYMPOSIUM ON VORONOI DIAGRAMS IN SCIENCE AND ENGINEERING》 * |
| 孙小丹等: "农用地遥感影像信息的角提取方法", 《农业工程学报》 * |
| 彭仪普等: "Delaunay三角网与Voronoi图在GIS中的应用研究", 《测绘工程》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107221001A (en) * | 2017-05-18 | 2017-09-29 | 苏州恩巨网络有限公司 | A kind of area information acquisition methods and device |
| CN107564077B (en) * | 2017-09-04 | 2019-12-10 | 浪潮集团有限公司 | method and device for drawing road network |
| CN107564077A (en) * | 2017-09-04 | 2018-01-09 | 济南浪潮高新科技投资发展有限公司 | A kind of method and device for drawing road network |
| CN107705002A (en) * | 2017-09-21 | 2018-02-16 | 中国矿业大学(北京) | The determination method of mining soil content of beary metal sampled point exception high level coverage |
| CN107705002B (en) * | 2017-09-21 | 2020-07-03 | 中国矿业大学(北京) | Method for determining influence range of abnormal high value of sampling point of heavy metal content in mining area soil |
| CN107766495A (en) * | 2017-10-19 | 2018-03-06 | 大连大学 | Distributed temporal index method based on Voronoi diagram |
| CN108021690A (en) * | 2017-10-19 | 2018-05-11 | 大连大学 | Arrange Voronoi indexes |
| CN107958087A (en) * | 2017-10-19 | 2018-04-24 | 大连大学 | The representative point acquisition methods of the distributed temporal index of the row's of falling Thiessen polygon of portable medical |
| CN107766496A (en) * | 2017-10-19 | 2018-03-06 | 大连大学 | Based on MapReduce and the extensive NN Query system for arranging Thiessen polygon |
| CN108961403A (en) * | 2018-06-22 | 2018-12-07 | 武汉大学 | A method of major trunk roads are extracted with open source street map |
| CN111768505A (en) * | 2020-07-01 | 2020-10-13 | 江西师范大学 | Plane point set shape reconstruction method and device and electronic equipment |
| CN111768505B (en) * | 2020-07-01 | 2023-10-17 | 江西师范大学 | Planar point set shape reconstruction method and device and electronic equipment |
| CN112330700A (en) * | 2020-11-16 | 2021-02-05 | 四川航天神坤科技有限公司 | Cultivated land plot extraction method of satellite image |
| CN112926029A (en) * | 2021-01-29 | 2021-06-08 | 四川省环境政策研究与规划院 | Residential area identification and division method for rural domestic sewage treatment |
| CN112926029B (en) * | 2021-01-29 | 2022-08-26 | 四川省环境政策研究与规划院 | Residential area identification and division method for rural domestic sewage treatment |
| CN112598724A (en) * | 2021-03-01 | 2021-04-02 | 武大吉奥信息技术有限公司 | Improved TIN-based planar element center line extraction method |
| CN114354464A (en) * | 2021-12-22 | 2022-04-15 | 南京大学 | Method for quantitatively analyzing similarity between hyperspectral libraries of different metal nanoparticles |
| CN114354464B (en) * | 2021-12-22 | 2024-04-09 | 南京大学 | Method for quantitatively analyzing similarity between hyperspectral libraries of different metal nanoparticles |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106548141B (en) | 2019-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106548141B (en) | A kind of object-oriented farmland information extraction method based on the triangulation network | |
| CN101840581B (en) | Method for extracting profile of building from satellite remote sensing image | |
| Blaschke et al. | Image segmentation methods for object-based analysis and classification | |
| CN105389559A (en) | System and method for identifying agricultural disaster scope based on high-resolution remote sensing image | |
| CN104156964B (en) | A kind of comprehensive MRF and the remote sensing image region segmentation method of Bayesian network | |
| CN103839267B (en) | Building extracting method based on morphological building indexes | |
| CN106709989B (en) | Geographic ontology modeling and semantic reasoning method based on object-oriented image features | |
| CN108985247A (en) | Multispectral image urban road identification method | |
| CN109919944A (en) | A kind of joint super-pixel figure of complex scene building variation detection cuts optimization method | |
| CN102096816A (en) | Multi-scale multi-level image segmentation method based on minimum spanning tree | |
| CN103294792A (en) | Polarimetric SAR (synthetic aperture radar) terrain classification method based on semantic information and polarimetric decomposition | |
| Hu et al. | Integrating CART algorithm and multi-source remote sensing data to estimate sub-pixel impervious surface coverage: a case study from Beijing Municipality, China | |
| CN107194857B (en) | Available land resource analysis system based on high-resolution satellite remote sensing data | |
| CN109766824A (en) | Main passive remote sensing data fusion classification method based on Fuzzy Evidence Theory | |
| Baluyan et al. | Novel approach for rooftop detection using support vector machine | |
| CN105701856B (en) | A kind of vegetation extracting method and system | |
| Guo et al. | Dual-concentrated network with morphological features for tree species classification using hyperspectral image | |
| Nelson et al. | Spatial statistical techniques for aggregating point objects extracted from high spatial resolution remotely sensed imagery | |
| CN103778413B (en) | A kind of remote sensing image less divided object automatic identifying method | |
| Zhang et al. | Urban road extraction from high-resolution remote sensing images based on semantic model | |
| CN112907567B (en) | SAR image ordered artificial structure extraction method based on spatial reasoning method | |
| CN112487977A (en) | Airport runway extraction method based on domestic GF2 remote sensing image | |
| Wang et al. | Combining SAR images with land cover products for rapid urban flood mapping | |
| Overton et al. | Flood modelling and vegetation mapping in large river systems | |
| Hu et al. | An commercial area extraction approach using time series nighttime light remote sensing data——Take Wuhan city as a case |
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 |