A method and device for automatically identifying ocean internal waves from satellite low-light cloud images

Abstract

The present invention discloses a method and device for automatically identifying ocean internal waves from satellite low-light cloud images, comprising the following steps: obtaining low-light data and converting it into a grayscale image; determining whether the low-light data meets a threshold range; calculating normalized radiation to obtain a lunar glare area; performing edge monitoring on the lunar glare area using a Canny algorithm and an envelope smoothing method to identify the boundary of the lunar glare area; determining the boundary of the transformation between bright and dark stripes in the lunar glare area by using different normalized radiation distributions caused by different sea surface roughness; geometrically correcting a local image containing ocean internal wave stripe characteristics within the boundary of the transformation between bright and dark stripes according to longitude and latitude information; and calculating the horizontal wavelength and propagation direction of the internal waves in the corrected image using a two-dimensional S-transformation model. The present invention can relatively systematically obtain samples of nighttime ocean internal wave low-light observation data, and obtain the horizontal wavelength and propagation direction of the ocean internal waves therefrom.

Description

Technical Field

[0001] The present invention relates to a method and device for automatically identifying ocean internal waves from satellite low-light cloud images, belonging to the field of remote sensing technology. Background Art

[0002] Changes in ocean roughness caused by internal waves can be observed in daytime visible light images, and the reasons for these phenomena are largely determined by the satellite observation geometry. Target areas on the sea surface that are rougher or smoother than the background may appear brighter or darker than the background in visible light images, depending on their position relative to the specular reflection point. Non-patent literature: Jackson CR, Alpers W. The role of the critical angle in brightness reversals on sunglint images of the sea surface [J]. Journal of Geophysical Research Oceans, 2010, 115 (C9) (Computing Normalized Radiation) first described the application of normalized radiation to identify bright and dark stripes in visible light images. Non-patent literature: Miller SD, Straka W, Mills SP, et al. Illuminating the capabilities of the Suomi National Polar-Orbiting Partnership (NPP) Visible Infrared Imaging Radiometer Suite (VIIRS) Day / Nightband [J]. Remote Sensing, 2013, 5(12): 6717-6766 (Elucidating the capabilities of the VIIRS / DNB channel of the Suomi-NPP satellite) first described that VIIRS / DNB can observe the characteristics of sea surface streaks caused by ocean internal waves at night. Non-patent literature: Hu S, Ma S, Yan W, et al. Measuring internal solitary wave parameters based on VIIRS / DNB data [J]. International Journal of Remote Sensing, 2019, 40(20): 7805-781 (Measuring internal solitary wave parameters based on VIIRS / DNB data) uses the low-light-level observation data of the VIIRS / DNB low-light-level imager with uniform spatial resolution to input the two-dimensional S-transform model to obtain the characteristic parameters of ocean internal waves.

[0003] Currently, the normalized radiometric algorithm is only applied to daytime visible light images and cannot handle the identification of flare areas under moonlight at night. Currently, most research on internal wave phenomena relies primarily on manual observations, lacking a systematic, automated algorithm for measuring ocean internal wave parameters. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method and device for automatically identifying ocean internal waves from satellite low-light cloud images, which can systematically obtain samples of ocean internal wave low-light observation data at night and derive the horizontal wavelength and propagation direction of the ocean internal waves from them.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a method for automatically identifying ocean internal waves from satellite low-light cloud images, comprising:

[0007] Obtain the low-light data of the ocean area collected in advance from the low-light channel of the satellite-borne low-light imager and convert it into a grayscale image with a grayscale value of 0-255;

[0008] Determining whether the twilight data meets a preset threshold range of the moon altitude angle and the moon phase angle; if so, retaining the data; otherwise, discarding the data;

[0009] The filtered low-light data are used to calculate the normalized radiation through the pre-acquired satellite observation geometry information and sea surface wind field information to obtain the lunar flare area formed by the lunar radiation.

