High-spectral remote sensing image cloud shadow identification method and system and storage medium

Abstract

The present application relates to high-resolution optical remote sensing image cloud shadow identification method, system and storage medium, including: obtaining high-resolution optical remote sensing data, calculating on-board reflectivity; according to the near-infrared band on-board reflectivity threshold condition, extracting low-value target area, calculating the on-board reflectivity data of near-infrared band iteratively to extract shadow area, calculating the difference image of the image after iteration and the initial image, and extracting the shadow area according to the difference image and the threshold condition; combining the cloud pixel position, the remote sensing image imaging geometric parameter and the cloud height range to calculate the set of potential cloud shadow area; respectively calculating the spatial overlap area of each element in the potential cloud shadow set and the shadow area, selecting the element with the largest spatial overlap area as the main shadow of the cloud, and extracting the final cloud shadow according to the main shadow and its buffer area and combining the near-infrared shadow area. The present application improves the degree of automation, reduces manual intervention, and has high accuracy of extraction result, and can quickly identify the cloud shadow in high-resolution remote sensing image.

Description

Technical Field

[0001] This invention relates to the technical field of surveying and mapping remote sensing, specifically to a method, system, and storage medium for cloud shadow recognition in high-resolution optical remote sensing images. Background Technology

[0002] Optical remote sensing imagery is heavily influenced by weather conditions. The presence of clouds and cloud shadows can obstruct or interfere with ground features, leading to inaccurate image recording. Statistics show that the global average probability of acquiring two cloud-free remote sensing images of the same area within any 80-day period is less than 30%. Therefore, detecting, identifying, and removing clouds and cloud shadows from optical remote sensing images is a crucial step in remote sensing data processing, effectively improving the utilization rate of optical remote sensing images. The spectral characteristics of clouds in optical remote sensing images are relatively distinct from other ground features, making them easier to identify, and mature detection algorithms already exist. However, cloud shadows present challenges. Firstly, most high-resolution remote sensing images lack synchronized thermal infrared remote sensing images, making it impossible to directly calculate cloud height and extract cloud shadows based on the geometric relationship between clouds and their shadows. Secondly, cloud shadows in high-resolution remote sensing images are highly similar to mountain shadows and tall building shadows, making direct extraction using spectral features difficult.

[0003] Existing methods for cloud shadow recognition in high-resolution optical remote sensing images mainly include manual interpretation, single-phase thresholding, and multi-phase detection.

[0004] Manual interpretation involves technicians visually interpreting cloud shadows directly from the spectral and texture information of high-resolution remote sensing images using specialized software. While this method offers high accuracy, it is time-consuming, labor-intensive, and inefficient, making it difficult to apply to processing massive amounts of remote sensing data.

[0005] The single-phase thresholding method involves using the spectral bands of the remote sensing image to be processed, and after calculating the exponential bands, selecting one or more features and setting corresponding thresholds to extract cloud shadows. This method has a relatively high degree of automation, but mountain shadows and tall building shadows have a significant impact on the results, so the accuracy of the extraction results cannot be guaranteed.

[0006] Multi-temporal detection involves comparing and analyzing cloudless images and images to be processed from the same region, and then using appropriate algorithms to extract cloud shadows. However, this method requires cloudless images to be prepared first, which is inefficient. In addition, if there are significant differences in the sunlight angle and observation angle between the two images, shadows from mountains and tall buildings can also affect the extraction results.

[0007] Therefore, it is necessary to develop a new method, system, and storage medium for cloud shadow recognition in high-resolution optical remote sensing images. Summary of the Invention

[0008] The purpose of this invention is to provide a method, system, and storage medium for cloud shadow recognition in high-resolution optical remote sensing images, which can improve the degree of automation, reduce manual intervention, and extract results with high accuracy, enabling rapid identification of cloud shadows in high-resolution remote sensing images.

[0009] In a first aspect, the cloud shadow recognition method for high-resolution optical remote sensing images according to the present invention includes the following steps:

[0010] Acquire high-resolution optical remote sensing data and calculate on-satellite reflectivity;

[0011] Low-value target areas are extracted based on the near-infrared band on-board reflectivity threshold conditions. The shadow area is extracted by iterative calculation of the near-infrared band on-board reflectivity data. The difference image between the iterated image and the initial image is calculated, and the shadow area is extracted based on the difference image and the threshold conditions.

