CN114170441A - Roadside tree automatic extraction method based on geographic national condition data and image classification - Google Patents
Roadside tree automatic extraction method based on geographic national condition data and image classification Download PDFInfo
- Publication number
- CN114170441A CN114170441A CN202210128588.3A CN202210128588A CN114170441A CN 114170441 A CN114170441 A CN 114170441A CN 202210128588 A CN202210128588 A CN 202210128588A CN 114170441 A CN114170441 A CN 114170441A
- Authority
- CN
- China
- Prior art keywords
- area
- region
- roadside tree
- tree
- roadside
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a roadside tree automatic extraction method based on geographical national condition data and image classification, which is characterized by comprising the following steps of: the method comprises the following steps: deriving road network data from the geographic national condition database, constructing road surfaces according to the road network data, and constructing road identification areas according to the road surfaces; leading out forest land pattern spots from the geographic national condition database, and judging a first area and/or an area to be supplemented by combining the forest land pattern spots and the road identification area; deriving a wayside tree region from the first region according to the wayside tree determining step, and/or deriving a supplemental wayside tree region from the region to be supplemented according to the wayside tree determining step; extracting information of the reserved roadside tree area and/or the supplementary roadside tree area; the invention has the beneficial effects that: the method has the advantages that the situations of losing effective information and data redundancy are prevented, the geographical national condition data and various images are combined, the roadside tree area can be rapidly extracted in the forest coverage rate calculation, the statistical time and the information extraction time are reduced, and the calculation cost is reduced.
Description
Technical Field
The invention belongs to the field of remote sensing image data processing and information extraction, and particularly relates to a roadside tree automatic extraction method based on geographical national condition data and image classification.
Background
The trees at four sides (also called as roadside trees) refer to various bamboos and woods planted at the sides of houses, villages, roads, water systems and the like with the area less than 0.067 hectare. The four-side trees are used as important components of forest coverage rate and forest area, the four-side trees can not only improve the carbon fixation amount of the forest and develop low-carbon economy, but also ensure the healthy development of forest resources and the potential of improving the carbon sink function, and have important functions of environmental regulation and social service. And with the increase of urban area and population density, Bush and Dzifa Adimle Puplampu and the like think that urban greenbelt composed of neighborhood trees can promote urban sustainable development and improve urban habitability, Khostataria and the like also think that urban greenbelt containing neighborhood trees in the neighborhood trees has the capability of preventing natural disasters and reducing damages caused by the natural disasters, highlight the importance of the neighborhood trees in urban construction, and with the increasing demand of obtaining high-precision woodland data, it is very important to apply modern technology to carry out accurate and efficient statistics on the neighborhood trees.
The quanshui trees have the characteristics of scattered distribution and difficulty in statistics, the vegetation research on the quanshui trees is started as early as 1987 in China, the Lisuo uses an area actual measurement method and a plant number conversion method to accurately measure and calculate the Hongqi village mountun, and the method is considered to be the most convenient and rapid method without measuring the plant number; korean Junyi and Menghai Yuan accurately check the resources of the trees in the four sides of Shanxi province in 2003 by combining a special sampling method with a global positioning system; the dynamic analysis of the quanshui tree resources in Jiangsu province in 2005 + 2015 was successfully realized through information statistics in Cao Hua and Wangkai; wangyingying et al accurately and thoroughly obtained the resource status of the 'quadbree' in the Jiyuan city in 2019 by performing village-by-village and road-by-road investigation on the 'quadbree' in the administrative village of the Jiyuan city. However, the on-site investigation method is adopted to extract the quadtree at present, the method has the problems of long investigation time, slow data updating and the like although the extraction precision is high, is difficult to deal with the daily supervision of forestry departments and the high-speed and high-efficiency development planning of modern cities, and has less application to emerging technologies such as remote sensing images, machine learning and the like. In 2008, a foreign learner Yashon O. Ouma utilizes a Quickbird satellite image to extract the area of an urban green land containing a four-side tree with higher precision; roope N ä si and the like utilize a small unmanned aerial vehicle to realize the health monitoring of urban forest lands in 2018; stefanie Lumnitz et al plotted trees along the road with higher accuracy in 2021 by means of deep learning.
However, the above classification studies have the following problems
1. The method for selecting the remote sensing image wave band is less considered, and the problems of effective information loss and data redundancy exist.
2. In the forest coverage rate calculation, due to the sparse distribution of the four-side trees and the field investigation and extraction method, the statistical time is long, the calculation cost is high, and the efficiency is low.
Disclosure of Invention
Aiming at the defects of the prior art, the technical problems to be solved by the invention are as follows: how to provide a roadside tree automatic extraction method based on geographical national condition data and image classification, prevent the problems of effective information loss and data redundancy, and reduce statistical time and calculation cost.
