Road domain water body recognition method, device and equipment based on unmanned aerial vehicle multispectral image, medium and product

Abstract

The invention discloses a road domain water body recognition method and device based on unmanned aerial vehicle multispectral images, equipment, a medium and a product, and relates to the technical field of image processing, and the method comprises the steps: obtaining a plurality of unmanned aerial vehicle multispectral images of a target road domain; processing the plurality of unmanned aerial vehicle multispectral images to obtain images containing different wavebands; performing water body and non-water body segmentation on the image map by adopting a threshold segmentation method to obtain an initial water body region image; preprocessing the initial water body area image to obtain the initial water body area image; calculating a normalized water body index of the initial water body region image; performing water body region extraction on the initial water body region image by using the normalized water body index to obtain a final water body region; the final water body area is converted into vector data, area calculation is carried out on planar elements in the vector data, the water body area of the target road domain is obtained, the multi-spectral image can be obtained through the unmanned aerial vehicle, and the road domain detection efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, device, equipment, medium and product for identifying road water bodies based on multispectral images of unmanned aerial vehicles. Background Art

[0002] With the rapid development of society, the demand for infrastructure construction in remote areas is increasing, especially in plateau areas. The construction of frozen soil roadbed is of great significance to ensure the smooth flow of transportation network. The stability of existing frozen soil roadbed is seriously affected by waterlogging, which not only threatens the safe operation of roads, but also has an adverse impact on the ecological environment. In addition, the distribution of ground monitoring stations is limited, the construction and operation costs are high, and the flexibility of monitoring data is low. Summary of the invention

[0003] The purpose of this application is to provide a method, device, equipment, medium and product for road area water body identification based on drone multispectral images, which can use drones to obtain multispectral images and improve road area detection efficiency.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] In a first aspect, the present application provides a method for identifying road water bodies based on multispectral images of unmanned aerial vehicles, and the method for identifying road water bodies based on multispectral images of unmanned aerial vehicles comprises:

[0006] Acquire multiple UAV multispectral images of the target road area;

[0007] Process multiple UAV multispectral images to obtain images containing different bands;

[0008] The image is segmented into water bodies and non-water bodies using a threshold segmentation method to obtain an initial water body area image;

[0009] Calculating a normalized water index of the initial water region image;

[0010] Using the normalized water body index to extract the water body region from the initial water body region image to obtain a final water body region;

[0011] The final water body area is converted into vector data, and the area of ​​the surface elements in the vector data is calculated to obtain the water body area of ​​the target road area.

[0012] Optionally, calculating the normalized water index of the initial water body region image specifically includes:

[0013] Extracting spectral features of a set band image in the initial water body area image;

[0014] A normalized water body index of the initial water body region image is calculated according to the spectral characteristics.

[0015] Optionally, the threshold segmentation method is a segmentation method based on a fixed threshold, a segmentation method based on an automatic threshold, or a segmentation method based on a multi-scale object-oriented method.

[0016] Optionally, the automatic threshold-based segmentation method is Otsu's global threshold method.

[0017] Optionally, the image map includes four bands, namely a green band, a red band, a near infrared band and a red edge band.

[0018] Optionally, multiple drone multispectral images are processed to obtain image maps containing different bands, specifically including:

[0019] The band alignment, reflectivity calibration and synthesis of multiple UAV multispectral images are performed in sequence to obtain images containing different bands.

[0020] In a second aspect, the present application provides a road water body recognition device based on drone multispectral images, and the road water body recognition device based on drone multispectral images includes:

[0021] UAV multispectral image acquisition module, used to acquire multiple UAV multispectral images of the target road area;

[0022] The image generation module containing different bands is used to process multiple UAV multispectral images to obtain image maps containing different bands;

[0023] An initial water body region image determination module is used to segment the image into water bodies and non-water bodies using a threshold segmentation method to obtain an initial water body region image;

[0024] A normalized water index calculation module, used to calculate the normalized water index of the initial water area image;

[0025] A final water body region determination module, used to extract the water body region from the initial water body region image using the normalized water body index to obtain a final water body region;

[0026] The water body area calculation module is used to convert the final water body area into vector data, calculate the area of ​​the surface elements in the vector data, and obtain the water body area of ​​the target road area.

[0027] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any of the above-mentioned methods for identifying road areas based on multispectral images of unmanned aerial vehicles.

[0028] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned methods for identifying road areas water bodies based on multispectral images of unmanned aerial vehicles.

[0029] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned methods for identifying road areas based on multispectral images of unmanned aerial vehicles.

