An underwater positioning method and system based on polarization characteristics and sky image restoration

Abstract

The present invention discloses an underwater positioning method and system based on polarization characteristics and underwater image restoration. In the process of underwater positioning using polarization characteristics, the underwater polarization pattern is used as a navigation reference compass for underwater positioning. First, underwater sky images in multiple polarization directions are acquired, and the images are preprocessed for denoising by using Wiener filtering combined with the BRDF model. Then, the underwater transmittance, the underwater background light intensity at infinity, and the backscattered light in the underwater physical imaging model are estimated by using the dark channel prior principle and polarization characteristics to achieve underwater image restoration. Next, a polarization perception model is designed to calculate the underwater polarization pattern. Finally, the position of the solar meridian is extracted according to the underwater polarization pattern, and the underwater azimuth information is calculated in combination with the solar azimuth angle. The present invention can ensure the accuracy of polarization information by denoising and restoring polarization images in a complex underwater environment, enabling the underwater vehicle to stably and reliably perform passive navigation and positioning tasks.

Description

Technical Field

[0001] The present invention belongs to the field of underwater positioning and navigation, and relates to an underwater positioning method and system based on polarization characteristics and sky image restoration. Specifically, after image denoising, the image is restored based on the dark channel prior and polarization characteristics, and the underwater polarization pattern is obtained by using a polarization perception model, and the azimuth information is obtained by using the position information of the sun meridian in it. Background Art

[0002] The navigation system is a key technology for autonomous underwater vehicles to complete underwater detection tasks independently. It plays an important role in the research of marine resource exploration, target recognition, positioning and communication, etc. Due to its special mission requirements, it needs to submerge and standby underwater for a long time, and as the navigation range continues to increase, its navigation system needs to have the characteristics of high accuracy, good stability, good persistence and low risk. Currently, underwater navigation systems include inertial navigation systems, satellite navigation systems, radio navigation systems, astronomical navigation systems, geomagnetic navigation systems, etc. Among them, the inertial navigation has a high update rate, short-term accuracy and good stability. The disadvantages are that the positioning error accumulates over time and the initial alignment time is relatively long before each use. The satellite navigation works by radio waves, is vulnerable to attack during wartime, and its dynamic characteristics are not as good as those of the inertial navigation system. The astronomical navigation is not affected by the electromagnetic field, does not radiate electromagnetic waves outward, has good concealment, high orientation and positioning accuracy, but is limited by clouds and meteorological conditions and is difficult to be used in real scenarios. In recent years, polarization navigation technology has gradually become a research hotspot. It only depends on the polarization light information in the environment and has the advantages of being not easily interfered by the outside world, small volume, high integration, low cost, and the error does not accumulate over time, providing a new idea for underwater navigation and positioning.

[0003] Since the polarized light source in the natural underwater environment mainly comes from skylight, the polarization characteristics of underwater light are closely related to the polarization characteristics of skylight and contain position information that can be used for navigation and positioning. Therefore, it is feasible to apply bionic polarization navigation sensors underwater by studying the distribution pattern of underwater polarized light. Summary of the Invention

[0004] Object of the Invention: Aiming at the problems existing in the prior art, the object of the present invention is to provide an underwater positioning method and system based on polarization characteristics and sky image restoration, which can effectively complete the positioning application of autonomous underwater vehicles only relying on the polarization light information in the environment, and has the advantages of being not easily interfered by the outside world, good concealment, low calculation cost, and the error does not accumulate over time. Combining with other navigation methods can solve many existing problems.

[0005] Technical Solution: To achieve the above object of the invention, the present invention adopts the following technical solutions:

[0006] An underwater positioning method based on polarization characteristics and sky image restoration mainly includes the following steps:

[0007] (1) Obtain sky images in multiple polarization directions underwater, and use Wiener filtering combined with the BRDF model to denoise the sky polarization images to improve the quality of underwater polarization images;

[0008] (2) Based on the underwater physical imaging model, use the dark channel prior principle and polarization characteristics to estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model to achieve underwater polarization image restoration;

[0009] (3) For the restored polarization images, obtain polarization information based on the polarization opponency perception model, calculate the underwater polarization pattern, and the polarization opponency perception model uses a weight factor to regulate the information entropy of the underwater polarization pattern to obtain the optimal underwater polarization pattern;

[0010] (4) Extract the position of the solar meridian in the underwater polarization pattern, and combine it with the solar azimuth angle to solve the azimuth information of the underwater vehicle.

[0011] Preferably, in step (1), using Wiener filtering combined with the BRDF model to denoise the sky polarization images includes: using Wiener filtering to process the blurring and noise of the images, describing the image blurring process through the point spread (PSF) function and analyzing the texture features and frequency domain properties through the power spectral density (PSD) function to remove image noise and restore the detailed information of the original image; using the bidirectional reflectance distribution function (BRDF) model to estimate the light propagation process in the underwater environment, calculate the scattering of light in the underwater environment, and then use the calculated scattering information to correct the underwater images in the Wiener filtering process to eliminate the image blurring noise caused by underwater scattering.