In order to solve the technical problems, the invention adopts the following technical scheme:
the roadside tree automatic extraction method based on geographic national condition data and image classification comprises the following steps:
deriving road network data from the geographic national condition database, constructing road surfaces according to the road network data, and constructing road identification areas according to the road surfaces;
leading out forest land pattern spots from the geographic national condition database, and judging a first area and/or an area to be supplemented by combining the forest land pattern spots and the road identification area;
deriving a wayside tree region from the first region according to the wayside tree determining step, and/or deriving a supplemental wayside tree region from the region to be supplemented according to the wayside tree determining step;
and extracting information of the reserved roadside tree region and/or the supplementary roadside tree region.
In some embodiments, in the method for automatically extracting a roadside tree based on geographic national conditions data and image classification, deriving a forest map spot from a geographic national conditions database, and determining a first region and/or a region to be supplemented by combining the forest map spot and a road identification region, the method includes:
if the forest map spots and the road identification area are intersected, the intersected area is a first area;
and if the forest map spots and the road identification area do not have the intersected areas, the areas which do not have the intersections are the areas to be supplemented.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a roadside tree region from the first region according to the roadside tree determination step includes:
screening the first region according to the standard area of the quadtree;
the area of the forest map spot in the first area, which is larger than the standard area of the quadtree, is cut off;
reserving a region, of which the forest map spot area is smaller than the standard area of the trees beside the forest, in the first region to obtain a first screening region;
a wayside tree region is derived from the first screening region according to the wayside tree determination step.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a roadside tree region from the first screening region according to a roadside tree determination step includes:
judging whether the first screening area is completely positioned in the road identification area or not;
if the first screening area is completely located in the road identification area, the first screening area is a roadside tree area;
if the first screening area part exists in the road identification area, the first screening area which does not exist in the road identification area is obtained as a first remaining area;
and combining the intersection area of the first screening area and the road identification area, the first residual area and the standard area of the four-side tree to obtain the road-side tree area.
In some embodiments, in the method for automatically extracting a roadside tree based on geographical national conditions data and image classification, the step of obtaining the roadside tree region by combining the intersection area of the first screening region and the road identification region, the remaining region and the standard area of the roadside tree includes:
if the first residual region is larger than the difference value between the standard area of the quadtree and the intersection area, the first residual region is discarded;
and if the first residual area is less than or equal to the difference between the standard area of the roadside tree and the intersection area, reserving the first residual area as the roadside tree area.
In some embodiments, the method for automatically extracting a roadside tree based on geographical national conditions data and image classification includes, before deriving a supplementary roadside tree region from a region to be supplemented according to the roadside tree determination step:
selecting a second region from the region to be supplemented according to the image wave band selection classification and the spectral characteristics;
deriving a supplemental roadside tree region from the second region according to the roadside tree determination step.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a supplementary roadside tree region from the second region according to the roadside tree determination step includes:
screening the second region according to the standard area of the quadtree;
the area of the forest map spot in the second area, which is larger than the standard area of the quadtree, is cut off;
reserving a region, of which the forest map spot area is smaller than the standard area of the trees on four sides, in the second region to obtain a second screening region;
deriving a supplemental roadside tree region from the second screening region according to the roadside tree determination step.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a supplementary roadside tree region from the second filtered region according to the roadside tree determination step includes:
judging whether the second screening area is completely positioned in the road identification area or not;
if the second screening area is completely located in the road identification area, the second screening area is a supplementary roadside tree area;
if the second screening area part exists in the road identification area, the second screening area which does not exist in the road identification area is obtained as a second residual area;
and obtaining a supplementary roadside tree region by combining the intersection area of the second screening region and the road identification region, the second residual region and the standard area of the roadside tree.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national condition data and image classification, the obtaining a supplementary roadside tree region by combining the intersection area of the second screening region and the road identification region, the second remaining region and the standard area of the roadside tree includes:
if the second residual area is larger than the difference value between the standard area of the quadtree and the intersection area, the second residual area is discarded;
and if the second residual area is less than or equal to the difference between the standard area of the roadside tree and the intersection area, reserving the second residual area as a supplementary roadside tree area.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national condition data and image classification, the extracting information of the reserved roadside tree region and/or the supplementary roadside tree region includes:
obtaining a final roadside tree region according to the roadside tree region and/or the supplementary roadside tree region;
acquiring a roadside tree position in the final roadside tree region;
calculating the roadside tree area in the final roadside tree region;
and judging the change condition of the roadside tree in the final roadside tree region.
The invention has the beneficial effects that:
1. and selecting by selecting and classifying image wave bands to prevent the situations of effective information loss and data redundancy.
2. By combining geographical national condition data with various images, the roadside tree (quadtree) area can be rapidly extracted in forest coverage rate calculation, the statistical time and the information extraction time are reduced, and the calculation cost is reduced.
Drawings
FIG. 1 is a flow chart of the present invention.
Fig. 2 is a constructed road surface information map.
Fig. 3 is a diagram of a constructed buffer.
FIG. 4 is a schematic view of the forest map after introduction of the spots.
FIG. 5 is a determination diagram of intersection of forest map spots and road identification areas.