[0030] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0031] The present application provides a method, device, equipment, medium and product for identifying water bodies in road areas based on multispectral images of unmanned aerial vehicles. A plurality of multispectral images of unmanned aerial vehicles are processed to obtain an image map containing different bands; a threshold segmentation method is used to segment the image map into water bodies and non-water bodies to obtain an initial water body area image; the normalized difference water index (Normalized Difference Water Index, NDWI) is used to extract the water body area of ​​the initial water body area image to obtain the final water body area, thereby realizing the determination of the final water body area through two water body area extractions based on multispectral images of unmanned aerial vehicles, realizing rapid detection of road areas in remote areas, and improving the efficiency of road area detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 A schematic diagram of a flow chart of a method for identifying road water bodies based on multispectral images of unmanned aerial vehicles provided in one embodiment of the present application;

[0034] Figure 2 A schematic diagram of the principle of a method for identifying road water bodies based on multispectral images of unmanned aerial vehicles provided in one embodiment of the present application;

[0035] Figure 3is the original visible light image;

[0036] Figure 4 A road water body recognition method based on multispectral images of unmanned aerial vehicles is provided in an embodiment of the present application. Figure 3 Schematic diagram of the area extracted after identifying the road area water body;

[0037] Figure 5 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0040] This application provides a road water body recognition method based on UAV multispectral images. Figure 1 and Figure 2 As shown, the road water body recognition method based on UAV multispectral image includes:

[0041] Step 101: Acquire multiple UAV multispectral images of the target road area.

[0042] Step 102: Process multiple UAV multispectral images to obtain image maps containing different bands.

[0043] Step 103: Segment the image into water bodies and non-water bodies using a threshold segmentation method to obtain an initial water body region image.

[0044] Step 104: Calculate the normalized water index of the initial water region image.

[0045] Step 105: extracting the water area from the initial water area image using the normalized water index to obtain a final water area.

[0046] Step 106: Convert the final water body area into vector data, calculate the area of ​​the surface elements in the vector data, and obtain the water body area of ​​the target road area.

[0047] Based on drone multispectral images, this application determines the final water area through two water area extractions, which can achieve rapid detection of road areas in remote areas and improve the efficiency of road area detection. In addition, this application has the advantages of low cost, wide monitoring range and flexible monitoring area.

[0048] In an exemplary embodiment, the multispectral image of the UAV in step 101 has a high resolution, thereby improving the accuracy of the water body extraction result.

[0049] In an exemplary embodiment, step 102 specifically includes: sequentially performing band alignment, reflectance calibration and synthesis on multiple drone multispectral images to obtain image maps containing different bands. More specifically, a multispectral camera is mounted on a drone and then photographs the water body to obtain original data, the original data is a single band, a single photo, and the original single-band drone multispectral data is band aligned, reflectance calibrated, and synthesized to generate images of different bands. The original data can be stored in a drone or a drone system or a drone cloud platform.

[0050] A multispectral camera is a device that can capture light of different wavelengths. It can decompose the captured light into different colors to obtain rich spectral information. Multispectral cameras carried on drones can be used in water body detection, vegetation cover classification, pest and disease monitoring, geological and soil surveys and other fields. For example, in water body detection, by analyzing the reflectance spectra of different bands, the existence of water bodies and their water quality conditions can be identified. Specifically, the normalized difference water index (NDWI) can be used to highlight water body characteristics, help identify water body boundaries and monitor water body changes. In this application, the multispectral camera used has 4 bands, namely the green band, red band, near infrared band and red edge band.

[0051] The image map specifically includes four bands, namely the green band, red band, near-infrared band and red edge band.

[0052] After spectrally synthesizing the collected raw data, a spectral image of the water body is generated. Each water body image sample contains multiple attribute features such as grayscale value, spectral characteristics, shape characteristics, and texture characteristics. Based on these sample data sets, the grayscale values ​​of pixels in the image are calculated and traversed through a grid search strategy to identify and judge the water body area. The grid search algorithm can exhaustively enumerate all possible parameter combinations, and use the hyperparameter combination with the highest average score as the optimal parameter set for model optimization. By analyzing and optimizing the grayscale values ​​of each pixel position in the image, a water body extractor based on the threshold method is constructed to improve the accuracy of water body recognition.

[0053] In an exemplary embodiment, in step 103, the threshold segmentation method is a fixed-threshold based segmentation method, an automatic-threshold based segmentation method, or a multi-scale object-oriented segmentation method.

[0054] Perform density segmentation on the calculated normalized index of the image map, and visually compare through the distribution characteristics of the histogram to obtain a suitable segmentation threshold, that is, the set threshold in the fixed-threshold based segmentation method. Generally, the set threshold can be selected as -0.2.