[0012] Preferably, in step (2), using the dark channel prior principle and polarization characteristics to estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model to achieve underwater polarization image restoration includes the following steps:

[0013] (2.1) Estimate the transmittance and underwater background light intensity at infinity in the underwater imaging physical model according to the dark channel prior principle. Perform dark channel processing on the image with a 3x3 window, find the darkest pixel in the local area of the current pixel to obtain the dark channel image; select the positions of the top 1% brightest pixel points in the dark channel image, and take the light intensity of the original image corresponding to their maximum window average pixel value as the underwater background light intensity at infinity; use the estimated underwater background light intensity at infinity and the total image light intensity, and according to the dark channel principle, estimate the minimum transmittance in the window where the current pixel is located, and solve the global transmittance of the image.

[0014] (2.2) Estimate the backscattered light of each pixel point based on polarization information. The light intensity information has a great influence on the degree of polarization. In order to suppress the influence of transmitted light on polarization information, the polarization angle is selected as the parameter for estimation. The polarization angle with the highest occurrence probability is selected in the polarization angle map, and the degree of polarization corresponding to the pixel positions of this polarization angle is screened out. The maximum value is selected as the degree of polarization of the backscattered light, and then the backscattered light of the image is solved.

[0015] (2.3) According to the estimated underwater transmittance, the underwater background light intensity at infinity, and the backscattered light, the restored polarization image is obtained by combining with the underwater physical imaging model.

[0016] Preferably, the formula for restoring the underwater polarization image is:

[0017]

[0018] where (x, y) represents the image pixel coordinates, L(x, y) represents the restored image, I(x, y) represents the total light intensity of the image to be restored, t(x, y) represents the underwater transmittance, A ∞ represents the underwater background light at infinity, I s (x, y) represents the scattered light component of the image, and its formula is I s (x, y) = A ∞ (1 - t(x, y)).

[0019] Preferably, in step (3), calculating the underwater polarization pattern based on the polarization opposition perception model includes the following steps:

[0020] (3.1) Design a polarization perception model, and collect sky images (I 0° , I 45° , I 90° , I 135° ) in four polarization directions (0°, 45°, 90°, 135°) respectively. The polarized light between mutually perpendicular angles forms a group of polarization channels, and weight factors are added before each polarization channel for the enhancement and suppression of the polarization channels to solve the polarization information of the underwater sky image, that is, the underwater polarization pattern;

[0021] (3.2) Solve the optimal weight factor in an adaptive optimization manner to solve the optimal information entropy of the underwater polarization pattern map. First, set the initial weight factor and the information entropy of the underwater polarization pattern map, then calculate the information entropy according to the current weight factor, and then update the weight factor. If the image information entropy in the current state is greater than the image information entropy in the previous state, update the weight factor, otherwise the weight factor increases at a fixed step length until the weight factor corresponding to the optimal information entropy is selected to obtain the underwater polarization pattern with the optimal information entropy.

[0022] Preferably, in step (4), the position of the solar meridian is extracted according to the underwater polarization pattern, and underwater positioning calculation is performed in combination with the solar azimuth angle, including the following steps:

[0023] (4.1) Extract the solar meridian according to the underwater polarization pattern to obtain the angle between the solar meridian and the body axis;

[0024] (4.2) Calculate the solar azimuth angle through the solar calendar;

[0025] (4.3) Establish a relationship model between the underwater space coordinate system and the solar meridian, and calculate the azimuth information of the underwater vehicle according to the extracted solar meridian position and solar azimuth angle information.

[0026] Based on the same inventive concept, the present invention provides an underwater positioning system based on polarization characteristics and sky image restoration, including: a polarization image detection and processing module, configured to acquire sky images in multiple polarization directions underwater, perform denoising processing on the sky polarization images using Wiener filtering combined with the BRDF model, estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model based on the underwater physical imaging model, and adopt the dark channel prior principle and polarization characteristics to achieve underwater polarization image restoration; a polarization information processing module, configured to obtain polarization information based on the polarization opponency perception model for the restored polarization images, calculate the underwater polarization pattern, and the polarization opponency perception model uses a weight factor to regulate the information entropy of the underwater polarization pattern to obtain the optimal underwater polarization pattern; and a positioning module, configured to extract the solar meridian in the underwater polarization pattern, and combine the solar azimuth angle to calculate the azimuth information of the underwater vehicle.

[0027] Based on the same inventive concept, the present invention provides a computer system, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the computer program is loaded into the processor, the steps of the underwater positioning method based on polarization characteristics and sky image restoration are implemented.