[0012] The set of potential cloud shadow areas is calculated by combining cloud pixel locations, remote sensing image imaging geometric parameters, and cloud height range.

[0013] Calculate the spatial overlap area between each element in the potential cloud shadow area set and the shadow area, select the element with the largest spatial overlap area as the main cloud shadow, and extract the final cloud shadow based on the main shadow and its buffer, combined with the shadow area.

[0014] Optionally, before calculating the on-satellite reflectivity, preprocessing of the high-resolution optical remote sensing imagery is also included, specifically:

[0015] The raw pixel values ​​of high-resolution optical remote sensing data are digitally quantized values. These values ​​are then converted into radiance values ​​received by the remote sensing sensor by combining the deviation and gain coefficients in the corresponding metadata of the high-resolution optical remote sensing data.

[0016] Optionally, the on-board reflectivity is calculated, specifically as follows:

[0017]

[0018] in, d is the on-planet reflectivity, d is the Earth-Sun distance, and L is the radiance data. The solar constant, This represents the solar zenith angle during imaging.

[0019] Optionally, low-value target regions are extracted based on the on-board reflectivity threshold conditions in the near-infrared band, specifically:

[0020] Low-value pixels in the near-infrared band that satisfy the first condition are extracted and designated as the low-value target region T1. The first condition is:

[0021]

[0022] in: The first threshold; The on-board reflectance in the near-infrared band of the initial image; To obtain the maximum value.

[0023] Optionally, the shadow region can be extracted by iterative calculation of the near-infrared on-board reflectivity data, specifically:

[0024] Iterative calculations are performed on the near-infrared band data, specifically as follows:

[0025]

[0026] =

[0027] in: To represent a certain pixel, for The value calculated after the first iteration. To obtain the minimum value, For the eight neighboring regions of a pixel, for The initial near-infrared on-board reflectivity value, for The near-infrared on-board reflectivity value after the nth iteration. for (Near-infrared on-board reflectivity value after the (n+1)th iteration).

[0028] The condition for the iteration to end is:

[0029] - =0

[0030] in: This represents the on-board reflectivity value in the near-infrared band after the nth iteration.

[0031] The image after the iteration is completed is obtained.

[0032] Optionally, the difference image between the iterated image and the initial image is calculated, and the shadow area is extracted based on the difference image and threshold conditions, specifically:

[0033] Extract the pixels that meet the second condition as the shadow area. The second condition is:

[0034]

[0035] in, The image after the iteration is complete. This is the second threshold.

[0036] Optionally, the set of potential cloud shadow areas is calculated by combining cloud pixel locations, remote sensing image imaging geometry parameters, and cloud height range, specifically:

[0037] For any cloud pixel Calculate the shadow pixel corresponding to the cloud pixel. ,in, The x-coordinate of the cloud pixel in the image. The vertical coordinate of the cloud pixel in the image. The image x-coordinate of the shadow pixel. The vertical coordinate of the shadow pixel in the image;

[0038]

[0039]

[0040] in: The solar zenith angle at the time of imaging. The zenith angle observed during imaging. This is the solar azimuth angle at the time of imaging. The azimuth angle during imaging. Let h1 be the height of the cloud relative to the Earth's surface, and h1 ≤ h ≤ h2; for all clouds within this range... The calculations are performed using the formulas above, resulting in the set of potential cloud shadow areas corresponding to all heights, denoted as . .

[0041] Calculate the set of potential cloud shadow regions separately Each element in the shadow area Based on the spatial overlap area, select the element with the largest spatial overlap area. The main shadow of the cloud.

[0042] Optionally, the final cloud shadow is extracted based on the primary shadow and its buffer, combined with the shadow area, specifically:

[0043] along The edges expand outwards Each pixel, the expanded area is denoted as . Extract the final cloud shadows The The calculation formula is as follows:

[0044]

[0045] in, The main shadow of the cloud, These are cloud pixels extracted from high-resolution optical remote sensing images.

[0046] Secondly, the cloud shadow recognition system for high-resolution optical remote sensing images described in this invention employs a memory and a controller. The memory stores a computer-readable program, which, when invoked by the controller, can execute the steps of the cloud shadow recognition method for high-resolution optical remote sensing images as described in this invention.

