CN111721267A

CN111721267A - Method for predicting building height by using satellite image

Info

Publication number: CN111721267A
Application number: CN202010711233.8A
Authority: CN
Inventors: 卢鹤立; 刘桂芳; 夏浩铭; 冯亚飞
Original assignee: Henan University
Current assignee: Henan University
Priority date: 2020-07-22
Filing date: 2020-07-22
Publication date: 2020-09-29
Also published as: LU102459B1

Abstract

The invention relates to a method for predicting the height of a building by utilizing a satellite image, which belongs to the technical field of building height measurement and calculation, wherein the method measures the heights of a small number of buildings in a target area, performs ratio operation on the measured heights and the corresponding shadow lengths in the satellite image, and then takes the average value of all ratios as a ratio coefficient for combining the shadow lengths of all buildings in a satellite image to calculate the estimated height of each building; and then, correcting the estimated height of each building by using the determined correction model to obtain a high-precision corrected height. The method disclosed by the invention is simple in principle, easy to realize, free of calculation of the solar azimuth angle, the solar altitude angle and the like, small in workload, high in calculation precision and calculation efficiency and high in application value.

Description

Method for predicting building height by using satellite image

Technical Field

The invention belongs to the technical field of building height measurement and calculation, and particularly relates to a method for predicting building height by using satellite images.

Background

The building is a major component of a city, and the increase in height reflects the expansion of the vertical space of the city. The method has the advantages that the height information of the buildings in the residential area is extracted quickly and accurately, and the method has great significance for researching the volume ratio of the buildings, improving the living quality of urban residents and the like.

In the prior art, chinese patent publication No. CN108765488A proposes a shadow-based high-resolution remote sensing image building height estimation method, which determines a shadow length by performing shadow detection on an original image to be detected, and then finds a solar azimuth angle, and further calculates a solar altitude angle, and estimates the height of a building by using the obtained shadow length and solar altitude angle. The method has the following disadvantages: the method is complex, the solar azimuth angle, the solar altitude angle and the like need to be obtained, the workload is large, the calculation efficiency is low, and the calculation precision does not meet the requirement.

Disclosure of Invention

The invention aims to provide a method for predicting the height of a building by using a satellite image, which is used for solving the problems of complexity and low calculation efficiency and precision of the existing method.

Based on the above purpose, a method for predicting the height of a building by using a satellite image has the following technical scheme:

(1) acquiring a satellite image of a target area, extracting shadows of buildings in the satellite image, and calculating the shadow length of each building in the target area;

(2) selecting N buildings in a target area for height measurement, wherein N is more than or equal to 2, and the selection conditions of the N buildings are as follows: the heights of the selected buildings are different from each other; acquiring actual measurement heights of the N buildings, respectively calculating ratios between shadow lengths of the N buildings and the actual measurement heights, calculating an average value of the N ratios, and determining a ratio coefficient;

(3) calculating to obtain the estimated height of each building in the target area by utilizing the ratio coefficient and combining the shadow length of each building in the target area;

(4) acquiring a correction model for representing the relationship between the corrected height and the estimated height of the building, wherein the parameters of the correction model are obtained by fitting the actual measured heights and the estimated heights of the N buildings in the target area; and calculating the corrected height of each building in the target area by combining the estimated height of each building in the target area by using the correction model.

The beneficial effects of the above technical scheme are:

the prediction method comprises the steps of measuring the heights of a small number of buildings in a target area, carrying out ratio operation on the measured heights and the corresponding shadow lengths in a satellite image, and then taking the average value of all the ratios as a ratio coefficient for combining the shadow lengths of all the buildings in the satellite image to obtain the estimated height of each building; and then, correcting the estimated height of each building by using the determined correction model to obtain a high-precision corrected height. The method disclosed by the invention is simple in principle, easy to realize, free of calculation of the solar azimuth angle, the solar altitude angle and the like, small in workload, high in calculation precision and calculation efficiency and high in application value.