[0028] Beneficial effects: The underwater positioning method based on polarization characteristics provided by the present invention mainly realizes underwater navigation and positioning through the underwater polarization pattern. First, an imaging polarization sensor is used to detect the sky polarization light, and the image is denoised by using Wiener filtering combined with the BRDF model to enhance the image details. Then, based on the underwater physical imaging model, the underwater transmittance, the underwater background light at infinity, and the backscattered light parameters in the model are estimated by using the dark channel prior principle and polarization characteristics to obtain the restored result of the image after removing scattering. Next, a polarization perception model is proposed to calculate the underwater polarization pattern, and the underwater polarization pattern with the optimal information entropy is optimized according to the weight factor. Finally, the spatial position of the solar meridian is extracted according to the underwater polarization pattern, and the heading information is determined in combination with the solar azimuth angle. Compared with the existing navigation methods of autonomous underwater vehicles using satellites, inertia, geomagnetism, and terrain matching, the present invention is inspired by the ability of insects such as mantis shrimp to distinguish directions using polarization light in the environment, accurately predicts the distribution of sky polarization characteristics under natural water surfaces, and locates the carrier according to the underwater polarization pattern. The present invention optimizes the underwater polarization pattern to make the pattern obtain the highest information entropy, so as to more accurately obtain the solar meridian, and can comprehensively improve the accuracy of orientation using polarization light in cloudy weather and complex underwater environments. Description of the Drawings

[0029] Figure 1 is the flowchart of the method of the embodiment of the present invention.

[0030] Figure 2 is the schematic diagram of the underwater space coordinate system.

[0031] Figure 3 is the schematic diagram of azimuth calculation principle.

[0032] Figure 4 is the experimental device diagram.

[0033] Figure 5 is the schematic diagram of solar meridian extraction. Among them, the (a) row is the polarization angle distribution pattern detected at different times and calculated, the (b) row is the polarization degree distribution pattern detected at different times and calculated, the (c) row is the spatial position feature point map extracted at different times through thresholding, and the (d) row is the solar meridian map after extraction at different times.

[0034] Figure 6 is the error comparison diagram. Detailed Embodiment

[0035] In order to clearly highlight the purpose and advantages of the present invention, the present invention will be further described below in conjunction with the drawings in the embodiments of the present invention.

[0036] As Figure 1As shown in the figure, in an underwater positioning method based on polarization characteristics and sky image restoration disclosed in an embodiment of the present invention, the implementation process mainly includes the following steps:

[0037] Step 1: Denoising and restoration processing of underwater sky images. First, obtain sky images in multiple polarization directions underwater, use Wiener filtering combined with the BRDF model to perform denoising processing on the sky polarization images, then, with the underwater physical imaging model as the background, adopt the dark channel prior principle and polarization characteristics to estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the model, and finally achieve the restoration of underwater polarization images.

[0038] Build an underwater polarization light acquisition system to obtain the original sky images in four polarization directions underwater, and improve the image quality through denoising and restoration algorithms for the original polarization images. In this embodiment, Wiener filtering combined with the BRDF model is used to perform denoising processing on the sky polarization images, including: using Wiener filtering to process the blur and noise of the image, describing the image blur process through the point spread (PSF) function and analyzing the texture features and frequency domain properties through the power spectral density (PSD) function to remove image noise and restore the detailed information of the original image; using the bidirectional reflectance distribution function (BRDF) model to estimate the light propagation process in the underwater environment, calculate the scattering situation of light in the underwater environment, and then use the calculated scattering information to correct the underwater image in the Wiener filtering process to eliminate the image blur noise caused by underwater scattering. The specific image denoising process is as follows:

[0039] Collect sky images in different polarization directions, namely I 0° ,I 45° ,I 90° and I 135° During the image acquisition process, due to factors such as shooting or the environment, the images degenerate. Denoise each polarization image according to the Wiener filtering method. The degenerate image model is:

[0040] g(x, y) = h(x, y) * f(x, y) + n(x, y)

[0041] In the formula, g(x, y) is the noisy degenerate image, f(x, y) is the restored image, n(x, y) is the additive noise model, and h(x, y) represents the system function (degradation function) that combines the degradation factors. If both the degradation function and noise are considered, the general inverse filtering effect will become very poor due to the existence of noise. Wiener filtering can be performed based on the estimation of the frequency response of the image to reduce the influence of noise and improve the image quality. Its expression in the frequency domain is:

[0042]

[0043] Among them, G(u, v) is the degraded image, F(u, v) is the restored image, and |H(u, v)| 2 represents the power spectral density (PSD) of the image degradation function H(u, v), which refers to the energy distribution in the frequency domain. |H(u, v)| 2 = H * (u, v)H(u, v), where H * (u, v) is the complex conjugate of H(u, v). H(u, v) is the image degradation function and can be calculated from the point spread function (PSF). The PSF describes the distribution of an ideal point light source on the imaging plane after passing through the optical system. Its spatial domain expression is:

[0044]

[0045] where a is a parameter measuring the scale of the PSF, x and y are spatial coordinates, and h(x, y) is transformed into the frequency domain as H(u, v).