[0047] Thirdly, the present invention provides a medium having a computer-readable program stored in its memory, which, when invoked by a controller, can execute the steps of the high-resolution optical remote sensing image cloud shadow recognition method as described in the present invention.

[0048] The beneficial effects of this invention are as follows: High-resolution remote sensing images are characterized by high spatial resolution but a small number of bands. Because the reflectance spectral characteristics of clouds differ significantly from those of other land features, they are relatively easy to identify using automated methods. However, the differences between mountain shadows, high-rise building shadows, and cloud shadows are smaller, resulting in higher accuracy for automated cloud shadow extraction. Furthermore, most high-resolution remote sensing data lack synchronized thermal infrared images, making it impossible to accurately invert cloud heights and directly calculate cloud shadow locations based on the geometric relationship between clouds and shadows. This invention first automatically iteratively calculates and extracts shadow areas from low-value regions in the near-infrared bands of high-resolution remote sensing images. Then, based on the estimated cloud height range and cloud pixels, it automatically calculates potential cloud shadow areas. Finally, it integrates shadow areas and potential cloud shadow areas to establish a cloud shadow identification method for high-resolution remote sensing images. This method requires no manual selection of training samples or setting of complex parameters, and it does not require cloudless reference images. It is highly automated and can effectively remove errors related to mountain shadows and high-rise building shadows in current automated identification methods. The extraction results are highly accurate and can quickly identify cloud shadows in high-resolution remote sensing images. Attached Figure Description

[0049] Figure 1 This is a flowchart of the cloud shadow recognition method for high-resolution optical remote sensing images described in the embodiments of this application;

[0050] Figure 2 This is an example diagram showing the results of cloud shadow identification in an embodiment of this application;

[0051] Figure 3 This is a schematic diagram of the cloud shadow recognition system for high-resolution optical remote sensing images in the embodiments of this application. Detailed Implementation

[0052] The following description, with reference to the accompanying drawings and preferred embodiments, illustrates the implementation of the technical solution of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0053] like Figure 1 As shown in the embodiments of this application, a method for cloud shadow recognition in high-resolution optical remote sensing images includes:

[0054] Acquire high-resolution optical remote sensing data and calculate on-satellite reflectivity;

[0055] Low-value target areas are extracted based on the near-infrared band on-board reflectivity threshold conditions. The shadow area is extracted by iterative calculation of the near-infrared band on-board reflectivity data. The difference image between the iterated image and the initial image is calculated, and the shadow area is extracted based on the difference image and the threshold conditions.

[0056] The set of potential cloud shadow areas is calculated by combining cloud pixel locations, remote sensing image imaging geometric parameters, and cloud height range;

[0057] Calculate the spatial overlap area between each element in the potential cloud shadow area set and the shadow area, select the element with the largest spatial overlap area as the main cloud shadow, and extract the final cloud shadow based on the main shadow and its buffer, combined with the shadow area.

[0058] The following combination Figure 1 The cloud shadow recognition method for high-resolution optical remote sensing images described in the embodiments of this application is explained in detail:

[0059] 1. High-resolution optical remote sensing image preprocessing

[0060] The original pixel values ​​of high-resolution optical remote sensing data are digitally quantized values. These values ​​are then converted into radiance values ​​L received by the remote sensing sensor by combining the deviation and gain coefficients in the corresponding metadata of the high-resolution optical remote sensing data.

[0061] Then, the on-board reflectivity is calculated using formula (1), as follows:

[0062] (1)

[0063] In the formula: d is the on-planetary reflectivity, d is the Earth-Sun distance, and L is the radiance data (unit: W / m²). 2 / sr / μm), The solar constant, This represents the solar zenith angle during imaging.

[0064] 2. Cloud pixel recognition

[0065] Cloud pixels in an image are identified and extracted using multi-threshold segmentation, supervised classification, or other extraction methods. The extracted cloud pixels are denoted as... .

[0066] 3. Extract shadow areas based on low-value pixels in the near-infrared band.

[0067] Low-value pixels in the near-infrared band that meet the following conditions are extracted and designated as the low-value target region T1.

[0068]

[0069] in: The first threshold is set to 0.2, which is generally taken as the reflectance characteristic of the near-infrared band of remote sensing images. This is the initial image.