Further, in order to reduce the calculation error, in the step (1), the shadow length of each building in the target area is calculated by adopting a box-line graph method, and the steps are as follows:

1) according to the sun azimuth angle when the satellite image is shot, intersecting the simulated sun rays with the shadow plane of the building in the satellite image, determining a plurality of line segments in the shadow plane, and arranging the line segments from small to large according to the line segment values to form an observation sample;

2) determining an upper quartile and a lower quartile in the observation sample, calculating a difference value between the upper quartile and the lower quartile, and determining abnormal value truncation points Y1 and Y2 according to the difference value, wherein the calculation formula is as follows:

Y1＝X_0.75+1.5×IQR

Y2＝X_0.25-1.5×IQR

wherein Y1 is the first outlier cutoff point, Y2 is the second outlier cutoff point, and X_0.25Is the lower quartile, X_0.75The number of upper quartiles is defined as the IQR, and the number of lower quartiles is defined as the difference between the upper quartile and the lower quartile;

3) and eliminating the line segment values which are greater than the abnormal value truncation point Y1 and less than the abnormal value truncation point Y2 in the observation sample, and averaging the remaining line segment values to obtain the shadow length of the building.

The abnormal value of the line segment value in each shadow plane is automatically and quickly identified by using a box line graph mathematical model (namely, a box line graph method), and is deleted, so that the automatic screening of the abnormal value is realized, and the detection efficiency is high.

Further, in order to improve the accuracy of inverting the building height, the calculation formula of the correction model is as follows:

in the formula (II), H'_n,xFor modifying the height of the building, H_n,xAnd a, b and c are parameters in the correction model and are obtained by fitting the actual measured height and the estimated height of the known sample point in the target area.

Further, the calculation formula of the ratio coefficient in the step (2) is as follows:

in the formula, K_nIs a coefficient of ratio, S_n,xIs the shadow length of the x-th building, h_n,xIs the actual measured height of the x-th building and N is the number of buildings.

Further, the estimated height of each building in step (3) is calculated as follows:

in the formula, H_n,xIs the estimated height of the building.

Drawings

FIG. 1 is a flow chart of a method for predicting building height in an embodiment of the present invention;

FIG. 2 is a front view of a target area in an embodiment of the present invention;

FIG. 3 is a schematic diagram of 34 building shadows in an orthographic view according to an embodiment of the present invention;

fig. 4 is a box plot of the 28 th shaded surface of a building shaded surface in an embodiment of the present invention.

Detailed Description

The following further describes embodiments of the present invention with reference to the drawings.

The present embodiment provides a method for predicting building height using satellite images, which has the following basic ideas: the method comprises the steps of shooting a plurality of buildings in a target area by using an orthographic camera on a resource third satellite to obtain an orthographic image of the buildings, simulating solar rays by using a solar azimuth angle, intersecting the simulated solar rays with a building shadow plane in the orthographic image, determining a plurality of line segments in the shadow plane, eliminating abnormal values of the line segments in the building shadow plane by using a box line graph method, averaging the rest line segments, and calculating to obtain the shadow length of the buildings in the orthographic image, namely the shadow length.

Then, the actual heights (i.e., actual measured heights) of the set number of buildings in the target area are measured, for example, the actual heights of three buildings are measured respectively, the ratios between the shadow lengths and the actual heights of the three buildings are calculated respectively, and the average value of the three ratios is obtained to obtain a ratio coefficient.

Finally, by utilizing the ratio coefficient and combining the length of the shadow surface of each building in the front-view image, the height estimation value of each building can be determined; and calculating the corrected height value of each building in the target area through an inversion model between the height estimated value and the corrected height value of the building to obtain the accurate height of the building, wherein the overall process is shown in fig. 1. The implementation steps of the method are specifically set forth below:

step 1, extracting the shadow of a building, which comprises the following specific steps:

according to the front-view camera on the resource three satellite (namely ZY-3), the front-view image of the target area is taken as shown in FIG. 2, the building shadows in the front-view image of the residential area are extracted according to the brightness difference of the image, and 34 building shadows in the front-view image of FIG. 2 are shown in FIG. 3.

Step 2, calculating the shadow length according to the building shadow in fig. 3, and the specific implementation method is as follows:

(1) according to the sun azimuth angle when the front-view image is shot, the sun rays are simulated and drawn in the figure 3, the simulated sun rays are intersected with the shadow plane of the building in the front-view image, and a plurality of line segments are determined in the shadow plane.

(2) In order to reduce errors and improve screening efficiency, a box-line graph method is adopted to eliminate abnormal values of line segments in shadow planes of various buildings, and the method is exemplified by eliminating the abnormal value of the line segment in the shadow plane of one of the buildings, and comprises the following specific steps:

s01, measuring n line segments which are total in the shadow surface of the building, correspondingly obtaining n sample observation values, and arranging the sample observation values into X1, X2, X3 … and Xn according to the sequence of the observation values from small to large to form observation samples, wherein when n is 20, the arrangement is shown in the following table 1.