[0046] is the ratio of the power spectral density of the scattering noise to the power spectral density of the non-degraded image. Its formula is:

[0047]

[0048] where N(u, v) is the scattering noise, F(u, v) is the non-degraded image, and N(u, v) / F(u, v) is the noise-to-signal ratio. Since the value of F(u, v) is to be solved, the true noise-to-signal ratio can be obtained by measuring the scattering coefficient. The bidirectional reflectance distribution function (BRDF) model can describe the underwater scattering situation. Its model is as follows:

[0049]

[0050] where d is the differential symbol, ω i and ω r respectively represent the solid angles of the incident direction and the observation direction. There is θ and represent the incident angle and azimuth angle of a beam of light. L r (ω r ) represents the radiation intensity in the observation direction ω r . L i (ω i ) represents the radiation intensity in the incident direction ω i . f r (ω i , ω r ) represents the ratio of the light intensities in the observation direction ω r and the incident direction ω i . Solving gives the scattering coefficient 1 - f r (ω i, ω r ) as the signal-to-noise ratio.

[0051] The image restoration method based on the underwater physical imaging model starts from image denoising and degradation, and obtains the restoration result through inversion using the imaging model by estimating parameters. Specifically, the transmittance and the underwater background light intensity at infinity in the underwater imaging physical model are estimated according to the dark channel prior principle, the backscattered light of each pixel point is estimated based on polarization information, and the restored polarization image is obtained by combining the estimated underwater transmittance, the underwater background light intensity at infinity, and the backscattered light with the underwater physical imaging model.

[0052] In this embodiment, according to the sky images I in four polarization directions after denoising 0° , I 45° , I 90° , I 135° , restoration processing is performed on them. Usually, the total light intensity of an image is composed of transmitted light and backscattered light, and its expression is:

[0053] I(x, y) = I t (x, y) + I s (x, y)

[0054] where (x, y) represents the image pixel coordinates, I(x, y) is the total light intensity of the image, I s (x, y) is the backscattered light, and I t (x, y) is the underwater target transmitted light, and its expression is:

[0055] I t (x, y) = L(x, y)t(x, y)

[0056] t(x, y) = e -βz

[0057] In the formula, L(x, y) is the image to be restored, t(x, y) is the underwater transmittance, z represents the distance between the optical imaging system and the target, and β is the light attenuation coefficient, which represents the attenuation function of the absorption and scattering of light by the water body.

[0058] I s (x, y) is the backscattered light, indicating that part of the light enters the camera after passing through the scattering particles, resulting in a decrease in imaging quality, and its expression is:

[0059] I s (x, y) = A ∞ (1 - t(x, y))

[0060] In the formula, A ∞ represents the underwater background light intensity value at infinity, and t(x, y) is the underwater transmittance. The total expression of the underwater physical imaging model is:

[0061] I(x, y) = L(x, y)t(x, y) + A ∞ (1 - t(x, y))

[0062] That is:

[0063]

[0064] It can be seen from the above formula that by solving t(x, y) and A ∞ parameters, the original image L(x, y) can be restored. According to I t (x, y) and the common t(x, y) in I s (x, y), the following formula can be obtained:

[0065]

[0066] The underwater restored image L(x, y) can be obtained from the above formula:

[0067]

[0068] It can be seen from the above formula that by obtaining A ∞ and I s (x, y), the restored image L(x, y) can be obtained.

[0069] Specifically, according to the dark channel prior principle, the underwater background light intensity A ∞ at infinity and the transmittance t(x, y) are estimated. The expression of the dark channel is:

[0070]

[0071] where J C (x, y) is each channel of the color image, and Ω(x, y) represents the window centered on the pixel (x, y). A window of size 3*3 is selected to find the darkest pixel in the local area of the current pixel, and the dark channel image J dark (x, y) is obtained. Then, among the pixels with the top 1% highest brightness in the dark channel, the maximum value of the average light intensity of the 3*3 window of the selected pixel is used as the underwater background light intensity A ∞ at infinity, that is

[0072]

[0073] In the formula is the pixel value of the top 1% highest pixels in the dark channel of the image, represents finding the maximum average pixel value of the 3*3 window centered on (x, y), represents the light intensity value corresponding to the window where the selected highest pixel is located, that is, the underwater background light intensity A ∞ .

