Urban mass building age efficient acquisition method

Abstract

The invention relates to an efficient acquisition method for urban mass building years, which is characterized in that on the basis of a satellite remote sensing image, building screening and building age determination in a target area can be quickly obtained through the steps of remote sensing image preprocessing, identification model training, building base identification, picture vectorization, data cleaning, building age determination and the like. The method can effectively improve the precision and efficiency of parameter generation required by urban building energy consumption calculation, and plays a positive role in scientific development of low-carbon city construction and realization of a dual-carbon target.

Description

Technical Field

[0001] The present invention belongs to the field of measurement technology, in particular to the design of environmental protection and energy saving of urban buildings, and specifically provides a method for quickly and efficiently obtaining the ages of massive urban buildings. Background Art

[0002] Global warming threatens the survival of all humanity. Low-carbon city development is a key solution for mitigating global warming and achieving carbon neutrality. Whether in existing urban renewal or incremental construction, scientific policy guidance is essential. Calculating urban building energy consumption can empower rational policy formulation, but accurately setting its parameters is extremely difficult, a bottleneck restricting the further development of this technology.

[0003] Currently, methods for setting parameters required for calculating urban building energy consumption include the prototype building method and the urban characteristic method. The prototype building method categorizes urban buildings into several types based on age and function, then assigns values to each prototype type according to standards and guidelines. The urban characteristic method leverages big data on urban buildings and, through machine learning, establishes relationships between various parameters and urban characteristics, including age, area, and population density. Therefore, building age is a key parameter required for calculating urban building energy consumption. For individual buildings, age information can be obtained through online research or on-site consultation, which is manageable and has a high probability of success. However, at the urban scale, with thousands or tens of thousands of buildings, these methods are difficult to replicate. Existing research on urban building energy consumption calculations primarily relies on government collaboration to obtain building data and extract age information. This method is not universally applicable and is typically applied only to a few hundred buildings, making it difficult to meet demand.

[0004] With the "dual carbon" goals approaching, urban building energy consumption calculation technology must shift from theoretical to practical applications to truly promote low-carbon city construction. This requires: 1) universal building age acquisition methods, not limited to government data; and 2) applicable to thousands or tens of thousands of buildings. This requires continuous research to explore solutions that meet these needs. Summary of the Invention

[0005] In order to address the deficiencies of the prior art, the present invention proposes a method for efficiently acquiring the ages of massive urban buildings to meet the needs of urban building energy consumption calculation.

[0006] The technical problem to be solved by the present invention is achieved through the following technical solutions:

[0007] A method for efficiently obtaining the ages of massive urban buildings comprises the following steps:

[0008] S1: Remote sensing image preprocessing: obtain remote sensing images of the target area in different years and preprocess them;

[0009] S2: Recognition model training: Using the database, train an image recognition model that can identify building bases in remote sensing images;

[0010] S3: Building base identification: Relying on the identification model, obtain the building base in remote sensing images of different ages in the target area;

[0011] S4: Image vectorization and data cleaning: vectorize the rasterized recognition images and delete the "noise" buildings in the images;

[0012] S5: Determination of building age: Compare the building base plane layers of each age, and take the year when the "newly added" building base first appears as the age of the corresponding building.

[0013] In the present invention, the remote sensing image preprocessing method in S1 includes: radiometric calibration, atmospheric correction, orthorectification and remote sensing image fusion.

[0014] In the present invention, the building base recognition model in S2 is constructed based on the U-net algorithm, and the samples required for its training include two types of databases: public databases and private databases. Among them, the public database must contain several labeled building base photos to enable the model to initially have the ability to recognize building bases. The private database relies on the remote sensing image preprocessed in S1, divides it into several 512×512 pixel images, and randomly selects a certain number of images for pixel-level annotation to form a training library to enable the model to have the ability to recognize building bases in specific areas.

[0015] In the present invention, the image vectorization operation in S4 includes: resampling, reclassification and vectorization. The data cleaning operation is for areas less than 300m 2 Delete the building base.

[0016] In the present invention, the building age determination in S5 includes two situations: accurate building age determination and building age time period determination: if the "new additions" of satellite remote sensing building base are compared year by year, the building age can be accurately determined; if the "new additions" of satellite remote sensing building base with a time span are compared, the time period of the building age can be obtained.

