A method and computer equipment for extracting rooftop photovoltaic distribution images

Abstract

This application discloses a method and computer device for extracting rooftop photovoltaic distribution images. The method comprises: acquiring a remote sensing image, preprocessing the remote sensing image to obtain a preprocessed image, and performing multi-scale segmentation on the preprocessed image according to a multi-scale segmentation algorithm to obtain a segmented image; removing shadow areas from the segmented image to obtain a shadow-free image, and performing band operations on the segmented image to obtain a photovoltaic feature image; fusing the shadow-free image and the photovoltaic feature image to obtain a multi-feature segmented image, and obtaining the rooftop photovoltaic distribution image based on the multi-feature segmented image. This application solves the technical problems of poor accuracy in extracting rooftop photovoltaic distribution images and high computing environment and computational requirements in the prior art.

Description

Technical Field

[0001] The present application relates to the technical field of remote sensing image processing, and in particular to a method and computer equipment for extracting rooftop photovoltaic distribution images. Background Art

[0002] Energy is a crucial material foundation of the national economy and a key area of ​​focus in the 14th Five-Year Plan. During the 14th and 15th Five-Year Plans, my country will continue to optimize the development of wind and solar power. While continuing to promote the construction of centralized bases, it will fully support the development of distributed wind and photovoltaic power generation. Rooftop photovoltaic installations are not restricted by geographical location and are closely integrated with the building, improving the utilization of land resources. Currently, the country strongly supports the development of the photovoltaic industry and provides financial subsidies based on the construction area. Traditionally, the measurement of rooftop photovoltaic construction area is typically done through manual field surveys, which wastes significant manpower and resources and is inefficient. With the rapid development of remote sensing platforms, sensors, communications, and data processing technologies in recent years, remote sensing, as a key means of collecting Earth data and information on its changes, has become capable of providing all-day, all-weather, and comprehensive remote sensing Earth observations of multiple types, resolutions, temporal phases, and angles. The emergence of high-resolution satellite imagery, in particular, has enabled more precise identification of fine surface features.

[0003] Currently, the extraction of photovoltaic distribution images on building rooftops using high-resolution imagery, both domestically and internationally, is typically based on image processing techniques or deep learning algorithms for rooftop photovoltaic identification. However, as strong absorbers, rooftop photovoltaics often suffer from shadow misclassification in practical applications of remote sensing imagery-based rooftop photovoltaic identification. Furthermore, the use of deep learning algorithms for rooftop photovoltaic identification places high demands on the computing environment and requires a large number of training samples. Summary of the Invention

[0004] The technical problem addressed by this application is that, in response to the poor accuracy of existing techniques for extracting rooftop photovoltaic distribution images and their high computing environment and computational requirements, this application provides a method and computer device for extracting rooftop photovoltaic distribution images. The solution provided in the embodiments of this application, on the one hand, determines the area corresponding to rooftop photovoltaics from a multi-feature segmented image with shadows removed, thereby avoiding misclassification of rooftop photovoltaics from shadows in rooftop photovoltaic identification applications, thereby improving the accuracy of rooftop photovoltaic distribution images. Furthermore, it eliminates the need for deep learning algorithms for rooftop photovoltaic identification, reducing the computing environment and computational requirements.

[0005] In a first aspect, an embodiment of the present application provides a method for extracting a rooftop photovoltaic distribution image, the method comprising:

[0006] Acquiring a remote sensing image, preprocessing the remote sensing image to obtain a preprocessed image, and performing multi-scale segmentation on the preprocessed image according to a multi-scale segmentation algorithm to obtain a segmented image;

[0007] removing shadow areas in the segmented image to obtain a shadow-free image, and performing band operation on the segmented image to obtain a photovoltaic characteristic image;

[0008] The shadow-free image and the photovoltaic feature image are fused to obtain a multi-feature segmented image, and the roof photovoltaic distribution image is obtained based on the multi-feature segmented image.

[0009] Optionally, preprocessing the remote sensing image to obtain a preprocessed image includes: performing geometric correction, radiometric calibration, and orthorectification on the remote sensing image to obtain the preprocessed image.

[0010] Optionally, removing the shadow area in the segmented image to obtain a shadow-free image includes: determining the area and brightness of each of one or more objects in the segmented image, and determining the shadow area based on the area and brightness of each object; and removing the shadow area from the segmented image to obtain a shadow-free image.