[0074] According to the underwater background light intensity A at infinity ∞ the transmittance t(x, y) can be further solved. According to the general expression of the underwater physical imaging model, it is transformed into:

[0075]

[0076] Regarding t(x, y) in the formula as a constant and taking the minimum value of both sides of the above formula twice, we get:

[0077]

[0078] According to the dark channel prior principle, there is at least one pixel in the sky region of the non-scattering image whose value is approximately 0, that is, J dark (x, y) → 0, we can get Then the transmittance is:

[0079]

[0080] Considering the actual situation, the particles in water will inevitably affect the imaging system. To obtain a more natural restored image, a correction factor ω with a value of 0.95 is introduced, and the correction formula is as follows:

[0081]

[0082] where ω is the correction factor, A ∞ is the underwater background light intensity at infinity, and I(x, y) is the total light intensity of the image.

[0083] Determine the backscattered light I s (x, y). The light intensity information has a great influence on the degree of polarization. In order to suppress the influence of the transmitted light on the polarization information, the polarization angle is selected as the parameter estimation. The polarization angle with the highest occurrence probability is selected in the polarization angle map, and the polarization degree corresponding to the pixel position of this polarization angle is screened out. The maximum value is selected as the polarization degree of the backscattered light, and then the backscattered light of the image is solved. The specific steps are as follows:

[0084] ① Calculate the polarization angle image and the polarization degree image through the Stokes vector parameters;

[0085] According to the sky image I 0° (x, y), I 45° (x, y), I 90° (x, y), I 135° (x, y), calculate the corresponding Stokes vector parameters, that is:

[0086]

[0087] Among them, I represents the total light intensity of polarized light, Q represents the light intensity difference between the horizontal polarization component and the vertical polarization component, U represents the light intensity difference between the polarization component in the 45° direction and the polarization component in the 135° direction, and V represents the light intensity difference between the right-handed polarization component and the left-handed polarization component of light. The degree of polarization DOP and the distribution map of the angle of polarization AOP are calculated from the Stokes vector parameters.

[0088]

[0089]

[0090] ② Conduct distribution statistics on the angle-of-polarization map of the sky region, and select the angle of polarization Ψ with the highest occurrence probability. s ;

[0091] ③ Screen out the degrees of polarization corresponding to the pixel positions that satisfy the angle of polarization Ψ s , and select the maximum polarization value as the degree of polarization P of the backscattered light. s .

[0092] According to the polarized light imaging theory, an image can be decomposed into a pair of mutually orthogonal polarized images. The light intensity value of the image passing through the polarized light is I / 2, then the transmitted light intensity is I(1 - p) / 2. The polarized image is composed of the transmitted light and the polarized part of the light intensity. According to Malus' law, the light intensity I SP (x, y) of the polarized part of the polarized image collected in the 0° direction is:

[0093]

[0094] In the formula, I(0°) represents the light intensity in the 0° polarization direction, p is the degree of polarization, and I is the total light intensity. Ψ s is the angle of polarization with the highest occurrence probability. According to the definition of the degree of polarization, that is, the ratio of the light intensity of the polarized part to the total light intensity, the scattered light intensity I s of each pixel can be obtained as:

[0095]

[0096] According to the above-obtained A ∞ and t(x, y) or A ∞ and I s (x, y), the restored image based on the underwater physical imaging model is:

[0097]

[0098] Thus, the denoising and restoration processing of the underwater image are completed.

[0099] Step 2: For the restored polarization image, design a polarization opponency perception model to obtain polarization information and calculate the underwater polarization pattern. Calculating the underwater polarization pattern according to the polarization opponency perception model specifically includes:

[0100] (1) Based on the processed polarization image I 0° , I 45° , I 90° , I 135° , calculate the polarization parameters of each channel, and its expression is:

[0101] S 1 = k 1 I 90° - k 2 I 0° - c 1 I 45°

[0102] S 2 = k 3 I 0° - k 4 I 90° - c 2 I 135°

[0103] S 3 = k 5 I 45° - k 6 I 135° - c 3 I 90°

[0104] S 4 = k 7 I 135° - k 8 I 45° - C 4 I 0°

[0105] Among them, the polarization parameter S represents the polarization light intensity difference after enhancement or suppression between angles, k represents the enhancement or suppression factor in each polarization channel, and c represents the suppression factor of adjacent groups of polarization light intensities.