Fig. 6 is a diagram before the first screening area is determined to obtain the roadside tree area when the first screening area is completely located in the road identification area.
Fig. 7 is a diagram after the first screening area determines that the roadside tree area is obtained when the first screening area is completely located in the road identification area.
Fig. 8 is a schematic diagram of a forest map shift where the second screening area intersects with the road identification area when the forest map shift is completely located in the road identification area.
Fig. 9 is a schematic diagram of a supplemental roadside tree region finally determined and selected by the second screening region when the second screening region is completely located in the road identification region.
Fig. 10 is a schematic diagram of the second screening area intersecting the road identification area when a portion of the second screening area exists in the road identification area.
Fig. 11 is a schematic diagram of a supplemental roadside tree region that is finally obtained when a part exists in a road identification area.
FIG. 12 is an exemplary diagram of a combined image.
FIG. 13 is a woodland classification diagram.
Fig. 14 is a schematic diagram of determining whether the woodland classification patch intersects with the road identification area.
Fig. 15 shows the result of the intersection determination.
Fig. 16 is a schematic diagram of the final roadside tree region.
FIG. 17 is a roadside tree pattern spot statistical area table.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The roadside tree automatic extraction method based on geographic national condition data and image classification comprises the following steps:
s100: deriving road network data from the geographic national condition database, constructing road surfaces according to the road network data, and constructing road identification areas according to the road surfaces;
deriving road network data from LRDL and LCTL layers of geographical national condition data to obtain road center line and road width information;
constructing a road surface:
when a road surface is constructed, the road center line is used as a middle axis, the road width of 1/2 is used as a radius, the road surface is constructed by extending along the road center line, and the construction result is shown in fig. 2;
when deriving the road network data, a coordinate system of the road network vector data is required to be used as a projection coordinate system;
if the road network coordinate system is not the projection coordinate system, the coordinate system is converted into the projection coordinate system, the unit is modified into meter, and then the road width is in meter when the road surface data is constructed.
Constructing a road identification area:
the method for constructing the distance of the road identification area used at this time is the Euclidean distance, and the calculation mode is as follows:
in the formula (I), the first and second groups of the compound,in order to obtain the Euclidean distance,andto classify the obtained geographical coordinates of the woodland edges,andfor the boundary extending outwardly by 25 metersGeographic coordinates.
The buffer is constructed in such a way that a given road boundary line is presentDetermining its radius toNeighborhood of metersI.e. the midlineAnd its aggregated buffer is:
In the formula (II), the first and second groups of the compound,is the minimum Euclidean distance;is composed ofThe buffer area(s) of (a),is all the way from the road boundary lineNot exceedingThe set of all points of (i.e. the union of the individual target buffers) the result of the construction is shown in fig. 3.
S200: leading out forest land pattern spots from the geographic national condition database, and judging a first area and/or an area to be supplemented by combining the forest land pattern spots and the road identification area;
s300: deriving a wayside tree region from the first region according to the wayside tree determining step, and/or deriving a supplemental wayside tree region from the region to be supplemented according to the wayside tree determining step;
s400: and extracting information of the reserved roadside tree region and/or the supplementary roadside tree region.
In some embodiments, in the method for automatically extracting a roadside tree based on geographic national conditions data and image classification, deriving a forest map spot from a geographic national conditions database, and determining a first region and/or a region to be supplemented by combining the forest map spot and a road identification region, the method includes:
if the forest map spots and the road identification area are intersected, the intersected area is a first area;
and if the forest map spots and the road identification area do not have the intersected areas, the areas which do not have the intersections are the areas to be supplemented.
The forest map spots are derived from the LDTB map spots of the geographical national condition data, and the introduction of the forest map spots is shown in FIG. 4. And judging whether the generated road identification area edge is intersected with the woodland pattern spot in the national condition data, if the generated road identification area edge is intersected with the woodland pattern spot, judging that a roadside tree possibly exists near the road according to a judgment result, and judging the intersected area as a first area.
Because the minimum map area of the forest map spots in the existing geographic national condition database is 400 square meters, if no intersected forest map spots exist, the judgment result is that no roadside trees with the area larger than 400 square meters exist near the road, and the areas which are not intersected are all the areas to be supplemented and are judged as the areas to be supplemented.
The method for judging the intersection of the identification area and the forest map spot comprises the following steps:
theoretically, if there is one distance among the distances from the forest map spot target to the road boundarySatisfy the requirement ofThe radius of the road identification area can be regarded as the forest map spot and roadThe way identification areas intersect. Distance thereofThe calculation formula of (2) is as follows:
In the formula (III)The coordinates of the end points of the road identification area (namely two end points of the edge line of the road surface) are identified,the intersection result for any point of the forest map spot is shown in fig. 5.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a roadside tree region from the first region according to the roadside tree determination step includes:
screening the first region according to the standard area of the quadtree;
the area of the forest map spot in the first area, which is larger than the standard area of the quadtree, is cut off;
reserving a region, of which the forest map spot area is smaller than the standard area of the trees beside the forest, in the first region to obtain a first screening region;
a wayside tree region is derived from the first screening region according to the wayside tree determination step.