[0011] Optionally, the shadow area is determined based on the area and brightness of each object, including: separately judging whether the area of ​​each object in one or more objects is greater than a first threshold and the brightness is less than a second threshold; if there is a first object whose area is greater than the first threshold and its brightness is less than the second threshold, then the area corresponding to the first object is the shadow area.

[0012] Optionally, band operation is performed on the segmented image to obtain a photovoltaic characteristic image, including: respectively determining the image value of each object corresponding to the red light band and the blue light band in the segmented image; and subtracting the image value of each object in the red light band from the image value of the object in its corresponding blue light band to obtain the photovoltaic characteristic image corresponding to the object.

[0013] Optionally, the rooftop photovoltaic distribution image is obtained according to the multi-feature segmented image, including: performing photovoltaic-non-photovoltaic pixel-level classification on the multi-feature segmented image according to a support vector machine (SVM) algorithm to obtain a classification result; performing band fusion with the classification result and the photovoltaic feature image, determining the category label corresponding to each object according to the fused image, and obtaining the rooftop photovoltaic distribution image according to the category label, wherein the category label includes rooftop photovoltaic and non-rooftop photovoltaic.

[0014] Optionally, determining the category label corresponding to each object according to the fused image includes: voting on the category label corresponding to each object, and using the category label with the maximum number of votes as the category label of each object.

[0015] In a second aspect, the present application provides a computer device, comprising:

[0016] a memory for storing instructions executed by at least one processor;

[0017] The processor is configured to execute instructions stored in the memory to perform the method described in the first aspect.

[0018] In a third aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions. When the computer instructions are executed on a computer, the computer executes the method described in the first aspect.

[0019] Compared with the prior art, the embodiments of the present application have at least the following beneficial effects:

[0020] In the solution provided in the embodiment of the present application, a segmented image is obtained by performing multi-scale segmentation on the pre-processed image, the shadow area in the segmented image is removed to obtain a shadow-free image, and the segmented image is subjected to band operation to obtain a photovoltaic feature image. The shadow-free image and the photovoltaic feature image are then fused to obtain a multi-feature segmented image, and the roof photovoltaic distribution image is obtained through the multi-feature segmented image. That is, in the solution provided in the embodiment of the present application, on the one hand, by determining the area corresponding to the roof photovoltaic from the multi-feature segmented image after removing the shadow, the misclassification of the roof photovoltaic and the shadow is avoided in the roof photovoltaic identification application, thereby improving the accuracy of the roof photovoltaic distribution image. On the other hand, there is no need to use a deep learning algorithm to identify the roof photovoltaic, which reduces the requirements for the computing environment and computing requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a process for extracting rooftop photovoltaic distribution images provided in an embodiment of the present application;

[0022] Figure 2 A schematic diagram of a simplified process of a method for extracting rooftop photovoltaic distribution images provided in an embodiment of the present application;

[0023] Figure 3 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The embodiments provided in the embodiments of this application are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0025] The following is a further detailed description of a method for extracting rooftop photovoltaic distribution images provided by the embodiment of the present application in conjunction with the accompanying drawings. The specific implementation of the method may include the following steps (the method flow is as follows: Figure 1 shown):

[0026] Step 101: Acquire a remote sensing image, preprocess the remote sensing image to obtain a preprocessed image, and perform multi-scale segmentation on the preprocessed image according to a multi-scale segmentation algorithm to obtain a segmented image.

[0027] As an example, the remote sensing image is a high-resolution remote sensing image, which may include rooftop photovoltaics or not, and is not limited here.

[0028] Furthermore, in the solution provided in the embodiments of the present application, after acquiring the remote sensing image, the remote sensing image needs to be preprocessed. In one possible implementation, preprocessing the remote sensing image to obtain a preprocessed image includes: performing geometric correction, radiometric calibration, and orthorectification on the remote sensing image to obtain the preprocessed image.

[0029] Furthermore, after preprocessing the remote sensing image to obtain the preprocessed image, it is necessary to use a multi-scale segmentation algorithm to perform multi-scale segmentation on the preprocessed image to obtain a segmented remote sensing image. In addition, in the process of performing multi-scale segmentation on the preprocessed image using the multi-scale segmentation algorithm, different scale segmentations can be used according to actual needs. As an example, in the solution provided in the embodiment of the present application, the scale segmentation parameters set for the multi-scale segmentation are: the scale parameter is set to 30, and the shape and compact reading parameters are set to 0.3 and 0.9 respectively.