[0106] (2) Weight factor optimization. Taking the information entropy of the polarization parameter as a reference index, its expression is:

[0107]

[0108] Where L represents the gray value of the image, p(i) is the ratio of the number of pixels with gray value i to the total number of pixels. The larger the information entropy E, the more information the image contains. When E reaches the maximum value, the weight factor is the optimal solution. The present invention uses an adaptive optimization method to solve the weight factor. The weight factors in the opponent perception model are mainly divided into: the enhancement factor k of the interlayer photoreceptor cells i (i = 1, 3, 5, 7), the inhibition factor k i (i = 2, 4, 6, 8), and the inhibition factor C of the photoreceptor cells in adjacent groups i (i = 1, 2, 3, 4). The weight factor is optimized by the exhaustive search method, and the main steps are as follows:

[0109] (2.1) Initialize the weight factor. Set k 1 = 1 and the value range is 1 to 10, k 1 = 0.1 and the value range is 0 to 1, c 1 = 0.001 and the value range is 0.001 to 0.05. The initial information entropy of the polarization-sensitive parameter image is N 0 representing the initial state;

[0110] (2.2) Calculate the information entropy of the polarization-sensitive image. Calculate the current information entropy according to the current weight factor;

[0111] (2.3) Update the weight factor. If the information entropy of the current state is higher than that of the previous state, that is N i and N i-1 represent the current state and the previous state, then update the current weight factor. Otherwise, the weight factor increases at a fixed step until the maximum information entropy value of the polarization-sensitive image is found, and the corresponding optimal weight factor value is recorded.

[0112] (3) Calculate the underwater polarization pattern according to the polarization perception model, that is, the degree of polarization DOP distribution map and the angle of polarization AOP distribution map. The formula is:

[0113]

[0114]

[0115] Thus, the acquisition of the underwater polarization pattern is completed.

[0116] Step 3: Finally, extract the solar meridian according to the underwater polarization pattern, and combine it with the solar azimuth angle to calculate the angle between the body axis direction and the geographical due north direction First, extract the solar meridian from the underwater polarization pattern to obtain the angle between the solar meridian and the body axis. Then, calculate the solar azimuth through the solar calendar. Finally, establish a relationship model between the underwater space coordinate system and the solar meridian, and calculate the underwater vehicle azimuth information based on the extracted solar meridian position and solar azimuth information.

[0117] Specifically, the method for extracting the position of the solar meridian in the underwater polarization pattern is as follows: initially set a threshold T α , and the distribution characteristics of the characteristic points of the solar meridian position satisfy 90° - T α <AOP<90° and -90°<AOP<T α - 90°, and the characteristic points of the solar meridian position are segmented from the background area through threshold segmentation. The position of the solar meridian is obtained by using the linear fitting method, that is, the angle between the solar meridian and the body axis

[0118] Figure 2 Figure for the underwater polarization positioning space coordinate system, which describes the process of light entering the vehicle observation position from the atmosphere. To more intuitively describe the underwater azimuth information calculation process, it is transformed into a two-dimensional geographical plane coordinate system, as Figure 3 shown.

[0119] represents the solar azimuth, which is the angle between the sun and the due north direction. According to astronomical knowledge, the solar altitude angle θ at the current moment can be obtained from the geographical latitude β, the solar declination angle δ, and the time hour angle t s and the solar azimuth Its calculation formula is:

[0120] θ s =arcsin(sinδsinβ + cosδcosβcost)

[0121]

[0122] In the formula, β represents the geographical latitude of the observation point, t is the solar hour angle, which is calculated from the true solar time, and δ is the solar declination angle.

[0123] is the angle between the solar meridian and the vehicle. By extracting the characteristic points of the solar meridian position in the underwater polarization pattern and using the linear fitting method, the slope k of the solar meridian can be obtained, thereby calculating the angle between the solar meridian and the vehicle That is:

[0124]

[0125] Establish the spatial relationship between the underwater space coordinate system and the solar meridian. Taking the underwater observation position as the origin, establish a northeast-up coordinate system, and determine the solar meridian as the reference benchmark for underwater positioning. According to the calculated solar azimuth The included angle between the extracted solar meridian and the body axis Thus, calculate the included angle between the carrier direction and the true north of the geography Azimuth The calculation formula is:

[0126]

[0127] So far, the calculation of the carrier azimuth angle is completed, and the application of underwater positioning using polarization characteristics and sky image restoration is completed.

[0128] Based on the same inventive concept, an underwater positioning system disclosed in an embodiment of the present invention based on polarization characteristics and sky image restoration includes: a polarization image detection and processing module, configured to obtain sky images in multiple polarization directions underwater, and use Wiener filtering combined with the BRDF model to perform denoising processing on the sky polarization images. Taking the underwater physical imaging model as the background, adopt the dark channel prior principle and polarization characteristics to estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model, and realize underwater polarization image restoration; a polarization information processing module, configured to, for the restored polarization images, obtain polarization information based on the polarization opponency perception model, calculate the underwater polarization pattern, and the polarization opponency perception model uses a weight factor to regulate the information entropy of the underwater polarization pattern to obtain the optimal underwater polarization pattern; and a positioning module, configured to extract the solar meridian in the underwater polarization pattern, and combine the solar azimuth angle to calculate the azimuth information of the underwater vehicle.

[0129] Based on the same inventive concept, a computer system disclosed in an embodiment of the present invention includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the computer program is loaded into the processor, the steps of the underwater positioning method based on polarization characteristics and sky image restoration are implemented.