TABLE 1

S02, calculating the quartile and the median of the sample observed value, wherein the calculation formula is as follows:

wherein n is the number of samples, and Xp is the p-quantile (0) of the samples<p<1) When p is 0.25, X_0.25Represents the lower quartile; when p is 0.75, X_0.75Representing the upper quartile.

According to the formula, X is calculated according to the data in the table 1_0.25And X_0.75The calculation results are as follows:

X_0.25＝1/2*(X5+X6)＝1/2*(20.5861671774+21.1124590567)＝20.84931

X_0.75＝1/2*(X15+X16)＝1/2*(25.7960830656+25.9315853827)＝25.86383

s03, calculating two quantiles X_0.25And X_0.75Difference between (i.e. quantile distance):

IQR＝X_0.75-X_0.75＝5.01452

s04, calculating two abnormal value truncation points, wherein the calculation formula of the two abnormal value truncation points is as follows:

Y1＝X_0.75+1.5×IQR＝33.38561

Y2＝X_0.25-1.5×IQR＝13.32153

in the formula, Y1 is the first outlier cutoff point, and Y2 is the second outlier cutoff point.

S05, abnormal value judgment:

when the sample observed value Xn<X_0.25-1.5 IQR or Xn>X_0.75+1.5 × IQR indicates an abnormal value of Xn. Therefore, by performing the judgment based on the abnormal value cutoff point calculated in step S04, it can be determined that the abnormal value cutoff point is located in the section [13.32153,33.38561]The other observation sample data Xi are abnormal values. As can be seen from table 1, the observation sample X1 ═ 0.389242773 and the observation sample X2 ═ 5.591321954 cut off the abnormal value interval [13.32153,33.38561 ═ 5.591321954]Out of the range of (a), so that the abnormal values X1 and X2 are rejected; sample X3 is 19.69029248, which is the minimum value of the number of samples after the removal, and automatically becomes the "lower edge" of the box plot, sample X20 is 28.30353732, which is the maximum value of the number of samples after the removal, and becomes the "upper edge" of the box plot, and the box plot after the removal of the abnormal value is shown in fig. 4, and the abscissa "length 28" in the figure represents the line length of the 28 th shaded surface.

(3) After the abnormal value is eliminated, the remaining line segments (namely the sunlight rays) in the shadow plane of each building are averaged, and the shadow length of each building in the front-view image, namely the shadow length, is obtained through calculation. The shadow lengths of 34 buildings are calculated by numbering the buildings according to the sequence of 0-33, and are shown in the following table 2.

TABLE 2

And 3, calculating a ratio coefficient, wherein the specific calculation method comprises the following steps:

firstly, in a research area, according to different heights of buildings, three sample points are selected, the actual heights (namely the actual measurement heights) of the three sample points (buildings) are measured, and the ratio of the shadow length to the actual heights is calculated and used for constructing a ratio coefficient K_nAs shown in table 3.

TABLE 3

Numbering	Shadow length (meter)	Actual height (rice)	Shadow length/actual height
				0	26.927	31.9	0.844
9	36.638	43.5	0.842
				24	40.132	52.2	0.769

From the data in Table 3, the ratio coefficient K was calculated_nThe calculation formula is as follows:

in the formula, subscripts 1,2, 3 denote three sample points, S, at which the actual height of the building has been taken_n,1、S_n,2、S_n,3Length of the facade shadow plane (i.e. shadow length), h, of three sample points_n,1、h_n,2、h_n,3For the actual measured height of the building corresponding to the facade building shadow surface of the three sample points, the correlation values of the three sample points in the table 3 are taken into the formula (1), and the ratio coefficient K is obtained through calculation_nIs 0.818.

In the step, the selection conditions of the sample points are that the heights of the selected buildings need to be different from each other, and more preferably, the larger the height difference between the selected buildings is, the better the ratio coefficient is, the ratio coefficient of the buildings in the research area can be represented more reasonably and has representativeness. Therefore, three buildings with large differences are selected in this step, as another embodiment, another number (more than two) of buildings may be selected, which is representative, and the specific number is not limited, and when N buildings are selected, the corresponding calculation formula of the ratio coefficient:

And 4, calculating the estimated height of the building, which comprises the following specific steps:

according to the ratio coefficient K in the step 3_nAnd calculating the estimated height of the building by combining the shadow length obtained in the step 2, wherein the estimated height is calculated by the following formula:

in the formula, H_n,xEstimating a height for the estimated building, S_n,xThe length of the shadow of the building is shown as x, which is the building number 1,2, …, 34.