[0070] Then, iterative calculations are performed on the near-infrared band data, specifically:

[0071] (2)

[0072] = (3)

[0073] In formula (2): To represent a certain pixel, for The value calculated after the first iteration, when hour, = ,when hour, = .

[0074] In formula (3): Indicates the maximum value. This represents the minimum value. Represents a pixel The eight neighboring areas, for The initial near-infrared on-board reflectivity value; for The near-infrared on-board reflectivity value after the nth iteration. for The on-board reflectivity value in the near-infrared band after the (n+1)th iteration;

[0075] The condition for the iteration to end is:

[0076] - =0 (4)

[0077] In equation (4): This represents the on-board reflectivity value in the near-infrared band after the nth iteration. , This means that all pixels in the entire image must satisfy the condition shown in formula (4); that is, the results of the two iterations are exactly the same. Finally, the pixels that satisfy the first condition are extracted as the shadow area. The first condition is:

[0078] (5)

[0079] in, The image after the iteration is complete. The second threshold is typically set to 0.02. The shadow areas extracted in this step include cloud shadows, mountain shadows, and tall building shadows.

[0080] 4. Estimating the set of potential cloud shadow areas based on cloud pixels and cloud height range

[0081] For any cloud pixel , in, The x-coordinate of the cloud pixel in the image. Given the image ordinate of the cloud pixel, calculate the corresponding shadow pixel using formulas (6) and (7). ,in The image x-coordinate of the shadow pixel, The vertical coordinate of the shadow pixel in the image.

[0082] (6)

[0083] (7)

[0084] In the formula: The solar zenith angle at the time of imaging. The zenith angle observed during imaging. This is the solar azimuth angle at the time of imaging. The azimuth angle during imaging. This represents the height of the cloud relative to the Earth's surface. (Setting) The range is h1-h2, where h1 and h2 can take values ​​of 500 meters and 6000 meters respectively. All values ​​within this range... The calculations are performed according to formulas (6) and (7), and the shadow areas corresponding to all cloud pixels can be obtained at any height. The set of all potential cloud shadow regions corresponding to all heights is denoted as . .

[0085] 5. Cloud shadow recognition

[0086] Calculate the set of potential cloud shadow areas All elements and shadow areas The spatial overlap area is used to determine the element with the largest spatial overlap area (let's assume it's 1). ( ) represents the main shadow of the cloud.

[0087] Also taking into account errors caused by terrain undulations, cloud edge errors, etc., along The edges expand outwards One pixel, The general value is 3, and the expansion area is denoted as... Finally, the final cloud shadow is extracted according to formula (8). :

[0088] (8)

[0089] in, The main shadow of the cloud, These are cloud pixels extracted from high-resolution optical remote sensing images.

[0090] like Figure 2 As shown, this is an example of the results of cloud shadow identification. In the figure, (a) is a remote sensing image, (b) is a cloud pixel, (c) is the shadow area extracted in the near-infrared band, (d) is a set of potential cloud shadow areas (partial), and (e) is the cloud shadow.

[0091] like Figure 3 As shown in the embodiments of this application, a cloud shadow recognition system for high-resolution optical remote sensing images employs a memory and a controller. The memory stores a computer-readable program, which, when invoked by the controller, can execute the steps of the cloud shadow recognition method for high-resolution optical remote sensing images as described in the embodiments of this application.

[0092] In this embodiment of the application, a storage medium has a computer-readable program stored in its memory. When the computer-readable program is invoked by a controller, it can execute the steps of the cloud shadow recognition method for high-resolution optical remote sensing images as described in this embodiment of the application.