[0010] Using the Canny algorithm and envelope smoothing method to monitor the edge of the lunar glare area and identify the boundary of the lunar glare area;

[0011] In the lunar glare region, the boundary of the transition between bright and dark stripes is determined by different normalized radiation distributions caused by different sea surface roughness;

[0012] The local image containing the ocean internal wave stripe characteristics within the boundary of the bright and dark stripes is geometrically corrected according to the longitude and latitude information;

[0013] The corrected image is used to calculate the horizontal wavelength and propagation direction of the internal waves using a two-dimensional S-transform model.

[0014] Furthermore, the preset threshold ranges of the moon altitude angle and the moon phase angle are: the moon altitude angle is less than 60° and the moon phase angle is less than 90°.

[0015] Furthermore, the filtered twilight data is used to calculate normalized radiation using pre-acquired satellite observation geometry information and sea surface wind field information to obtain the lunar glare area formed by lunar radiation, including:

[0016] The glare area is identified by the normalized radiance at the maximum wind speed on the sea surface. The calculation process is as follows:

[0017]

[0018] Among them, p(α,σ 2 ) is the symmetric slope probability density function, α is the sea level inclination, σ 2 is the sea surface roughness variance, and α is obtained by geometric calculation of satellite observations:

[0019]

[0020] σ 2 It is related to the sea surface wind speed w and is obtained through the empirical formula:

[0021] σ 2 =0.003+0.00512w (3)

[0022] The lunar radiation received by the satellite sensor in the glare area is obtained from the probability density function:

[0023]

[0024] Among them, F0 is the incident moon radiation in the flare area, and ρ(ω) is the Fresnel reflection coefficient. Finally, the normalized moon radiation F is obtained. n :

[0025]

[0026] Furthermore, the use of the Canny algorithm and envelope smoothing method to monitor the edge of the lunar glare area and identify the boundary of the lunar glare area includes:

[0027] The grayscale image is filtered using the Gaussian function, and the formula is as follows:

[0028]

[0029] The image gradient is calculated using the Sobel operator. Based on the weighted difference in grayscale between adjacent pixels, the gradient operator in the horizontal and vertical directions is calculated to obtain the extreme value at the edge.

[0030] By comparing the current pixel gradient intensity with the adjacent pixel gradients in the positive and negative gradient directions, if the current pixel gradient intensity is the local maximum, it is retained as an edge point, otherwise the gradient value is suppressed to 0;

[0031] Mark edge pixels greater than a high threshold as strong edge pixels, and pixels less than a high threshold and greater than a low threshold as weak edge pixels. If weak edge pixels are isolated, they are suppressed. Otherwise, if they are continuous, they are retained as edge points. If edge pixels are less than a low threshold, they are suppressed.

[0032] The boundary line is subjected to envelope smoothing to obtain an image for identifying the boundary of the lunar glare area.

[0033] Furthermore, in the lunar glare area, different normalized radiation distributions are caused by different sea surface roughness to determine the boundary of the transformation of bright and dark stripes, including: different normalized radiation distributions caused by different sea surface roughness caused by different sea surface wind speeds, comparing the normalized radiation distribution areas of the target area and the background area, extracting the normalized radiation boundary, and thus determining the boundary of the transformation of bright and dark stripes.

[0034] Furthermore, the local image containing the ocean internal wave stripe characteristics within the boundary of the conversion of bright and dark stripes is geometrically corrected according to the longitude and latitude information, and the corrected local image includes a horizontal axis and a vertical axis, wherein the horizontal axis represents the longitude value of the pixel point and the vertical axis represents the latitude value of the pixel point.

[0035] Furthermore, the calculation process of the two-dimensional S-transform model is as follows:

[0036]

[0037] Among them, h(x,y) is the input two-dimensional image, S(τ x ,τ y ,k x ,k y ) is the four-dimensional wave number spectrum, τ x and τ y are the translation variables in the x and y directions respectively. x and k y is the spatial angular wave number in the x and y directions, and c is the stretch parameter.