The trees are defined as the forest lands which are planted in the vicinity of roads, water, houses and villages within 25 meters and have the area less than 670 square meters.
And judging the calculated areas of the forest map spots in the first region, and sequentially arranging the areas from small to large. And (4) with reference to the definition of the standard area of the four-side tree, cutting off the intersected forest land area with the area larger than 670 square meters in the first area, and reserving the forest land area which can become the road-side tree to obtain a first screening area.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a roadside tree region from the first screening region according to a roadside tree determination step includes:
judging whether the first screening area is completely positioned in the road identification area or not;
if the first screening area is completely located in the road identification area, the first screening area is a roadside tree area;
if the first screening area part exists in the road identification area, the first screening area which does not exist in the road identification area is obtained as a first remaining area;
and combining the intersection area of the first screening area and the road identification area, the first residual area and the standard area of the four-side tree to obtain the road-side tree area.
Judging whether the reserved first screening area is completely in the road identification area, namely judging all distances from forest map spot targets in the reserved first screening area to the road boundarySatisfy the requirement ofAnd if the radius of the road identification area is meter, the forest map spot can be considered to be completely positioned in the road identification area.
In the formula (IV)Two end points on the edge line of the road surface,for any point of the forest map spot,the number of pixels in the forest floor.
When the woodland pattern of the geographical national conditions is completely located in the road identification area, the woodland pattern is retained and is judged as a roadside tree area, the result before judgment is shown in fig. 6, and the result after judgment is shown in fig. 7.
In some embodiments, in the method for automatically extracting a roadside tree based on geographical national conditions data and image classification, the step of obtaining the roadside tree region by combining the intersection area of the first screening region and the road identification region, the remaining region and the standard area of the roadside tree includes:
if the first residual region is larger than the difference value between the standard area of the quadtree and the intersection area, the first residual region is discarded;
and if the first residual area is less than or equal to the difference between the standard area of the roadside tree and the intersection area, reserving the first residual area as the roadside tree area.
If not, judging the area of the rest forest land outside the road identification area (namely the area of the first rest area), judging whether the area of the first rest area is larger than the judgment area, wherein the judgment area is as follows: and (3) subtracting the area of the intersected forest map spots (the intersection area of the first screening area and the road identification area) by 670 square meters, if the first residual area is larger than the judgment area, discarding the land, and if the first residual area is smaller than or equal to the judgment area, reserving the first residual area and judging the first residual area as the roadside tree area.
In some embodiments, the method for automatically extracting a roadside tree based on geographical national conditions data and image classification includes, before deriving a supplementary roadside tree region from a region to be supplemented according to the roadside tree determination step:
selecting a second region from the region to be supplemented according to the image wave band selection classification and the spectral characteristics;
deriving a supplemental roadside tree region from the second region according to the roadside tree determination step.
As most of the woodland pattern spots in the geographic national condition data are woodland pattern spots with the area larger than 400 square meters, most of the roadside trees are small-area pattern spots which are smaller than 670 square meters and distributed scattered, and the geographic national condition data are difficult to completely count, in order to count the roadside trees more accurately, the wave band is used to be at least larger than 3, the image resolution is at least higher than 15 meters, the cloud number of the area where the image is located is less than 0.05 percent, and the multiband remote sensing images without cloud coverage in the area where the roadside trees are extracted are classified, so that the remote sensing images are used as supplements to the roadside tree information, and the supplemented area is called as a supplemented roadside tree area.
The step of screening out the second region from the region to be supplemented by using the image band selection classification and the spectral features comprises the following steps:
the method comprises the following steps: calculating the single-band information quantity;
in order to calculate the information quantity of a single wave band in the roadside tree image, calculating the standard deviation of the single wave band by using the corresponding image primitive wave band value, the average value of the wave band values and the total number of pixels of the image, and preferentially selecting the wave band with large standard deviation from the wave band combination of the subsequent four-sided tree image;
the formula used is:
In the formula (V):is the standard deviation of the measured data to be measured,is the number of the image pixels,for the value of the band of each picture element,is the average of the band values. The image pixel number and the wave band value can be obtained from the image, the average value of the wave band value can be obtained by dividing the sum of the pixel wave band values by the total number of the pixels, and the output result is the standard deviation value of each wave band of the image.
Step two: calculating the correlation degree between the wave bands;
in order to calculate the correlation degree between the image bands of the roadside trees, the covariance between the bands is divided by the respective standard deviation to calculate the band correlation coefficient, and the band with the small correlation coefficient between the bands is preferentially selected in the subsequent four-sided tree image band combination;
the formula used is:
In the formula (six):andexpressed as the standard deviation of the image band i and the band j,the covariance of the i and j bands of the image. Wherein the image standard deviation can be calculated in the first step, and the covariance can be obtained by the following formula:
In formula (seven):andfor two different image bands i and j, n is the pixel number of the band image, isThe band value of each picture element of the band i,is the average of the band values of the band i,andsimilarly, the above parameters can be obtained from the first step.
