A bionic polarized light heading measurement method based on an improved polarized field singularity point model

Abstract

The present invention discloses a bionic polarized light heading measurement method based on an improved polarization field singular point model. First, an image-based bionic polarized light sensor and a high-precision fiber-optic inertial navigation system are initialized. Next, the bionic polarized light sensor captures an image and transmits it to a host computer to determine the observation vector, polarization degree, and polarization angle information. Then, an improved polarization field singular point model for the full-sky polarization pattern is constructed using singular point correction coefficients. Based on a vector consisting of the solar azimuth, solar altitude, atmospheric turbidity parameters, and correction coefficients in the polarized light sensor system, the polarization angle and polarization degree corresponding to a single pixel are calculated. The polarization angle information is converted from the sensor coordinate system to the observation meridian coordinate system. Finally, a minimization objective function is constructed, and the least squares optimization method is used to determine the solar azimuth in the polarized light sensor system. Combined with the solar azimuth in the world coordinate system, the heading is ultimately determined. This invention addresses the problems of existing image-based polarized light sensor heading measurement methods, which are all based on a single Rayleigh scattering model and cannot adapt to weather conditions with thick aerosols and high solar altitudes. Furthermore, existing sky polarization pattern characterization models suffer from high computational complexity and poor characterization capabilities.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bionic polarized light navigation, and particularly to a bionic polarized light heading measurement method based on an improved polarized field singular point model. BACKGROUND

[0002] Absolute heading measurement under global navigation satellite system (GNSS) failure or magnetic interference conditions is a challenging problem. Inertial measurement units cannot meet the long-time and fully autonomous navigation requirements due to the cumulative error. Many biological studies have shown that mantis shrimp, octopus, dragonfly larvae, and sand ant can perceive the polarization of light using their unique visual structure and use the polarization information for navigation, foraging, and migration. Some archaeological studies have shown that ancient Vikings can also use the regular polarization distribution in the sky to determine the heading of a ship and achieve long-distance navigation.

[0003] Inspired by this, scholars have conducted a large number of bionic polarized light absolute heading measurement related research. Polarized light sensors are mainly divided into point source type and image type. However, the point source type polarized light sensor can only measure the polarization information in one direction at a time, and is easily affected by environmental interference, and has poor robustness. Therefore, the image type polarized light sensor has attracted more and more attention, and how to use the collected polarization image to obtain more robust and more accurate navigation information has gradually become a research hotspot.

[0004] The single scattering theory assumes that the maximum polarization direction of the polarization light at any point in the sky is always perpendicular to the plane composed of the sun vector and the observation vector. There are two main ways to measure the heading using bionic polarized light sensors. On the one hand, by solving the different representations of the sun vector in the carrier coordinate system and the navigation coordinate system, the heading can be solved by using the difference of the sun azimuth angle in different coordinate systems. On the other hand, the projection of the sun meridian in the sensor detection plane can be directly solved by using the line features and symmetry of the polarization image, and the sun azimuth angle in the sensor system can also be obtained. The above heading measurement methods are based on the full-sky polarization representation model, which is the Katrian model. In the case of high atmospheric turbidity, the existence of the neutral point will significantly affect the robustness of the heading measurement when the sun elevation angle changes. The polarized field singular point model considers the neutral point caused by the depolarization effect using the polarized field singular point analytical function, which reduces the computational complexity, but the AOP image features of the model have a large deviation from the actual situation.

[0005] The existing image type polarized light sensor heading measurement method is based on single Rayleigh scattering model, which cannot adapt to the weather condition of thick aerosol and high solar elevation angle, and the existing sky polarization pattern representation model has problems of large calculation amount and poor representation ability. Therefore, the heading measurement method based on improved polarization field singular point model is proposed, the singular point correction coefficient is introduced into the traditional sky polarization pattern polarization field singular point representation model to improve the representation ability of the model, and the minimum objective function about AOP value is constructed, and the least square method is used to solve the heading to improve the heading measurement precision. SUMMARY

[0006] The application aims to provide a bionic polarized light heading measurement method based on improved polarization field singular point model, which can improve the heading measurement robustness under different solar elevation angle conditions.