[0130] In order to verify the effects of the underwater positioning method and system using polarization characteristics and sky image restoration proposed by the present invention, an experimental device was built for experiments. During the experiment, a camera was used to photograph the zenith, and the camera axis was consistent with the body axis direction. The experimental device is as Figure 4 shown.

[0131] In order to verify the effectiveness of the present invention, the underwater polarization positioning performance at different times was experimentally tested. The experimental results use the heading angle error as the evaluation criterion. The experimental results are as Figure 5As shown in the figure, (a) is the sky polarization angle distribution diagram; (b) is the sky polarization degree distribution diagram; (c) is the position feature point fusion diagram; (d) is the solar meridian extraction result.

[0132] The experiment was carried out from 8:00 am to 5:00 pm, keeping the actual azimuth in the due south direction, and the test was carried out every hour. The experiment recorded the time of each hour, the solar altitude angle and solar azimuth angle corresponding to the longitude and latitude of the test location, the information entropy of the underwater polarization pattern, the azimuth angle, and the azimuth error. As shown in the following table, E 1 and E 2 are the Stokes vector parameters and the information entropy of the underwater polarization pattern obtained by the present invention respectively, and θ 1 and θ 2 are the Stokes vector parameters and the azimuth information calculated under the underwater polarization pattern obtained by the present invention respectively, and δ 1 and δ 2 are the Stokes vector parameters and the azimuth error under the underwater polarization pattern obtained by the present invention respectively.

[0133] Table 1 Experimental data table

[0134]

[0135]

[0136] From Table 1 and Figure 6 the data result analysis shows that the sky is easily affected by clouds and water mist, etc., resulting in the fuzziness of the polarization azimuth angle information. In the sky polarization pattern obtained by testing under these weather conditions, the solar meridian area will be more blurred. The traditional method is greatly affected by noise, while the method and system proposed by the present invention have better robustness and stability, which is of great significance for underwater polarization positioning.

Claims

1. An underwater positioning method based on polarization characteristics and sky image restoration, characterized in that, it includes the following steps: (1) Obtain sky images in multiple polarization directions underwater, and use Wiener filtering combined with the BRDF model to denoise the sky polarization images; (2) Based on the underwater physical imaging model, adopt the dark channel prior principle and polarization characteristics to estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model, and realize the restoration of underwater polarization images; (3) For the restored polarization images, obtain polarization information based on the polarization opponency perception model, calculate the underwater polarization pattern, and the polarization opponency perception model uses a weight factor to regulate the information entropy of the underwater polarization pattern to obtain the optimal underwater polarization pattern; (4) Extract the position of the solar meridian in the underwater polarization pattern, and combine it with the solar azimuth angle to solve the azimuth information of the underwater vehicle.

2. The underwater positioning method based on polarization characteristics and sky image restoration according to claim 1, characterized in that, in the step (1), using Wiener filtering combined with the BRDF model to denoise the sky polarization images includes: using Wiener filtering to process the blur and noise of the image, describing the image blur process through the point spread (PSF) function and analyzing the texture features and frequency domain properties through the power spectral density (PSD) function to remove image noise and restore the detail information of the original image; using the bidirectional reflectance distribution function (BRDF) model to estimate the light propagation process in the underwater environment, calculate the scattering of light in the underwater environment, and then use the calculated scattering information to correct the underwater image in the Wiener filtering process to eliminate the image blur noise caused by underwater scattering.

3. The underwater positioning method based on polarization characteristics and sky image restoration according to claim 1, characterized in that, in the step (2), adopting the dark channel prior principle and polarization characteristics to estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model to realize the restoration of underwater polarization images includes the following steps: (2.1) Estimate the transmittance and underwater background light intensity at infinity in the underwater imaging physical model according to the dark channel prior principle, perform dark channel processing on the image with a 3x3 window, find the darkest pixel in the local area of the current pixel to obtain the dark channel image; select the brightest set proportion of pixel positions in the dark channel image, and take the maximum value of the window average pixel values of them corresponding to the light intensity of the original image as the underwater background light intensity at infinity; use the estimated underwater background light intensity at infinity and the total image light intensity to estimate the minimum transmittance in the window where the pixel is located according to the dark channel principle to obtain the global transmittance of the image; (2.2) Estimate the backscattered light of each pixel point based on polarization information, select the polarization angle with the highest occurrence probability in the polarization angle map, screen out the polarization degree corresponding to the pixel positions of this polarization angle, select the maximum value as the backscattered light polarization degree, and then solve the backscattered light of the image; (2.3) Based on the estimated underwater transmittance, underwater background light intensity at infinity, and backscattered light, the restored polarization image is obtained by combining with the underwater physical imaging model.