[0038] Furthermore, the calculation process of the horizontal wavelength and propagation direction is as follows:

[0039] λ(τ x , τ y )=(k x (τ x , τ y )) 2 +(k y (τ x , T y )) 2 (8)

[0040]

[0041] Among them, λ(τ x ,τ y ) is the horizontal wavelength, θ(τ x ,τ y ) is the propagation direction.

[0042] In a second aspect, the present invention provides an automatic ocean internal wave identification device based on satellite low-light cloud images, comprising a processor and a storage medium;

[0043] The storage medium is used to store instructions;

[0044] The processor is configured to operate according to the instructions to execute the steps of any of the aforementioned methods.

[0045] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of any of the aforementioned methods when executed by a processor.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] This invention transplants the glare region identification algorithm from the original daytime visible light image to low-light level image, achieving the first recognition of moonlit glare regions at night. It also proposes a new method for determining the transition boundary between bright and dark internal wave fringes using different normalized radiance distributions. A two-dimensional S-transform model is used to calculate the horizontal wavelength and propagation direction of internal wave data samples in geometrically corrected local low-light level images. All of the steps in this invention constitute a relatively systematic algorithm for automatic ocean internal wave identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 FIG2 is a flow chart of an embodiment of an algorithm for automatically identifying ocean internal waves in nighttime low-light-level images according to the present invention;

[0049] Figure 2 Shown is a grayscale image of an embodiment of the present invention based on an automatic ocean internal wave recognition algorithm in nighttime low-light level images;

[0050] Figure 3 Shown is an image of the glare zone boundary determination based on an embodiment of the automatic ocean internal wave recognition algorithm in nighttime low-light level images according to the present invention;

[0051] Figure 4 Shown is a normalized radiation distribution image based on an embodiment of an automatic ocean internal wave recognition algorithm in nighttime low-light level images;

[0052] Figure 5 The figure shows a boundary image converted from bright and dark stripes based on an embodiment of an automatic ocean internal wave recognition algorithm in nighttime low-light images according to the present invention;

[0053] Figure 6 Shown is a partial image after geometric correction according to an embodiment of the present invention based on an automatic ocean internal wave recognition algorithm in nighttime low-light level images;

[0054] Figure 7 Shown is a new partial image of an embodiment of the present invention based on an automatic ocean internal wave recognition algorithm in nighttime low-light level images;

[0055] Figure 8 Shown is a distribution diagram of the horizontal wavelength and propagation direction of ocean internal wave bright stripes according to an embodiment of the present invention based on an automatic ocean internal wave identification algorithm in nighttime low-light level images.

[0056] Figure 9 Shown is a distribution diagram of the horizontal wavelength and propagation direction of dark stripes of ocean internal waves based on an embodiment of the present invention's automatic identification algorithm for ocean internal waves in nighttime low-light level images. DETAILED DESCRIPTION

[0057] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0058] Example 1

[0059] This embodiment introduces a method for automatically identifying ocean internal waves from satellite low-light cloud images, including:

[0060] Obtain the low-light data of the ocean area collected in advance from the low-light channel of the satellite-borne low-light imager and convert it into a grayscale image with a grayscale value of 0-255;

[0061] Determining whether the twilight data meets a preset threshold range of the moon altitude angle and the moon phase angle; if so, retaining the data; otherwise, discarding the data;

[0062] The filtered low-light data are used to calculate the normalized radiation through the pre-acquired satellite observation geometry information and sea surface wind field information to obtain the lunar flare area formed by the lunar radiation.