The estimated heights of 34 buildings in the target area were calculated according to the above formula (2), and the calculation results are shown in table 4.

TABLE 4

Numbering	Building estimated height (rice)
		0	32.9181840709
1	32.0917660269
		2	35.4763414059
3	33.9285851467
		4	33.0473877017
5	38.2165656846
		6	31.7652906968
7	30.9561071149
		8	35.1326747188
9	44.7897835452
		10	34.0660152812
11	41.6780893154
		12	33.2144755134
13	38.6476679218
		14	44.7123556112
15	38.1330807213
		16	32.5998575672
17	33.1198353178
		18	45.1857157946
19	33.8966864548
		20	34.6249938998
21	47.416317066
		22	32.7029683741
23	49.1557320905
		24	49.0609188998
25	38.7303794988
		26	45.068242335
27	31.1487885819
		28	29.3406661125
29	28.8366111614
		30	30.0909739976
31	31.8327183741
		32	30.0845996699
33	32.3694284108

And 5, establishing a correction model, and correcting the height of the building based on the correction model to obtain a corrected height value of the building. The method comprises the following specific steps:

estimating height H with estimated building_n,xAs independent variable, the corrected height H 'of the building is adopted'_n,xEstablishing a correction model by using a quadratic equation of a single element as a dependent variable, wherein the expression of the correction model is as follows:

in the formula (II), H'_n,xTo correct the building height (i.e., the corrected height), H_n,xFor the estimated building height (i.e. estimated height), a, b, c are all parameters in the calibration model, which can be obtained by fitting the actual building height (i.e. actual measured height) of known sample points in the area under study to the estimated building height. In this example, parameters a, b, c were determined to be 0.0657, -4.1246, 96.5325, respectively, and the resulting correction model was:

after the correction model is determined, the 34 groups of building estimated height data in the step 4 are respectively substituted into the correction model, so that the corrected height values of 34 buildings in the target area can be calculated, and the accuracy is high.

In order to verify the accuracy of the results, the actual heights of 34 buildings were actually measured, and the corrected height values of 34 groups of buildings were calculated according to the method of steps 1 to 5 in this embodiment, as shown in table 5. The result shows that the calculation accuracy of 24 buildings is improved by 4.55% on average compared with the accuracy calculated without using the unary quadratic correction model by using the constructed unary quadratic correction model in the extracted 34 buildings, and the calculation accuracy of 10 buildings is reduced by 3.88% on average, so that the method has good feasibility for inverting the height of the building based on the unary quadratic correction model.

TABLE 5

The prediction method of the invention has the following advantages:

1) the method is simple and easy to realize, does not need to calculate the solar azimuth angle, the solar altitude angle and the like, has small workload and high calculation efficiency;

2) the prediction precision is high, and the method is mainly embodied from two aspects:

on one hand, the height is measured by selecting a small representative building degree, a reasonable ratio coefficient is determined, and the proportional relation between the shadow length of the building and the actually measured height can be accurately represented; and moreover, the abnormal value of the line segment in each shadow plane can be quickly identified and deleted by using a box line graph method, so that the rationality of the ratio coefficient is further improved, and the estimated height of the building with higher precision can be determined.

On the other hand, the accuracy of inverting the height of the building can be effectively improved by constructing a correction model for representing the relationship between the corrected height and the estimated height of the building.

In conclusion, the method can quickly and accurately invert the building height of the residential area, and has good feasibility.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.

Claims

1. A method for predicting building height by using satellite images is characterized by comprising the following steps:

2. The method for predicting the height of the building by using the satellite images as claimed in claim 1, wherein in the step (1), the shadow length of each building in the target area is calculated by using a box-line diagram method, and the steps are as follows:

Y1＝X_0.75+1.5×IQR

Y2＝X_0.25-1.5×IQR

3. The method of claim 1, wherein the correction model is calculated as follows:

in the formula (II), H'_n,xFor modifying the height of the building, H_n,xFor the estimated height of the building, a, b, c are all parameters in the calibration model.

4. The method for predicting the height of a building using satellite images as claimed in claim 1, wherein the ratio coefficients in step (2) are calculated as follows:

5. The method according to claim 4, wherein the estimated height of each building in step (3) is calculated by the following formula:

in the formula, H_n,xIs the estimated height of the building.