[0017] Furthermore, the determination of “newly added” buildings is based on the following two criteria: 1) When the building base centroid in the new remote sensing image does not appear in the building base plane in the old remote sensing image, it is determined to be a “newly added” building; 2) Although the building base centroid in the new remote sensing image appears in the building base plane in the old remote sensing image, it is also determined to be a “newly added” building when it satisfies formula (1):

[0018]

[0019] Where S 交 is the intersection area of the building base in the new remote sensing image and the old remote sensing image (m 2 ), S 并 is the union area of the building base in the new remote sensing image and the old remote sensing image (m 2 ).

[0020] Compared with the prior art, the present invention has the following advantages:

[0021] (1) The implementation cost of the present invention is low: the data required by the present invention is only open remote sensing images, and the implementation cost is low, which has a price advantage;

[0022] (2) The present invention has wide coverage: The present invention can obtain the age information of thousands or tens of thousands of urban buildings, with a wide coverage;

[0023] (3) The present invention has high accuracy: the present invention can obtain building age information more accurately, which is conducive to improving the accuracy of urban building energy consumption calculation;

[0024] (4) The present invention has high implementation efficiency: the present invention can automatically obtain the building age, with less manual intervention and high overall efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is the overall flow chart of the present invention;

[0026] Figure 2 This is a schematic diagram of satellite remote sensing image preprocessing according to the present invention;

[0027] Figure 3 Examples of public databases (a) and private databases (b) used for recognition training of the present invention;

[0028] Figure 4 The present invention is based on Figure 2 Schematic diagram of the identified building base;

[0029] Figure 5 The present invention is based on Figure 4 Schematic diagram of image vectorization and processing and cleaning;

[0030] Figure 6 A schematic diagram illustrating the newly added building according to the present invention. DETAILED DESCRIPTION

[0031] The present invention will be further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the scope of protection of the present invention is not limited thereby.

[0032] In order to show the embodiment of the present invention more clearly and easily, the present invention selects a certain district in a certain city (district a in city A) to describe the embodiment. Figure 1 As shown, the present invention proposes a method for efficiently obtaining the ages of massive urban buildings, and the specific method includes the following steps:

[0033] S1: Remote sensing image preprocessing:

[0034] The remote sensing image acquired this time is WorldView-3, with a panchromatic resolution of 0.3m / pixel and a multispectral resolution of 1.2m / pixel. Figure 1 The area included in is a 5 km × 5 km area selected from District A, City A. The remote sensing image is subjected to radiometric calibration, atmospheric correction, orthorectification, and remote sensing image fusion using ERDAS Imagine software, ultimately obtaining a color remote sensing image with a resolution of 0.3 m / pixel, as shown in the figure. Figure 2 shown.

[0035] S2: Recognition model training:

[0036] The building base recognition model is built based on the U-net algorithm, and its training is carried out based on public and private databases. The public database is the Wuhan University database, such as Figure 3 The example in (a) is obtained from the address http: / / gpcv.whu.edu.cn / data / building_dataset.html. 10,000 labeled photos are selected for preliminary training to enable the model to initially have the ability to recognize building bases. Private databases such as Figure 3 (b) It is constructed based on the WorldView-3 remote sensing image used, which is segmented into 1050 512×512 pixel images. 200 of them are selected for pixel-level annotation to form a training library, enabling the model to have the ability to identify the building base in a specific area.

[0037] S3: Building base identification:

[0038] Use the trained building base recognition model to identify the building bases in remote sensing images of different ages in the target area and obtain rasterized recognition images, such as Figure 4 shown.

[0039] S4: Image vectorization and data cleaning:

[0040] Image vectorization and data cleaning are implemented based on GIS (Geographic Information System): First, use "resampling" to reduce the grid side length of the rasterized image to 1 / 10 of the original. Then, use "reclassification" to adjust the RGB band values in the image from the original 9 categories to 2 categories to achieve image "binarization" classification (one category represents the building base and the other category represents the background). Then, use "vectorization" to convert the rasterized building base image into a vectorized building base plane. Finally, delete the base area less than 300m 2 "Noisy" buildings, such as Figure 5 shown.

[0041] S5: Determination of building age:

[0042] There are two situations for determining the age of a building: if the "new additions" of the building base in satellite remote sensing images are compared year by year, the specific age of the building can be obtained; if the "new additions" of the building base in satellite remote sensing images are not compared year by year, the time period to which the age of the building belongs can be obtained.