[0030] Step 102 : removing the shadow area in the segmented image to obtain a shadow-free image, and performing band operation on the segmented image to obtain a photovoltaic characteristic image.

[0031] After performing multi-scale segmentation on the remote sensing image to obtain the segmented image, the acquired remote sensing image will contain shadow areas in addition to the rooftop photovoltaic area. To avoid misclassification of shadow areas with photovoltaic areas, the shadow areas are identified and removed before extracting the rooftop photovoltaic image. There are several ways to remove shadow areas, and the following example illustrates one method.

[0032] In one possible implementation, removing shadow areas in a segmented image to obtain a shadow-free image includes: determining the area and brightness of each of one or more objects in the segmented image, and determining a shadow area based on the area and brightness of each object; and removing the shadow area from the segmented image to obtain a shadow-free image.

[0033] For example, since the segmented image is obtained through a multi-scale segmentation algorithm, the segmented image corresponds to multiple scales, and each scale corresponds to one or more objects. An object is an area composed of multiple pixels, and the pixels between multiple objects do not overlap. Furthermore, in order to determine the shadow area, it is necessary to judge each object. The specific judgment process is as follows:

[0034] First, the area and brightness of each object (e.g., the average pixel value of all pixels contained in the object) are determined, and then, based on the area and brightness of each object, whether the object is a shadow area is determined. In one possible implementation, determining the shadow area based on the area and brightness of each object includes: separately determining whether the area of ​​each of one or more objects is greater than a first threshold and the brightness is less than a second threshold; if there is a first object whose area is greater than the first threshold and whose brightness is less than the second threshold, then the area corresponding to the first object is a shadow area.

[0035] Specifically, in the solution provided in the embodiment of the present application, the brightness threshold (second threshold) and area threshold (first threshold) corresponding to the object are pre-stored. After determining the area and brightness of each object, an algorithm is used to determine whether the area corresponding to any object is a shadow area:

[0036] For object(x)

[0037] IF brightnessobject(x)<T1 AND sizeobject(x)> T2

[0038] THEN object(x)is identified as shadow.

[0039] END

[0040] Where T1 is the second threshold, T2 is the first threshold, and the empirical values ​​of the second and first thresholds are generally 30 and 100; object(x) represents the x-th object on the segmented image; brightnessobject(x) represents the brightness of the x-th object; sizeobject(x) represents the area of ​​the x-th object.

[0041] Furthermore, one or more first objects in the segmented image are determined by the above method, wherein the first object refers to an object whose corresponding area is a shadow area, and the one or more first objects are removed from the segmented image to obtain a shadowless image.

[0042] Furthermore, in order to obtain the photovoltaic distribution image of the roof, in addition to removing the shadow area in the segmented image, it is also necessary to perform band operation on the segmented image to obtain the photovoltaic feature image. The specific process of performing band operation on the segmented image is as follows:

[0043] In one possible implementation, band operations are performed on the segmented image to obtain a photovoltaic characteristic image, including: determining the image value (Digital Number) of each object corresponding to the red light band and the blue light band in the segmented image respectively; and subtracting the image value of each object in the red light band from the image value of the object in its corresponding blue light band to obtain the photovoltaic characteristic image corresponding to the object.

[0044] Specifically, in the solution provided in the embodiment of the present application, the segmented image includes multiple bands, each band includes multiple objects, and there is a certain correspondence between the objects included in different bands. For example, the segmented image includes a red light band and a blue light band, wherein the red light band contains 5 objects, namely object 1, object 2, object 3, object 4, and object 5; the blue light band also contains 5 objects, namely object 1, object 2, object 3, object 4, and object 5. Then, object 1 in the red light band corresponds to object 1 in the blue light band, object 2 in the red light band corresponds to object 2 in the blue light band, object 3 in the red light band corresponds to object 3 in the blue light band, object 4 in the red light band corresponds to object 4 in the blue light band, and object 5 in the red light band corresponds to object 5 in the blue light band. As an example, the image value corresponding to each object here can be the average of the image values ​​of all pixels in the object.