[0063] Using the Canny algorithm and envelope smoothing method to monitor the edge of the lunar glare area and identify the boundary of the lunar glare area;

[0064] In the lunar glare region, the boundary of the transition between bright and dark stripes is determined by different normalized radiation distributions caused by different sea surface roughness;

[0065] The local image containing the ocean internal wave stripe characteristics within the boundary of the bright and dark stripes is geometrically corrected according to the longitude and latitude information;

[0066] The corrected image is used to calculate the horizontal wavelength and propagation direction of the internal waves using a two-dimensional S-transform model.

[0067] The application process of the method for automatically identifying ocean internal waves from satellite low-light cloud images provided in this embodiment specifically involves the following steps:

[0068] The radiance data of the ocean area is obtained from the low-light channel of the satellite-borne low-light imager and converted into a grayscale image with a grayscale value of 0-255;

[0069] Determine whether the low-light data meets the conditions of the moon altitude angle less than 60° and the moon phase angle less than 90° to ensure sufficient moonlight conditions for observing the nighttime internal wave phenomenon. If the moon phase angle conditions are met, the data will be retained; otherwise, it will be discarded.

[0070] The normalized radiation is calculated by using the pre-acquired satellite observation geometry information and sea surface wind field information to obtain the lunar flare area formed by the lunar radiation.

[0071] Use the Canny algorithm and envelope smoothing method to monitor the edge of the glare area and identify the boundary of the glare area;

[0072] The boundary between bright and dark stripes is determined by different normalized radiation distributions caused by different sea surface roughness;

[0073] The local image containing the ocean internal wave stripe characteristics within the boundary is geometrically corrected according to the latitude and longitude information;

[0074] The corrected image is used to calculate the horizontal wavelength and propagation direction of the internal waves using a two-dimensional S-transform model.

[0075] This paper transplants the glare region identification algorithm from existing daytime visible light images to low-light-level images, enabling the identification of moonlit glare regions at night for the first time. It also proposes a new method for determining the transition boundary between bright and dark internal wave fringes using different normalized radiance distributions. Using a two-dimensional S-transform model, internal wave data samples in geometrically corrected local low-light-level images are used to calculate their horizontal wavelength and propagation direction. All of the steps in this invention constitute a relatively systematic algorithm for automatic ocean internal wave identification.

[0076] Example 2

[0077] Based on Example 1, this example introduces in detail the steps and methods of the algorithm, including moon phase angle screening, normalized radiation calculation, glare area boundary identification, bright and dark stripe conversion boundary judgment, image geometric correction, and calculation of the horizontal wavelength and propagation direction of ocean internal waves.

[0078] (1) Determine the lunar phase angle and filter data

[0079] The low-light data needs to meet the conditions that the moon altitude angle is less than 60° and the moon phase angle is less than 90° to ensure sufficient moonlight conditions for observing internal wave phenomena at night.

[0080] (2) Identification of the lunar glare area

[0081] The glare area is identified by the normalized radiance at the maximum wind speed on the sea surface. The calculation process is as follows:

[0082]

[0083] p(α,σ 2 ) is the symmetric slope probability density function, α is the sea level inclination, σ 2 is the sea surface roughness variance, and α can be calculated through satellite observation geometry:

[0084]

[0085] σ 2 It is related to the sea surface wind speed w and can be obtained through the empirical formula:

[0086] σ 2 =0.003+0.00512w (3)

[0087] The lunar radiation received by the satellite sensor in the glare area can be obtained from the probability density function:

[0088]

[0089] Among them, F0 is the incident moon radiation in the flare area, and ρ(ω) is the Fresnel reflection coefficient. Finally, the normalized moon radiation F is obtained. n :

[0090]

[0091] The edge monitoring of the glare area is carried out using the Canny algorithm and envelope smoothing methods. It is mainly divided into five steps: Gaussian filtering image noise reduction, image gradient calculation by Sobel operator, non-maximum suppression, double threshold screening, and boundary envelope smoothing. Figure 3 .