The output result is the correlation coefficient value of each wave band of the image, the size of the information cross degree of each wave band is judged according to the result, the combination with small cross degree among the wave bands is selected, the wave band combination which has the largest information quantity and the smallest correlation degree and is most suitable for the local roadside tree classification is selected for image classification, and the synthetic result is shown in figure 12.
Step three: obtaining a classification sample;
and performing roadside tree sample selection on the combined image, and acquiring spectral characteristics of roadside trees used for subsequent image classification by a cutting method. The land information (such as grasslands, construction land, forest lands and the like) of the cut samples can be determined in the forms of corresponding geographic national condition data, high-definition satellite images or unmanned aerial vehicle images and the like. After the spectral feature information and the geographic position of the roadside tree and other land features are confirmed, corresponding clipping classification samples can be in the forms of clipped images, ROI files or AOI files, but the samples must be accurately and uniformly distributed in the image of the roadside tree and contain the spectral information and the position information of the clipping samples, and the sample result should have an explicit label of the land feature (such as attribute information of the forest land, the grassland and the like). Training data set obtained at this timeCan be expressed as:
In the formula (eight), the first and second groups,representing the corresponding band values of the input samples (3 bands are taken as an example in the formula, the number of bands can be greater than 3, L is the assigned class label,in order to classify the number of samples,is the number of the classified classes.
Step four: classifying the remote sensing images;
the acquired spectral feature information samples of the roadside trees and the images to be classified are led into a support vector machine classification algorithm, pixel values of all wave bands (such as R, G, B wave bands of a three-wave band image) of the classified samples with labels are projected into a high-dimensional (such as a three-wave band image needs to construct a three-dimensional space) space to create a hyperplane in the image classification process, a radial basis kernel function is used for assisting the hyperplane construction in the construction process, the hyperplane is used for segmenting the pixel values in the images to be classified to classify, and the classified image results of the roadside trees are obtained. Because the spectral characteristics of buildings and water bodies are more obvious in the images, the samples are divided into the following parts in the classification process: and classifying the water body, the non-water body, the building and the non-building, and removing the classified buildings and water bodies from the classification result to obtain the final woodland class.
At this point, after the image classification is finished and the building pattern spots, the water body pattern spots, the grassland and farmland pattern spots are discarded, the forest classification result is shown in fig. 13.
After the remote sensing image is classified, the corresponding coordinates exist in the classification result image spots, so that the judgment is only needed to judge whether the coordinates are in the road identification area, the judgment mode and the example are shown in figure 14,
the rectangle in the figure is a road identification area, firstly, a vertical line is drawn from the classified pixels to the y axis, and the image spots with possible intersection points are screened.
Assuming that the coordinates of the end points of the road identification area are (x 1, y 1) and (x 2, y 2), intersection is possible only when the x coordinate of the classified forest land is between x1 and x2, so that the non-intersected forest land pattern spots d are discarded.
Then, the number of the intersection points is judged.
a. when in useWhen the number of the intersection points is odd, outputting and judging intersected forest map spots a and f;
b. when in use=Outputting a forest map spot b which is used for judging the intersection on the boundary (one intersection point can be calculated);
c. when in useHowever, when there are even number of intersections, the patch is located outside the identified area, and therefore the disjoint forest map patch e is discarded.
Therefore, when the intersection points of the coordinates of the forest map spots and the boundary of the identification area are odd, the intersection points are intersected with the identification area; when the number is even: not within the identification zone. The final determination of the intersected woodland patches is shown in FIG. 15.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a supplementary roadside tree region from the second region according to the roadside tree determination step includes:
screening the second region according to the standard area of the quadtree;
the area of the forest map spot in the second area, which is larger than the standard area of the quadtree, is cut off;
reserving a region, of which the forest map spot area is smaller than the standard area of the trees on four sides, in the second region to obtain a second screening region;
deriving a supplemental roadside tree region from the second screening region according to the roadside tree determination step.
The trees are defined as the forest lands which are planted in the vicinity of roads, water, houses and villages within 25 meters and have the area less than 670 square meters.
And judging the calculated areas of the forest map spots in the second area, and sequentially arranging the areas from small to large. And (4) with reference to the definition of the standard area of the four-side tree, discarding the intersected forest land area with the area larger than 670 square meters in the second area, and reserving the forest land area which can become the road-side tree to obtain a second screening area.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, deriving a supplementary roadside tree region from the second filtered region according to the roadside tree determination step includes:
judging whether the second screening area is completely positioned in the road identification area or not;
if the second screening area is completely located in the road identification area, the second screening area is a supplementary roadside tree area;
if the second screening area part exists in the road identification area, the second screening area which does not exist in the road identification area is obtained as a second residual area;
and obtaining a supplementary roadside tree region by combining the intersection area of the second screening region and the road identification region, the second residual region and the standard area of the roadside tree.