[0055] Analyze the normalized index, mainly using two bands, band Green and band NIR. Generally, the normalized index exhibited by water bodies is less than 0. In the fixed-threshold based segmentation method, first perform NDWI histogram statistics, and select a suitable threshold based on the histogram. The automatic-threshold based segmentation method uses the Otsu global threshold method, which can automatically calculate a suitable threshold. However, it is suitable for the situation where there are obvious "peaks and valleys" in the pixel histogram of the image, and the greater the difference between the peak value and the valley value, the better the water body extraction effect, which has certain limitations. In the multi-scale object-oriented segmentation method, an additional band data (band RED) is added for rough extraction and secondary extraction, which can effectively process images with complex backgrounds. The threshold selected by the multi-scale object-oriented segmentation method is 0, and a very high accuracy can be achieved.

[0056] The calculation formula of the fixed-threshold based segmentation method is:

[0057]

[0058] Where B(x, y) represents the output binary image, T is the set threshold, I(x, y) represents the pixel value of the input image map at position (x, y), B(x, y) = 1 represents water body, and B(x, y) = 0 represents non-water body.

[0059] The automatic-threshold based segmentation method is the Otsu global threshold method. The automatic-threshold based segmentation method is applicable to the situation where there are obvious "peaks and valleys" in the pixel histogram of the image, and the greater the difference between the peak value and the valley value, the better the water body extraction effect.

[0060] The Otsu global threshold method specifically includes the following steps: Assume the image has L gray levels, and n i (0 < i < L - 1) is the number of pixels at gray level i, then the total number of pixels N is:

[0061] Normalize the gray histogram:

[0062] Then the probability of the gray level i of the pixel points in the image is:

[0063] Select a threshold T and divide the pixels into two intervals X0 and X1 according to different gray levels. X0 is the pixel with gray level in [0, T-1], and X1 is the pixel with gray level in [T, L-1]. Then the probability of the pixel falling into the two intervals is:

[0064] The grayscale means of X0 and X1 are: The between-class variance of X0 and X1 is: 2 =x0(μ0-μ) 2 +x1(μ1-μ) 2 =x0x1(μ1-μ0).

[0065] Among them, n i is the number of pixels at gray level i, p i is the probability of gray level i appearing, X0 is the pixel with gray level in [0, T-1], X1 is the pixel with gray level in [T, L-1], μ is the gray mean of the image, μ0 is the gray mean of X0, μ1 is the gray mean of X1, σ 2 is the between-class variance of X0 and X1.

[0066] Based on the multi-scale object-oriented segmentation method, the whole scene image is segmented into small objects of different sizes by setting relevant parameters through the object-oriented segmentation method. This module has high extraction accuracy, but it has high requirements on the time and computer hardware configuration required in the extraction process. The first segmentation and the second water body segmentation both use the Quick Shift segmentation method. QuickShift is a density-based image segmentation and clustering algorithm, which is mainly used for image clustering and feature extraction. It is an unsupervised learning method that draws on the Density Peak Clustering and Gradient Ascent methods. Unlike traditional segmentation methods (such as those based on thresholds, edge detection or region growing), QuickShift calculates the local density of pixels and finds density peaks to perform image segmentation.

[0067] The QuickShift algorithm performs image segmentation through the following steps:

[0068] (1) Calculate the distance between pixels. i and x j , calculate the Euclidean distance d(x i , x j ):

[0069]

[0070] In the formula, x iand x j is the pixel value. The NDWI water index data is used in this software.

[0071] (2) Gaussian kernel function. The Gaussian kernel function is used to measure the similarity between pixels. The Gaussian kernel function is defined as:

[0072]

[0073] Among them, σ is a parameter that controls the width of the kernel function and needs to be defined according to image features.

[0074] (3) Local density calculation. The local density of each pixel is calculated based on other pixels in the surrounding neighborhood. i , whose local density ρ(x i ) can be expressed as:

[0075]

[0076] Among them, K(x i , x j ) is the pixel x i and x j The similarity between them, ρ(x i ) reflects the pixel x i Relative density to its area.

[0077] (4) Displacement calculation (shift)

[0078] For each pixel x i , quickshift calculates the pixel density in the field, determines that the point migrates to the area with higher density, and updates the position of each pixel by calculating the gradient (i.e., density direction). Assume y i is the pixel x i The migration location is:

[0079]

[0080] Where η is the step size, is the local density gradient (the direction points to areas of higher density).

[0081] (5) For each pixel point, QuickShift will continuously adjust its position during the iteration process until the point no longer undergoes significant displacement, that is, it reaches the peak position of the local density.