[0092] (3) Propose a new method to determine the transition boundary between bright and dark stripes

[0093] The normalized radiation distribution obtained by different sea surface roughness caused by different sea surface wind speeds is different. By comparing the normalized radiation distribution areas of the target area and the background area, the normalized radiation boundary is extracted to determine the boundary of the bright and dark stripes transformation. Figure 4 、 5 .

[0094] (4) Image geometric correction

[0095] The local image containing the bright and dark stripes of the internal wave is geometrically corrected according to the longitude and latitude information. The corrected local image includes a horizontal axis and a vertical axis. The horizontal axis represents the longitude value of the pixel point, and the vertical axis represents the latitude value of the pixel point. Figure 6 .

[0096] (5) Calculation of the horizontal wavelength and propagation direction of ocean internal waves

[0097] The corresponding spatial resolution is calculated based on the corrected local image, and the radiance value of each pixel point corresponding to the spatial resolution and the image is input into the two-dimensional S-transform model. The horizontal wavelength and propagation direction parameter reference of the ocean internal wave are calculated by the two-dimensional S-transform model. Figure 8 、 Figure 9 .

[0098] The two-dimensional S-transform model includes formula (7), and the calculation process is as follows:

[0099]

[0100] Among them, h(x,y) is the input two-dimensional image, S(τ x ,τ y ,k x ,k y ) is the four-dimensional wave number spectrum, τ x and τ y are the translation variables in the x and y directions respectively. x and k y is the spatial angular wave number in the x and y directions, and c is the stretch parameter.

[0101] The horizontal wavelength and propagation direction of the ocean internal wave are obtained from equations (8) and (9), as follows:

[0102] λ(τ x , τ y )=(k x (τ x , τ y )) 2 +(k y (τ x , τ y )) 2 (8)

[0103]

[0104] Among them, λ(τ x ,τ y ) is the horizontal wavelength, θ(τ x ,τ y ) is the propagation direction.

[0105] In application, reference Figure 7 , the image texture features of the grayscale image, the local image of the extraction of the bright stripes of ocean waves in the lunar glare area in the grayscale image includes 164×177 pixels, and the longitude and latitude range of the new local image is (12.2-13.2°N, 95-96°E). According to the 742-meter spatial resolution of the original image of VIIRS / DNB and the method that the distance difference is 111km for every 1° difference in latitude, the spatial length occupied by each pixel point of the new local image is calculated to obtain the corresponding spatial resolution; similarly, the local image of the extraction of the dark stripes of ocean waves in the lunar glare area in the grayscale image includes 340×318 pixels, and the longitude and latitude range of the new local image is (8.9-9.7°N, 95.2-95.8°E). Reference Figure 8 The average horizontal wavelength of the ocean internal waves that show bright streaks in the figure is about 7.06 km, and the average propagation direction is 53°, that is, 53° east of south. Figure 9 The average horizontal wavelength of the ocean internal waves that show dark stripes in the figure is about 7.58 km, and the average propagation direction is -18°, that is, 18° east of north.

[0106] Example 3

[0107] This embodiment provides an automatic ocean internal wave identification device based on satellite low-light cloud images, including a processor and a storage medium;

[0108] The storage medium is used to store instructions;

[0109] The processor is configured to operate according to the instructions to execute the steps of the method according to any one of the first embodiments.

[0110] Example 4

[0111] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the steps of any one of the methods described in Embodiment 1 are implemented.