For the reserved second screening area, calculating the area of the land parcel for the result of the intersected pattern spots in the following way:
thus, the areas of various regions of the image classification are obtained, and the areas are arranged from small to large in sequence. With reference to the definition of the quadtree, all forest areas with area greater than 670 m are discarded, leaving areas that may become supplemental roadside trees.
When the forest map spot of the second screening area is completely located in the road identification area, the forest map spot is reserved and is judged as a supplementary wayside tree area, the judgment result is shown in fig. 8 and 9, the forest map spot in which the second screening area intersects with the road identification area is shown in fig. 8, and the supplementary wayside tree area finally selected by the second screening area is shown in fig. 9.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national condition data and image classification, the obtaining a supplementary roadside tree region by combining the intersection area of the second screening region and the road identification region, the second remaining region and the standard area of the roadside tree includes:
if the second residual area is larger than the difference value between the standard area of the quadtree and the intersection area, the second residual area is discarded;
and if the second residual area is less than or equal to the difference between the standard area of the roadside tree and the intersection area, reserving the second residual area as a supplementary roadside tree area.
If not, judging the forest land area of the second surplus region, judging whether the area of the second surplus region is larger than the judgment area, and judging that the area is as follows: and (3) subtracting the area of the intersected forest map spot (namely the intersected area of the second screening area and the road identification area) from 670 square meters, if the intersected area is larger than the judgment area, discarding the land, and if the intersected area is smaller than the reserved area, judging that the intersected area is a supplemented roadside tree area. The determination results are shown in fig. 10 and fig. 11, the black area in fig. 10 is the geographical national forest map spot intersecting the second screening area and the road identification area, and the reserved forest map spot in the second remaining area in fig. 11 is the supplemental roadside tree area.
When judging that the area of building and water body exist in the road buffer:
there are buildings:
1) firstly, constructing a 25-meter building identification area for the building pattern spots according to the method in the step S100, judging whether the forest pattern spots are located in the building identification area, if not, judging that the forest pattern spots are roadside trees, and if so, entering the next step;
2) judging whether the building map spots are map spots which can be used for people to live in for a long time, such as residential districts, unit courtyards and the like in the map layer of the geographic national condition data BUCP, if the building map spots meet the condition, the part of the forest lands should be separately extracted and marked as 'residential side trees', and if cultivated lands or garden lands in the geographic national condition data exist in the building identification area, the part of the forest lands should be marked as 'possible rural side trees', and then the part of the forest lands is confirmed in a field investigation mode.
The presence of a water body:
1) firstly, constructing a 25-meter water body identification area for the building pattern spots according to the step S100, judging whether the forest pattern spots are located in the water body identification area, if not, judging that the forest pattern spots are roadside trees, and if so, entering the next step;
2) whether the water body of the road identification area is in HYDA and HYDL pattern spots of the geographical and national condition data or not is judged, and if the water body is a river, a lake, a reservoir and the like in the pattern spots, the woodland pattern spots are divided into water side trees through the water body. If the water bodies are scattered or distributed in a regular pattern, judging whether the water bodies are paddy fields or pools by contrasting HYDA and HYDL pattern spots of geographical and national condition elements, and if the water bodies are the paddy fields or the pools, dividing the forest lands into roadside trees. If the water body is not in the geographic national condition data and the water body type cannot be distinguished, the segment of forest land is separately marked as 'possible water side tree', and is confirmed in a subsequent on-site investigation mode.
In some embodiments, in the above method for automatically extracting a roadside tree based on geographical national conditions data and image classification, the information extraction for the reserved roadside tree region and/or the supplementary roadside tree region includes:
obtaining a final roadside tree region according to the roadside tree region and/or the supplementary roadside tree region;
acquiring a roadside tree position in the final roadside tree region;
calculating the roadside tree area in the final roadside tree region;
and judging the change condition of the roadside tree in the final roadside tree region.
And summarizing the woodland patches judged as the wayside tree area and the supplementary wayside tree area to obtain a final wayside tree area, calculating the area of a wayside tree from the final wayside tree area by using a method for judging/supplementing the wayside tree area, and simultaneously acquiring the position of the woodland and the wayside patches which are not clearly distinguished according to the projection coordinates, thereby facilitating the subsequent on-site investigation. And the image data of many years can be used for classification, and the area and the position change of the recent roadside tree are compared.
The final wayside tree region and statistical area are shown in fig. 16 and 17.
The above is only a preferred embodiment of the present invention, and it should be noted that several modifications and improvements made by those skilled in the art without departing from the technical solution should also be considered as falling within the scope of the claims.
Claims (10)
1. The roadside tree automatic extraction method based on geographic national condition data and image classification is characterized by comprising the following steps of: the method comprises the following steps:
deriving road network data from a geographic national condition database, constructing a road surface according to the road network data, and constructing a road identification area according to the road surface;
leading out forest land pattern spots from a geographical national condition database, and determining a first area and/or an area to be supplemented by combining the forest land pattern spots and the road identification area;
deriving a wayside tree region from the first region according to a wayside tree determination step, and/or deriving a supplemental wayside tree region from the region to be supplemented according to the wayside tree determination step;
and extracting information of the reserved roadside tree region and/or the supplementary roadside tree region.