[0082] Threshold algorithms (fixed-threshold-based segmentation methods and automatic-threshold-based segmentation methods) are processes that classify pixels in an image by setting different thresholds. Each time image segmentation is performed, different threshold combinations are searched in the given pixel value space to achieve the optimal image segmentation effect. In the process of constructing a threshold segmentation model, randomness is mainly reflected in the steps of selecting different threshold ranges and applying local threshold segmentation in different regions. The specific threshold selection process is as follows:

[0083] Step 1: In the entire image pixel set P, there are a total of M pixel values. For each segmentation, a partial pixel region is randomly selected as the basis for calculating the current local threshold. For the threshold selection of each local region, a random selection method is adopted, so that each pixel has the possibility of being selected for calculating the local threshold or not being selected at all. As the number of local regions increases and the number of traversals increases, most pixel regions will eventually be included in the threshold calculation process.

[0084] Step 2: Assume that the total number of pixel features in each local region is f. When calculating each local threshold, k features (k < f) are randomly selected, and the optimal local threshold is calculated based on the gray values or other attributes of these features. Different local regions will calculate different local thresholds according to different random feature combinations, thereby achieving the optimization of image segmentation.

[0085] In this way, the threshold algorithm can flexibly select and optimize different thresholds at the global and local levels, thereby improving the accuracy of image segmentation.

[0086] After generating a water body extractor based on the threshold algorithm, the band spectral image is input into the water body extractor. The water body extractor calculates the Normalized Difference Water Index (NDWI) based on each pixel feature in the band spectral image to form a binary image. The threshold algorithm classifies the binary image by setting a specific threshold. In this embodiment, the classification result of the binary image of water is 1, and the classification result of other backgrounds is 0. According to the category of each pixel in the image, the water body area is marked, and finally a complete water body mask area is obtained.

[0087] In the statistical monitoring example of the accumulated water area of the G214 line along the permafrost area of the Qinghai-Tibet Plateau for multiple periods and multiple times. When the weather is good, high-quality and high-resolution multispectral image data can be collected. Since the reflectivity of water is small, it appears white or transparent gray as a whole in the visible light band of the image. The extraction result of the water body area by this algorithm is correct, and the accumulated water range and area are statistically calculated, which coincide with the accumulated water range in the high-quality and high-resolution visible light photos, and the coincidence rate reaches 90% or more. The example analysis results are as Figure 3 and Figure 4As shown, the pink area is the waterlogged area identified by a road water body identification method based on UAV multispectral imaging in this application. Compared with the original visible light image, it has a high degree of overlap.

[0088] In an exemplary embodiment, before the image is segmented by the segmentation method in step 103, the image is preprocessed first, and then the preprocessed image is segmented to obtain an initial water body area image.

[0089] The preprocessing in the segmentation method based on fixed threshold and the segmentation method based on automatic threshold includes image enhancement and image denoising.

[0090] For the multi-scale object-oriented segmentation method, the preprocessing includes image enhancement, image denoising, image reading, band operation and density segmentation.

[0091] In the process of water body extraction based on multi-scale object-oriented method, three band images need to be imported. Band operation and density segmentation refer to synthesizing the three input three-band images after importing the images, and slicing the images according to the set parameters. Image reading, band operation and density segmentation are carried out in sequence, and image enhancement and image denoising are mainly used to process the images after reading them.

[0092] In an exemplary embodiment, step 104 specifically includes: extracting spectral features of a set band image in the initial water body area image; and calculating a normalized water body index of the initial water body area image according to the spectral features.

[0093] The spectral characteristics of the set band image include reflectance in the green, red, and near-infrared bands.

[0094] Extracting the spectral characteristics of the band image and calculating the normalized water index specifically include:

[0095] Read the appropriate band image and extract its spectral characteristics.

[0096] The band image is divided into multiple local areas, and the local threshold of the normalized water index of each local area is calculated.

[0097] Based on the calculated local threshold of the normalized water index of each local area, the pixels in each local area are binarized.

[0098] Select a suitable threshold segmentation algorithm to realize the binary segmentation of the image.

[0099] The normalized water index calculation formula is expressed as:

[0100] Among them, NDWI represents the normalized water index, ρ Green represents the reflectance of the green band, ρ NIR Represents the reflectivity in the near-infrared band.

[0101] In an exemplary embodiment, step 105 specifically includes: extracting a water body area using a normalized water body index, marking the extracted water body area, generating a water body mask according to the marked position, and obtaining a final water body area in the form of raster map data according to the water body mask.

[0102] In an exemplary embodiment, step 106 is the process of converting raster data into vector data, specifically converting the final water body area in the form of raster map data into vector map data (vector data), calculating the area of ​​the surface elements in the vector data, and obtaining the water body area of ​​the target road area.