[0093] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0094] In the context of this application, a storage medium may be a tangible storage medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. The storage medium may be a machine-readable signal medium or a machine-readable storage medium. Storage media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0095] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method for cloud shadow recognition in high-resolution optical remote sensing images, characterized in that, Includes the following steps: Acquire high-resolution optical remote sensing data and calculate satellite reflectivity; Low-value target areas are extracted based on the near-infrared band on-board reflectivity threshold conditions. Shadow areas are extracted by iterative calculation of near-infrared band on-board reflectivity data. The difference image between the iterated image and the initial image is calculated, and the shadow area is extracted based on the difference image and the threshold conditions. The set of potential cloud shadow areas is calculated by combining cloud pixel locations, remote sensing image imaging geometric parameters, and cloud height range. Calculate the spatial overlap area between each element in the potential cloud shadow area set and the shadow area, select the element with the largest spatial overlap area as the main shadow of the cloud, and extract the final cloud shadow based on the main shadow and its buffer zone, combined with the shadow area. Low-value target regions are extracted based on the on-board reflectivity threshold conditions in the near-infrared band, specifically: Low-value pixels in the near-infrared band that satisfy the first condition are extracted and designated as the low-value target region T1. The first condition is: in: The first threshold; The on-board reflectance in the near-infrared band of the initial image; To obtain the maximum value; The shadow region is extracted by iterative calculation of on-board reflectivity data in the near-infrared band, specifically as follows: Iterative calculations are performed on the near-infrared band data, specifically as follows: = in: To represent a certain pixel, for The value calculated after the first iteration. To obtain the minimum value, For the eight neighboring regions of a pixel, for The initial near-infrared on-board reflectivity value, for The near-infrared on-board reflectivity value after the nth iteration. for The on-board reflectivity value in the near-infrared band after the (n+1)th iteration; The condition for the iteration to end is: - =0 in: This represents the on-board reflectivity value in the near-infrared band after the nth iteration. Obtain the image after the iteration is complete; Calculate the difference image between the iterated image and the initial image, and extract the shadow area based on the difference image and threshold conditions, specifically: Extract the pixels that meet the second condition as the shadow area. The second condition is: in, The image after the iteration is complete. This is the second threshold.

2. The cloud shadow recognition method for high-resolution optical remote sensing images according to claim 1, characterized in that: Before calculating the on-satellite reflectivity, preprocessing of the high-resolution optical remote sensing imagery is also included, specifically: The original pixel values ​​of high-resolution optical remote sensing data are digitally quantized values. These values ​​are then converted into radiance values ​​received by the remote sensing sensor by combining the deviation and gain coefficients in the corresponding metadata of the high-resolution optical remote sensing data.

3. The cloud shadow recognition method for high-resolution optical remote sensing images according to claim 1, characterized in that: The on-board reflectivity is calculated as follows: in, d is the on-planet reflectivity, d is the Earth-Sun distance, and L is the radiance data. The solar constant, This represents the solar zenith angle during imaging.

4. The cloud shadow recognition method for high-resolution optical remote sensing images according to claim 1, characterized in that: The set of potential cloud shadow areas is calculated by combining cloud pixel locations, remote sensing image imaging geometry parameters, and cloud height range, specifically: For any cloud pixel Calculate the shadow pixel corresponding to the cloud pixel. ,in, The x-coordinate of the cloud pixel in the image. The vertical coordinate of the cloud pixel in the image. The image x-coordinate of the shadow pixel. The vertical coordinate of the shadow pixel in the image; in: The solar zenith angle at the time of imaging. The zenith angle observed during imaging. This is the solar azimuth angle at the time of imaging. The azimuth angle during imaging. Let h1 be the height of the cloud relative to the Earth's surface, and h1 ≤ h ≤ h2; for all clouds within this range... The calculations are performed using the formulas above, resulting in the set of potential cloud shadow areas corresponding to all heights, denoted as . ; Calculate the set of potential cloud shadow regions respectively Each element in the shadow area Based on the spatial overlap area, select the element with the largest spatial overlap area. The main shadow of the cloud.

5. The cloud shadow recognition method for high-resolution optical remote sensing images according to claim 4, characterized in that: Based on the primary shadow and its buffer, and combined with the shadow area, the final cloud shadow is extracted as follows: along The edges expand outwards Each pixel, the expanded area is denoted as . Extract the final cloud shadows The The calculation formula is as follows: in, The main shadow of the cloud, These are cloud pixels extracted from high-resolution optical remote sensing images.

6. A cloud shadow recognition system for high-resolution optical remote sensing images, characterized in that: The method employs a memory and a controller, wherein the memory stores a computer-readable program, which, when invoked by the controller, can perform the steps of the cloud shadow recognition method for high-resolution optical remote sensing images as described in any one of claims 1 to 5.

7. A medium, characterized in that: Its memory contains a computer-readable program, which, when invoked by the controller, can perform the steps of the cloud shadow recognition method for high-resolution optical remote sensing images as described in any one of claims 1 to 5.

Images