2. The method for automatically extracting the roadside tree based on the geographic national conditions data and the image classification as claimed in claim 1, wherein: the geographic national conditions database derives forest map spots, and the first area and/or the area to be supplemented are/is judged by combining the forest map spots and the road identification area, wherein the method comprises the following steps:
if the forest map spot and the road identification area are intersected, the intersected area is the first area;
and if the forest map spots and the road identification area do not have intersected areas, the areas which do not intersect are the areas to be supplemented.
3. The method for automatically extracting roadside trees based on geographical national conditions data and image classification as claimed in claim 2, wherein: the deriving a roadside tree region from the first region according to the roadside tree determination step includes:
screening the first region according to the standard area of the quadtree;
the area of the forest map spot in the first area, which is larger than the standard area of the quadtree, is cut off;
reserving a region, of which the forest map spot area is smaller than the standard area of the side trees, in the first region to obtain a first screening region;
a wayside tree region is derived from the first screening region according to a wayside tree determination step.
4. The method for automatically extracting roadside trees based on geographical national conditions data and image classification as claimed in claim 3, wherein: the deriving a roadside tree region from the first screening region according to the roadside tree determining step includes:
judging whether the first screening area is located in the road identification area or not;
if the first screening area is completely located in the road identification area, the first screening area is a roadside tree area;
if the first screening area part exists in the road identification area, the first screening area which does not exist in the road identification area is obtained as a first remaining area;
and combining the intersection area of the first screening area and the road identification area, the first residual area and the standard area of the roadside tree to obtain the roadside tree area.
5. The method for automatically extracting roadside trees based on geographical national conditions data and image classification as claimed in claim 4, wherein: the obtaining the roadside tree region by combining the intersection area of the first screening region and the road identification region, the residual region and the standard area of the roadside tree comprises:
if the first residual region is larger than the difference value between the standard area of the quadtree and the intersection area, the first residual region is discarded;
and if the first residual area is less than or equal to the difference value between the standard area of the roadside tree and the intersection area, reserving the first residual area as the roadside tree area.
6. The method for automatically extracting the roadside tree based on the geographic national conditions data and the image classification as claimed in claim 1, wherein: before the step of deriving a supplementary roadside tree region from the region to be supplemented according to the roadside tree determination step, the method includes:
screening out a second region from the region to be supplemented according to the image band selection classification and the spectral characteristics;
deriving a supplemental roadside tree region from the second region according to the roadside tree determination step.
7. The method for automatically extracting roadside trees based on geographical national conditions data and image classification as claimed in claim 6, wherein: said deriving a supplemental roadside tree region from the second region according to the roadside tree determination step, comprising:
screening the second region according to the standard area of the quadtree;
the area of the forest map spot in the second area, which is larger than the standard area of the quadtree, is cut off;
reserving a region, of which the forest map spot area in the second region is smaller than the standard area of the side trees, to obtain a second screening region;
deriving a supplemental roadside tree region from the second screening region according to a roadside tree determination step.
8. The method for automatically extracting roadside trees based on geographical national conditions data and image classification as claimed in claim 7, wherein: the deriving a supplemental roadside tree region from the second screening region according to the roadside tree determining step includes:
judging whether the second screening area is located in the road identification area or not;
if the second screening area is completely located in the road identification area, the second screening area is a supplementary roadside tree area;
if the second screening area part exists in the road identification area, the second screening area which does not exist in the road identification area is obtained as a second remaining area;
and combining the intersection area of the second screening area and the road identification area, the second residual area and the standard area of the roadside tree to obtain the supplementary roadside tree area.
9. The method for automatically extracting roadside trees based on geographical national conditions data and image classification as claimed in claim 8, wherein: the obtaining the supplementary roadside tree region by combining the intersection area of the second screening region and the road identification region, the second remaining region and the standard area of the roadside tree, includes:
if the second residual region is larger than the difference value between the standard area of the quadtree and the intersection area, the second residual region is discarded;
and if the second residual area is less than or equal to the difference value between the standard area of the quadtree and the intersection area, reserving the second residual area as the supplementary roadside tree area.