[0007] The technical scheme for achieving the application is as follows: a bionic polarized light heading measurement method based on improved polarization field singular point model, comprising the following steps:

[0008] Step one: install the image type bionic polarized light sensor and high precision fiber optic inertial navigation system, complete the start, preheating preparation, initial alignment, and use the output heading of the high precision fiber optic inertial navigation system as the reference value to evaluate the heading measurement result of the bionic polarized light sensor;

[0009] Step two: the bionic polarized light sensor collects images and transmits them to the host computer to solve the observation vector and polarization degree and polarization angle information; the polarization state of light can be described by Stokes vector as

[0010]

[0011] In the formula, s0 represents the total intensity of light, s1 represents the linearly polarized light component in the horizontal direction, s0 represents the linearly polarized light component in the 45° direction, Ip k , k = 1, 2, 3, 4 represent the light intensity output of four polarization channels, then the polarization angle (AOP) ψ i of the polarization unit is 0.5arctan (s2, s1), and the polarization degree (DOP) is

[0012] Step three: construct the improved polarization field singular point model w i (x) of the full sky polarization pattern, wherein i is the index of the current observation vector, x = [alpha, beta, A, k1, k2] T represents the vector composed of the solar azimuth angle alpha, the solar elevation angle beta, the atmospheric turbidity parameter A and the correction coefficient k1, k2 at the current time under the polarization light sensor system;

[0013] Step four: De-warping operation is performed on single pixel of fisheye polarization camera. Thus the correspondence between observation vector and pixel point (u, v) is obtained. Each pixel (u, v) of polarization sensor corresponds to an observation vector OP i and the corresponding polarization vector Then the observation vector OP i corresponds to the polarization angle and degree of polarization

[0014] Step five: In order to obtain the polarization angle information in the observation meridian plane coordinate system, certain conversion is needed. The azimuth angle of the observation vector is subtracted from the measured AOP to obtain the polarization angle φ i (x) = ψ i (x) - α OPi , ψ i represents the AOP of the sensor system, and α OPi represents the azimuth angle of the current observation vector. The AOP measurement value of the sensor also needs to be processed φ mi = ψ mi - α OPi , ψ mi is the measured value in the sensor system, and φ mi is the measured value in the meridian plane coordinate system.

[0015] Step six: The theoretical AOP value φ i (x) and the measured AOP value φ mi under the current parameters are used to construct a minimization objective function. The Powell least squares optimization algorithm is used to solve the objective function, thereby obtaining the estimated polarization sensor system's solar azimuth angle α. According to the local geographical location and time, the solar azimuth angle α s in the world coordinate system can be obtained. The heading angle at the current time can be obtained by subtracting the two. The solution of the heading angle has a 180° ambiguity, which can be determined by the value given by other navigation systems.

[0016] In step three, the improved polarization field singular point model of the full-sky polarization pattern is constructed, and the specific method is as follows:

[0017] In order to quantitatively describe the sky polarization information, the complex number form of the projection of a certain observation vector OP (x, y, z) on the celestial sphere in the horizontal plane is represented by the Cartesian coordinate system as ζ = x + iy. The complex number representation of the projection point in the polar coordinate system is ζ = rexp(iδ). When the solar meridian plane coincides with the yoz plane, ζ = 0, which represents the projection point of the zenith vector, and the height angle is β. Correspondingly, the radius of the projection circle on the horizontal plane is r = (1-tan(β / 2)) / (1+tan(β / 2)). And the anti-symmetric point of ζ is set as -1 / ζ *The polarization information can be represented by the complex form of the non-normalized Stokes parameters:

[0018] w(ζ) = <E x +iE y ) 2 > = <E x 2 - <E y 2 + 2i <E x E y > = |w(ζ)|exp(2iφ(ζ))

[0019] In the above equation, |w(ζ)| represents the degree of polarization information, and φ(ζ) contains the polarization angle information relative to the x-axis direction. w(ζ) is a function related to the observation vector, and its zero point can be represented as the neutral point of the sky polarization pattern. A constant A related to the atmospheric turbidity is set to consider the influence of multiple scattering particles in the atmosphere. If the projection line of the solar meridian is always coincident with the y-axis, when the solar elevation angle is β s , the projection point of the solar position is ζ s = iy s = i(1-tan(β s / 2)) / (1+tan(β s / 2)), the positions of the four neutral points are:

[0020]

[0021] The solar elevation angle under the polarization sensor system is now simply written as β, and the projection point of the solar vector is y m = (1-tan(β / 2)) / (1+tan(β / 2)). The neutral point coordinates are written in the vector form of the Cartesian coordinate system:

[0022]

[0023] The azimuth angle of the solar vector under the polarization light sensor coordinate system at the current time is defined as α, and the matrix for two-dimensional rotation on the ground plane is

[0024]

[0025] Then the neutral point vector is rotated in the ground plane coordinate system, and the neutral point coordinates at the current time and their complex representation form are obtained as

[0026]

[0027] [ζ1 ζ2 ζ3 ζ4] = ζ x +iζ y

[0028] Then the projection point ζ0=x0+iy0 of the horizontal plane under the meridional plane coincidence condition is also subjected to the rotation transformation to obtain the projection point coordinate ζ of the polarized light sensor system