[0103] Based on the same inventive concept, the embodiment of the present application also provides a road water body identification device based on drone multispectral images for implementing the road water body identification method based on drone multispectral images mentioned above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the road water body identification device based on drone multispectral images provided below can refer to the limitations of the road water body identification method based on drone multispectral images above, and will not be repeated here.

[0104] In an exemplary embodiment, the present application provides a road water body recognition device based on drone multispectral images, comprising:

[0105] The UAV multispectral image acquisition module is used to acquire multiple UAV multispectral images of the target road area.

[0106] The image map generation module containing different bands is used to process multiple UAV multispectral images to obtain image maps containing different bands.

[0107] The initial water body area image determination module is used to segment the image into water bodies and non-water bodies using a threshold segmentation method to obtain an initial water body area image.

[0108] A normalized water index calculation module, used to calculate the normalized water index of the initial water area image;

[0109] The final water body region determination module is used to extract the water body region from the initial water body region image using the normalized water body index to obtain the final water body region.

[0110] The water body area calculation module is used to convert the final water body area into vector data, calculate the area of ​​the surface elements in the vector data, and obtain the water body area of ​​the target road area.

[0111] The present application provides a road area water body identification device based on unmanned aerial vehicle multispectral images, which also includes a unmanned aerial vehicle, and the unmanned aerial vehicle is used to collect multispectral images of target road areas.

[0112] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store road water body identification data based on drone multispectral images. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a road water body identification method based on drone multispectral images is implemented.

[0113] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0114] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0115] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0116] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0117] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0118] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a data processing logic of a programmable logic device, etc., but is not limited thereto.

[0119] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0120] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A road water body recognition method based on UAV multispectral images, characterized in that: The road water body recognition method based on UAV multispectral images includes: Acquire multiple UAV multispectral images of the target road area; Process multiple UAV multispectral images to obtain images containing different bands; The image is segmented into water bodies and non-water bodies using a threshold segmentation method to obtain an initial water body area image; Calculating a normalized water index of the initial water region image; Using the normalized water body index to extract the water body region from the initial water body region image to obtain a final water body region; The final water body area is converted into vector data, and the area of ​​the surface elements in the vector data is calculated to obtain the water body area of ​​the target road area.

2. The road water body recognition method based on UAV multispectral images according to claim 1 is characterized in that: Calculating the normalized water index of the initial water area image specifically includes: Extracting spectral features of a set band image in the initial water body area image; A normalized water body index of the initial water body region image is calculated according to the spectral characteristics.

3. The road water body recognition method based on UAV multispectral images according to claim 1 is characterized in that: The threshold segmentation method is a segmentation method based on a fixed threshold, a segmentation method based on an automatic threshold, or a segmentation method based on a multi-scale object-oriented method.

4. The method for identifying road water bodies based on multispectral images of unmanned aerial vehicles according to claim 3 is characterized in that: The segmentation method based on automatic threshold is Otsu global threshold method.

5. The method for identifying road water bodies based on multispectral images of unmanned aerial vehicles according to claim 1 is characterized in that: The image includes four bands, namely green band, red band, near-infrared band and red edge band.

6. The method for identifying road water bodies based on multispectral images of unmanned aerial vehicles according to claim 1 is characterized in that: Process multiple UAV multispectral images to obtain images containing different bands, including: The band alignment, reflectivity calibration and synthesis of multiple UAV multispectral images are performed in sequence to obtain images containing different bands.

7. A road water body identification device based on multispectral images of unmanned aerial vehicles, characterized in that: The road water body recognition device based on the multispectral image of the unmanned aerial vehicle comprises: UAV multispectral image acquisition module, used to acquire multiple UAV multispectral images of the target road area; The image generation module containing different bands is used to process multiple UAV multispectral images to obtain image maps containing different bands; An initial water body region image determination module is used to segment the image into water bodies and non-water bodies using a threshold segmentation method to obtain an initial water body region image; A normalized water index calculation module, used to calculate the normalized water index of the initial water area image; A final water body region determination module, used to extract the water body region from the initial water body region image using the normalized water body index to obtain a final water body region; The water body area calculation module is used to convert the final water body area into vector data, calculate the area of ​​the surface elements in the vector data, and obtain the water body area of ​​the target road area.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the road area water body identification method based on unmanned aerial vehicle multispectral images as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying road water bodies based on multispectral images of unmanned aerial vehicles as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for identifying road water bodies based on multispectral images of unmanned aerial vehicles as described in any one of claims 1 to 6 is implemented.

Images