10. The method for automatically extracting the roadside tree based on the geographic national conditions data and the image classification as claimed in claim 1, wherein: performing information extraction on the reserved roadside tree region and/or the supplementary roadside tree region, including:
obtaining a final roadside tree region according to the roadside tree region and/or the supplementary roadside tree region;
acquiring a roadside tree position in the final roadside tree region;
calculating the roadside tree area in the final roadside tree region;
and judging the change condition of the roadside tree in the final roadside tree region.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210128588.3A CN114170441B (en) | 2022-02-11 | 2022-02-11 | Roadside tree automatic extraction method based on geographic national condition data and image classification |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210128588.3A CN114170441B (en) | 2022-02-11 | 2022-02-11 | Roadside tree automatic extraction method based on geographic national condition data and image classification |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114170441A true CN114170441A (en) | 2022-03-11 |
CN114170441B CN114170441B (en) | 2022-05-10 |
Family
ID=80489739
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210128588.3A Active CN114170441B (en) | 2022-02-11 | 2022-02-11 | Roadside tree automatic extraction method based on geographic national condition data and image classification |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114170441B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115546339A (en) * | 2022-10-14 | 2022-12-30 | 中国测绘科学研究院 | Urban forest drawing method, device and equipment by using landscape ecological index model |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2006024686A1 (en) * | 2004-07-26 | 2006-03-09 | Consejo Superior De Investigaciones Científicas | Method of differentiating and quantifying an olive grove with vegetative covers by means of remote sensing with high-resolution space sensors |
CN113284171A (en) * | 2021-06-18 | 2021-08-20 | 成都天巡微小卫星科技有限责任公司 | Vegetation height analysis method and system based on satellite remote sensing stereo imaging |
-
2022
- 2022-02-11 CN CN202210128588.3A patent/CN114170441B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2006024686A1 (en) * | 2004-07-26 | 2006-03-09 | Consejo Superior De Investigaciones Científicas | Method of differentiating and quantifying an olive grove with vegetative covers by means of remote sensing with high-resolution space sensors |
CN113284171A (en) * | 2021-06-18 | 2021-08-20 | 成都天巡微小卫星科技有限责任公司 | Vegetation height analysis method and system based on satellite remote sensing stereo imaging |
Non-Patent Citations (6)
Title |
---|
YASHON O. OUMA ET AL.: "On the optimization and selection of wavelet texture for feature extraction from high-resolution satellite imagery with application towards urban-tree delineation", 《INTERNATIONAL JOURNAL OF REMOTE》 * |
万泽敏 等: "四旁树抽样调查总体指标按行政区域分解方法――以上海市嘉定区为例", 《华东森林经理》 * |
党宇等: "基于深度学习AlexNet的遥感影像地表覆盖分类评价研究", 《地球信息科学学报》 * |
李春干 等: "基于卫星遥感数据空中抽样的大尺度森林资源动态监测", 《林业资源管理》 * |
马锁金 等: "四旁树数据库管理系统的研建", 《江苏林业科技》 * |
黄桂娥 等: "卫星遥感图像技术在森林二类清查中的应用研究", 《现代农业科技》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115546339A (en) * | 2022-10-14 | 2022-12-30 | 中国测绘科学研究院 | Urban forest drawing method, device and equipment by using landscape ecological index model |
Also Published As
Publication number | Publication date |
---|---|
CN114170441B (en) | 2022-05-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111666918B (en) | Coastline change identification method based on multiple factors | |
Bechtel et al. | Generating WUDAPT Level 0 data–Current status of production and evaluation | |
Feranec et al. | Inventory of major landscape changes in the Czech Republic, Hungary, Romania and Slovak Republic 1970s–1990s | |
CN101840581B (en) | Method for extracting profile of building from satellite remote sensing image | |
CN109389163B (en) | Unmanned aerial vehicle image classification system and method based on topographic map | |
CN108764193A (en) | Merge the city function limited region dividing method of POI and remote sensing image | |
Hurd et al. | Characterization of forest fragmentation and urban sprawl using time sequential Landsat imagery | |
CN109299673A (en) | The green degree spatial extraction method of group of cities and medium | |
CN104408463B (en) | High-resolution construction land pattern spot identification method | |
CN109766824B (en) | Active and passive remote sensing data fusion classification method based on fuzzy evidence theory | |
CN115452759B (en) | River and lake health index evaluation method and system based on satellite remote sensing data | |
Guo et al. | Urban impervious surface extraction based on multi-features and random forest | |
CN114170441B (en) | Roadside tree automatic extraction method based on geographic national condition data and image classification | |
CN116385867A (en) | Ecological land block monitoring, identifying and analyzing method, system, medium, equipment and terminal | |
CN113780459B (en) | Urban and rural settlement type automatic identification method and system based on spatial pedigree | |
CN113610873B (en) | Noctilucent remote sensing image data processing method and device | |
CN116129262A (en) | Cultivated land suitability evaluation method and system for suitable mechanized transformation | |
CN115062682A (en) | Target function area identification method, device, storage medium and equipment | |
Beiranvand et al. | Evaluation of changes in per capita green space through remote sensing data | |
Tian et al. | Suburban sprawl measurement and landscape analysis of cropland and ecological land: A case study of Jiangsu Province, China | |
CN112836590B (en) | Flood disaster monitoring method and device, electronic equipment and storage medium | |
CN117953387A (en) | Cross-resolution land coverage classification method and system based on knowledge consistency constraint | |
CN106021499A (en) | Construction land classification method and device based on geographic information of volunteer | |
CN115457386A (en) | Village land informatization generation method | |
Gafuraningtyas et al. | Changes in Land Cover in Temon District Impact of Development of Yogyakarta International Airport |
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 |