[0045] Furthermore, preliminary experiments revealed that the spectral characteristics of rooftop photovoltaics and buildings have completely opposite spectral variation trends from the blue band to the red band. This spectral pattern provides favorable support for the fine-grained distinction between buildings and rooftop photovoltaics. In view of this, band operations are performed on the segmented high-resolution image to obtain photovoltaic feature images, which serve as auxiliary features for subsequent rooftop photovoltaic identification. The band operation strategy is as follows:

[0046] solar(x)=band blue(x)-band red(x)

[0047] Where band_red(x) represents the image value of the x-th object in the red band of the segmented image, band_blue(x) represents the image value of the x-th object in the blue band of the segmented image, and solar(x) represents the calculated photovoltaic characteristic image.

[0048] Step 103: The shadow-free image and the photovoltaic feature image are fused to obtain a multi-feature segmented image, and the rooftop photovoltaic distribution image is obtained based on the multi-feature segmented image.

[0049] After processing the segmented image to obtain a shadow-free image and a photovoltaic feature image, the shadow-free image and the photovoltaic feature image are fused to obtain a multi-feature segmented image, and then the multi-feature segmented image is processed to obtain a rooftop photovoltaic distribution image. Specifically, the process of processing the multi-feature segmented image to obtain a rooftop photovoltaic distribution image is as follows:

[0050] In one possible implementation, the rooftop photovoltaic distribution image is obtained according to the multi-feature segmented image, including: obtaining the rooftop photovoltaic distribution image according to the multi-feature segmented image, including: performing photovoltaic-non-photovoltaic pixel-level classification on the multi-feature segmented image according to a support vector machine (SVM) algorithm to obtain a classification result; performing band fusion with the classification result and the photovoltaic feature image according to the fused image, determining the category label corresponding to each object according to the fused image, and obtaining the rooftop photovoltaic distribution image according to the category label, wherein the category label includes rooftop photovoltaic and non-rooftop photovoltaic.

[0051] First, the multi-feature segmentation image is classified at the photovoltaic-non-photovoltaic pixel level by the support vector machine algorithm (SVM), wherein the photovoltaic-non-photovoltaic pixel level classification refers to determining the pixels corresponding to photovoltaic and the pixels corresponding to non-photovoltaic in the multi-feature segmentation image, and classifying the pixels in the multi-feature segmentation image according to the pixels corresponding to photovoltaic and the pixels corresponding to non-photovoltaic. For example, the pixels corresponding to photovoltaic and the pixels corresponding to non-photovoltaic can be determined based on the pixel values ​​of the pixels. In addition, the parameters set by the support vector machine algorithm are different for different requirements. For example, the parameters set by the support vector machine algorithm are as follows: the penalty coefficient is set to 100, the RBF kernel function is selected, and the bandwidth is set to 1 / n, where n is the dimension of the input feature, which is not less than the positive integer 1).

[0052] Furthermore, in order to improve the pixel-level classification results, especially to optimize the extracted rooftop photovoltaic boundaries, in one possible implementation method, the category label corresponding to each object is determined based on the fused image, including: voting on the category label corresponding to each object, and using the category label with the maximum number of votes as the category label of each object.

[0053] Specifically, the decision is made according to the following algorithm:

[0054] For object(x)

[0055] Label_object(x)=Max_vote(label(x))

[0056] END

[0057] Where object(x) represents the xth segmented object on the segmented image, and label(x) represents the category label corresponding to the xth segmented object. Max_vote is the maximum vote for the category labels of the x segmented objects, and the winning category label is selected as the category label of the object. According to the above decision rules, the pixel-level classification results and the segmented image are fused object-oriented to obtain the object-oriented classification results and the final rooftop photovoltaic distribution image. For details, see Figure 2 , which is a brief flow chart of a method for extracting rooftop photovoltaic distribution images provided in an embodiment of the present application.

[0058] In addition, in the solution provided in the embodiment of the present application, it has been mentioned above that the remote sensing image may or may not include rooftop photovoltaics. When the remote sensing image includes rooftop photovoltaics, the rooftop photovoltaic distribution image can be obtained according to the above method; when the remote sensing image does not include rooftop photovoltaics, the rooftop photovoltaic distribution image will not be obtained. In order to prompt the user when the rooftop photovoltaic distribution image is not obtained, in one possible implementation method, the rooftop photovoltaic distribution image is obtained based on the multi-feature segmented image, and further includes: if the remote sensing image does not include rooftop photovoltaics, a prompt message is generated, and the prompt message is used to indicate that there is no rooftop photovoltaic in the remote sensing image.