[0112] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for automatically identifying ocean internal waves from satellite low-light cloud images, characterized in that: include: Obtain the low-light data of the ocean area collected in advance from the low-light channel of the satellite-borne low-light imager and convert it into a grayscale image; Determining whether the low-light data meets a preset threshold range of the moon altitude angle and the moon phase angle; if so, retaining the data; otherwise, discarding the data; The filtered twilight data is used to calculate the normalized radiation using pre-acquired satellite observation geometry information and sea surface wind field information to obtain the lunar flare area formed by lunar radiation; including: The glare area is identified by the normalized radiance at the maximum wind speed on the sea surface. The calculation process is as follows: （1）； in, is a symmetric slope probability density function, is the sea level inclination, is the sea surface roughness variance, The geometric calculation is obtained through satellite observation: (2)； It is related to the sea surface wind speed w and is obtained through the empirical formula: (3)； The lunar radiation received by the satellite sensor in the glare area is obtained from the probability density function: (4)； Among them, F0 is the incident lunar radiation in the flare region, is the Fresnel reflection coefficient; finally, the normalized moon radiation F is obtained n : (5)； Using the Canny algorithm and envelope smoothing method to monitor the edge of the lunar glare area and identify the boundary of the lunar glare area; In the lunar glare region, the boundary of the transition between bright and dark stripes is determined by different normalized radiation distributions caused by different sea surface roughness; The local image containing the ocean internal wave stripe characteristics within the boundary of the bright and dark stripes is geometrically corrected according to the longitude and latitude information; The horizontal wavelength and propagation direction of the internal wave in the corrected image are calculated using a two-dimensional S-transform model. The calculation process of the two-dimensional S-transform model is as follows: (7)； in, is the input two-dimensional image, is the four-dimensional wave number spectrum, and are the transformation variables in the x and y directions respectively; k x and k y is the spatial angular wave number in the x and y directions, c is the stretch parameter; The calculation process of the horizontal wavelength and propagation direction is as follows: (8)； (9)； in, is the horizontal wavelength, For the direction of transmission.

2. The method for automatically identifying ocean internal waves from satellite low-light cloud images according to claim 1, characterized in that: The preset threshold ranges of the moon altitude angle and the moon phase angle are: the moon altitude angle is less than 60° and the moon phase angle is less than 90°.

3. The method for automatically identifying ocean internal waves from satellite low-light cloud images according to claim 1, characterized in that: The method of using the Canny algorithm and the envelope smoothing method to monitor the edge of the lunar glare area and identify the boundary of the lunar glare area includes: The grayscale image is filtered using the Gaussian function, and the formula is as follows: (6)； The image gradient is calculated using the Sobel operator. Based on the weighted difference in grayscale between adjacent pixels, the gradient operator in the horizontal and vertical directions is calculated to obtain the extreme value at the edge. By comparing the current pixel gradient intensity with the adjacent pixel gradients in the positive and negative gradient directions, if the current pixel gradient intensity is the local maximum, it is retained as an edge point, otherwise the gradient value is suppressed to 0; Mark edge pixels greater than the high threshold as strong edge pixels, and pixels less than the high threshold and greater than the low threshold as weak edge pixels. If the weak edge pixels are isolated, they are suppressed. Otherwise, if they are continuous, they will be retained as edge points. If the edge pixels are less than the low threshold, they will be suppressed. The boundary line is subjected to envelope smoothing to obtain an image for identifying the boundary of the lunar glare area.

4. The method for automatically identifying ocean internal waves from satellite low-light cloud images according to claim 1, characterized in that: In the lunar glare area, different normalized radiation distributions are caused by different sea surface roughness to determine the boundary of the transformation of bright and dark stripes, including: different normalized radiation distributions caused by different sea surface roughness caused by different sea surface wind speeds, comparing the normalized radiation distribution areas of the target area and the background area, extracting the normalized radiation boundary, and thus determining the boundary of the transformation of bright and dark stripes.

5. The method for automatically identifying ocean internal waves from satellite low-light cloud images according to claim 1, characterized in that: The local image containing the ocean internal wave stripe characteristics within the boundary of the conversion of bright and dark stripes is geometrically corrected according to the longitude and latitude information. The corrected local image includes a horizontal axis and a vertical axis, wherein the horizontal axis represents the longitude value of the pixel point and the vertical axis represents the latitude value of the pixel point.

6. An automatic ocean internal wave recognition device based on satellite low-light cloud images, characterized by: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images