4. A method for underwater positioning based on polarization characteristics and sky image restoration according to claim 1, wherein, the formula for underwater polarization image restoration is: Or Among them, (x, y) represents the image pixel coordinates, L(x, y) represents the restored image, I(x, y) represents the total light intensity of the image to be restored, t(x, y) represents the underwater transmittance, and A ∞ represents the underwater background light intensity at infinity, and I s (x, y) represents the scattered light component of the image. There is I s (x, y) = A ∞ (1 - t(x, y)).

5. A method for underwater positioning based on polarization characteristics and sky image restoration according to claim 1, wherein, in step (3), calculating the underwater polarization pattern based on the polarization opposition perception model includes the following steps: (3.1) Design a polarization perception model to collect sky images I in four polarization directions of 0°, 45°, 90°, and 135° respectively 0° 、I 45° 、I 90° 、I 135° . The polarized light between mutually perpendicular angles forms a set of polarization channels. A weight factor is added before each polarization channel for the enhancement and suppression of the polarization channels to solve the polarization information of the underwater sky image, that is, the underwater polarization pattern; (3.2) Solve the optimal weight factor in an adaptive optimization manner to solve the optimal information entropy of the underwater polarization pattern map. First, set the initial weight factor and the information entropy of the underwater polarization pattern map, then calculate the information entropy according to the current weight factor, and then update the weight factor. If the image information entropy in the current state is greater than that in the previous state, update the weight factor; otherwise, the weight factor increases by a fixed step until the weight factor corresponding to the optimal information entropy is selected to obtain the underwater polarization pattern with the optimal information entropy.

6. A method for underwater positioning based on polarization characteristics and sky image restoration according to claim 1, wherein, in step (4), extracting the position of the solar meridian according to the underwater polarization pattern and performing underwater positioning calculation in combination with the solar azimuth angle includes the following steps: (4.1) Extract the solar meridian from the underwater polarization pattern to obtain the angle between the solar meridian and the body axis; (4.2) Calculate the solar azimuth angle through the solar calendar; (4.3) Establish a relationship model between the underwater space coordinate system and the solar meridian, and calculate the azimuth information of the underwater vehicle according to the extracted solar meridian position and solar azimuth angle information.

7. An underwater positioning system based on polarization characteristics and sky image restoration, wherein, comprising: a polarization image detection and processing module, configured to obtain sky images in multiple polarization directions underwater, perform denoising processing on the sky polarization images using Wiener filtering combined with the BRDF model, estimate the transmittance, underwater background light intensity at infinity, and backscattered light in the underwater imaging model based on the underwater physical imaging model, and realize underwater polarization image restoration; a polarization information processing module, configured to obtain polarization information based on the polarization opposition perception model for the restored polarization image, calculate the underwater polarization pattern, and the polarization opposition perception model uses a weight factor to regulate the information entropy of the underwater polarization pattern to obtain the optimal underwater polarization pattern; a positioning module, configured to extract the solar meridian in the underwater polarization pattern and calculate the azimuth information of the underwater vehicle in combination with the solar azimuth angle.

8. An underwater positioning system based on polarization characteristics and sky image restoration according to claim 7, wherein, The method for estimating the transmittance, the underwater background light intensity at infinity, and the backscattered light by the polarization image detection and processing module is as follows: According to the dark channel prior principle, the transmittance and the underwater background light intensity at infinity in the underwater imaging physical model are estimated. The dark channel processing is performed on the image with a 3x3 window to find the darkest pixel in the local area of the current pixel, and the dark channel image is obtained; in the dark channel image, the pixel positions of the brightest set proportion of pixels are selected, and the light intensity of the original image corresponding to the maximum value of their window average pixel values is taken as the underwater background light intensity at infinity; using the estimated underwater background light intensity at infinity and the total image light intensity, the minimum transmittance in the window where the pixel is located is estimated according to the dark channel principle to obtain the global transmittance of the image; the backscattered light of each pixel point is estimated based on the polarization information. The polarization angle with the highest occurrence probability is selected in the polarization angle map, the polarization degrees corresponding to the pixel positions of this polarization angle are screened out, and the maximum value is selected as the backscattered light polarization degree, and then the backscattered light of the image is solved.

9. An underwater positioning system based on polarization characteristics and sky image restoration according to claim 7, characterized in that, the formula for underwater polarization image restoration is: or Among them, (x, y) represents the image pixel coordinates, L(x, y) represents the restored image, I(x, y) represents the total light intensity of the image to be restored, t(x, y) represents the underwater transmittance, and A ∞ represents the underwater background light intensity at infinity, and I s (x, y) represents the scattered light component of the image, and there is I s (x, y) = A ∞ (1 - t(x, y)).

10. A computer system, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, when the computer program is loaded into the processor, the steps of an underwater positioning method based on polarization characteristics and sky image restoration according to any one of claims 1-6 are implemented.

Images