[0059] In the solution provided in the embodiment of the present application, a segmented image is obtained by performing multi-scale segmentation on the pre-processed image, the shadow area in the segmented image is removed to obtain a shadow-free image, and the segmented image is subjected to band operation to obtain a photovoltaic feature image. The shadow-free image and the photovoltaic feature image are then fused to obtain a multi-feature segmented image, and the roof photovoltaic distribution image is obtained through the multi-feature segmented image. That is, in the solution provided in the embodiment of the present application, on the one hand, by determining the area corresponding to the roof photovoltaic from the multi-feature segmented image after removing the shadow, the misclassification of the roof photovoltaic and the shadow is avoided in the roof photovoltaic identification application, thereby improving the accuracy of the roof photovoltaic distribution image. On the other hand, there is no need to use a deep learning algorithm to identify the roof photovoltaic, which reduces the requirements for the computing environment and computing requirements.

[0060] See also Figure 3 The present application provides a computer device, comprising:

[0061] Memory 301, used to store instructions executed by at least one processor;

[0062] Processor 302, configured to execute instructions stored in the memory Figure 1 The method described.

[0063] The present application provides a computer-readable storage medium storing computer instructions. When the computer instructions are executed on a computer, the computer executes Figure 1 The method described.

[0064] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage and optical storage, etc.) that contain computer-usable program code.

[0065] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0066] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0068] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for extracting rooftop photovoltaic distribution images, characterized in that: include: Acquiring a remote sensing image, preprocessing the remote sensing image to obtain a preprocessed image, and performing multi-scale segmentation on the preprocessed image according to a multi-scale segmentation algorithm to obtain a segmented image; removing shadow areas in the segmented image to obtain a shadow-free image, and performing band operation on the segmented image to obtain a photovoltaic characteristic image; fusing the shadow-free image and the photovoltaic feature image to obtain a multi-feature segmented image, and obtaining the rooftop photovoltaic distribution image based on the multi-feature segmented image; in: Removing the shadow area in the segmented image to obtain a shadow-free image includes: determining an area and brightness of each of the one or more objects in the segmented image, and determining a shadow area based on the area and brightness of each object; removing the shadow area from the segmented image to obtain a shadow-free image; Further, determining the shadow area according to the area and brightness of each object includes: Determining whether the area of ​​each of the one or more objects is greater than a first threshold and the brightness is less than a second threshold; If there is a first object whose area is larger than the first threshold and whose brightness is smaller than the second threshold, the area corresponding to the first object is a shadow area; Performing band operations on the segmented image to obtain a photovoltaic feature image, including: respectively determining the image value of each object corresponding to the red light band and the blue light band in the segmented image; Subtracting the image value of each object in the red light band from the image value of the object corresponding to the blue light band to obtain a photovoltaic characteristic image corresponding to the object; Obtaining the rooftop photovoltaic distribution image according to the multi-feature segmented image includes: Performing photovoltaic-non-photovoltaic pixel-level classification on the multi-feature segmented image according to a support vector machine (SVM) algorithm to obtain a classification result; Performing band fusion on the classification result and the photovoltaic feature image, determining a category label corresponding to each object based on the fused image, and obtaining the rooftop photovoltaic distribution image based on the category label, wherein the category label includes rooftop photovoltaic and non-rooftop photovoltaic; Furthermore, determining the category label corresponding to each object according to the fused image includes: performing maximum voting on the category label corresponding to each object and selecting the winning category label as the category label of the object.

2. The method according to claim 1, wherein Preprocessing the remote sensing image to obtain a preprocessed image includes: The remote sensing image is subjected to geometric correction, radiometric calibration and orthorectification to obtain the pre-processed image.

3. A computer device, characterized in that: include: a memory for storing instructions executed by at least one processor; A processor, configured to execute instructions stored in a memory to perform the method according to any one of claims 1 to 2.

4. A computer-readable storage medium, characterized in that include: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 2.

Images