[0043] For "newly added" buildings, such as Figure 6 As shown in the figure, there are mainly two situations: 1) newly built buildings on vacant land, and 2) newly built buildings after the demolition of old buildings or temporary buildings. For situation 1: use the "element conversion point" in GIS to extract the center of gravity of the building base in the new remote sensing image, and then use "spatial connection" to link it to the old remote sensing image layer. If the center of gravity does not fall on the building base plane of the layer, then the corresponding building is a "new" building; for situation 2: merge the building bases in the new and old remote sensing images, and use the "intersection" and "union" functions of GIS to calculate S respectively. 交 and S 并 , calculate the ratio of the two one by one, if the following formula (1) is satisfied, that is, S 交 With S 并 If the ratio is less than 75%, the building is considered a "new" building:

[0044]

[0045] Where S 交 is the intersection area of the building base in the new remote sensing image and the old remote sensing image (m 2 ), S 并 is the union area of the building base in the new remote sensing image and the old remote sensing image (m 2 ).

[0046] Take the 5km×5km area selected in District A of City A as an example. According to statistics, the area contained 2,175 buildings in 2020. Figure 2 、 Figure 4 and Figure 5 After image recognition, vectorization, and data cleaning, 2,103 buildings were obtained, with a recognition rate of 96.7%. From 2017 to 2020, 217 new buildings were added to the area, and the ages of 203 of them could be accurately identified, with an accuracy rate of 93.5%.

Claims

1. A method for efficiently obtaining the ages of massive urban buildings, characterized by: The specific steps include: S1: Remote sensing image preprocessing: including radiometric calibration, atmospheric correction, orthorectification and remote sensing image fusion; S2: Recognition model training: training an image recognition model that can identify building bases in remote sensing images; S3: Building base identification: Relying on the identification model, obtain the building base in remote sensing images of different ages in the target area; S4: Image vectorization and data cleaning: vectorize the rasterized recognition images and remove "noise" buildings in the images; S5: Determination of building age: Compare the base plane layers of buildings from different eras, and use the year when the "newly added" building base first appears as the age of the corresponding building.

2. The method for efficiently obtaining the ages of massive urban buildings according to claim 1 is characterized by: The method for preprocessing the remote sensing images in S1 includes preprocessing remote sensing images of the target area at different times.

3. The method for efficiently obtaining the ages of massive urban buildings according to claim 1 is characterized by: The building base recognition model in S2 is built based on the U-net algorithm and trained using public and private databases.

4. The method for efficiently acquiring the ages of massive urban buildings according to claim 3 is characterized by: The public database must contain several labeled photos of building bases to enable the model to initially have the ability to identify building bases. The private database relies on the remote sensing images preprocessed in S1, which are divided into several 512×512 pixel images, and a certain number of images are randomly selected for pixel-level annotation to form a training library.

5. The method for efficiently obtaining the ages of massive urban buildings according to claim 1 is characterized by: The recognition model in S3 has the ability to recognize the building base in a specific area.

6. The method for efficiently acquiring the ages of massive urban buildings according to claim 1 is characterized by: In the image vectorization in S4, the data cleaning operation is performed on images with an area less than 300m 2 Delete the building base.

7. The method for efficiently obtaining the ages of massive urban buildings according to claim 1 is characterized by: The building age determination in S5 includes two situations: accurate building age determination and building age time period determination: if the "new additions" of satellite remote sensing building base are compared year by year, the building age can be accurately determined; if the "new additions" of satellite remote sensing building base with a time span are compared, the time period of the building age can be obtained.

8. The method for efficiently acquiring the ages of massive urban buildings according to claim 7 is characterized by: The determination of "newly added" buildings is based on the following two criteria: 1) When the building base centroid in the new remote sensing image does not appear in the building base plane in the old remote sensing image, it is determined to be a "newly added" building; 2) Although the building base centroid in the new remote sensing image appears in the building base plane in the old remote sensing image, it is also determined to be a "newly added" building when formula (1) is satisfied: Where S 交 is the intersection area of the building base in the new remote sensing image and the old remote sensing image (m 2 ), S 并 is the union area of the building base in the new remote sensing image and the old remote sensing image (m 2 ).

Images