[0029] [x y] T = R Berry [x0 y0] T , ζ=x+iy

[0030] If the initial observation projection point ζ0 is known, the sky polarization pattern constraint condition based on the polarization field singular point model about the solar azimuth angle α, the elevation angle β and the atmospheric turbidity parameter A can be obtained as:

[0031] w(α,β,A)∝(ζ-ζ1)(ζ-ζ2)(ζ-ζ3)(ζ-ζ4)

[0032] In order to make the ∞ shape saturation of the polarization field singular point model closer to the true value, two coefficients k1 and k2 are introduced in the present application to correct the neutral point position on both sides of the anti-solar position, and the initial neutral point coordinate can be expressed as:

[0033]

[0034] Thus, the real-time representation function of the sky polarization pattern based on the improved polarization field singular point model containing the polarization information of the current observation direction is obtained:

[0035]

[0036] Compared with the prior art, the present application has the beneficial effects that:

[0037] (1) The singular point correction coefficient is introduced into the sky polarization pattern polarization field singular point representation model, and the representation ability of the polarization pattern is improved;

[0038] (2) The influence of the neutral point is considered in the process of solving the heading of the polarization light compass, and the heading measurement accuracy is improved;

[0039] (3) The method is simple and practical, and has high heading measurement robustness for polarization images of different field of view ranges. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 is a flowchart of the present application;

[0041] Figure 2 is a schematic diagram of different representation models of the sky polarization pattern;

[0042] Figure 3 is a schematic diagram of the imaging process of the image type polarization light sensor;

[0043] Figure 4 is a schematic diagram of the experimental equipment and process.

[0044] Figure 5 Full sky polarization pattern simulation and measurement;

[0045] Figure 6 Comparison of the robustness of heading measurements of different models to the field of view range. DETAILED DESCRIPTION

[0046] The following is combined with Figure 1 The present invention is further described in detail with reference to the flow diagram of the present invention.

[0047] The sky polarization pattern is a special distribution pattern in the sky formed by polarized light generated by sunlight scattered by particles. It has a significant distribution pattern. Under clear weather conditions, the scattering particles are mainly composed of atmospheric molecules, whose size is much smaller than the wavelength of light. Therefore, the first-order Rayleigh scattering model can be used to describe the atmospheric scattering process under clear weather conditions. That is, the direction of the E vector (the electric vibration vector in the light wave) of the scattered light is perpendicular to the scattering surface. Figure 2 As shown, O represents the position of the observer, S represents the direction of the sun on the celestial sphere, and the zenith angle γ is used. s and azimuth angle α s Indicates; P represents the observation direction, whose zenith angle and azimuth angle are γ and α respectively; φ is the polarization angle of the incident light, and θ is the scattering angle. According to the first-order Rayleigh scattering model, the polarization degree of the scattered light is

[0048]

[0049] The polarization angle φ of the incident light is

[0050]

[0051] This allows the sky polarization pattern (including the degree of polarization d and polarization angle φ) to be solved based on the first-order Rayleigh scattering model.

[0052] However, the actual atmospheric polarization pattern does not strictly satisfy the first-order Rayleigh scattering model. There will be obvious "neutral points" in the polarization distribution diagram, which will seriously affect the robustness of the heading angle solution. This phenomenon is caused by the multi-order scattering of aerosol particles of different sizes and shapes, anisotropic scattering, and the depolarization effect caused by ground reflection. The four neutral points are located on the main plane perpendicular to the sun and the zenith, and their distribution is as follows: Figure 2 As shown in Figure 2, the Babinet and Brewster neutral points are located on either side of the sun, while the Arago and fourth neutral points are located on either side of the anti-sun position. Under clear weather conditions, only the Babinet and Arago neutral points, or the Babinet and Brewster neutral points, can be observed at the same time.

[0053] The neutral points of the atmospheric polarization pattern are modeled accurately by the polarized field singularity theory. To quantitatively describe the sky polarization information, the complex number form of the projection of a certain observation vector OP(x, y, z) on the celestial sphere to the horizontal plane is expressed as ζ = x + iy in the Cartesian coordinate system. The complex number form of the projection point in the polar coordinate system is ζ = rexp(iδ). When the solar meridian plane coincides with the yoz plane, ζ = 0, which means the projection point of the zenith vector, and the height angle is β, and the radius of the projection circle on the horizontal plane is r = (1 - tan(β / 2)) / (1 + tan(β / 2)). And the anti-symmetric point of ζ is set as -1 / ζ * . The polarization information can be expressed as the complex number form of the non-normalized Stokes parameters:

[0054] w(ζ) = <E x +iE y ) 2 > = <E x 2 >- <E y 2 >+ 2i <E x E y > = |w(ζ)|exp(2iφ(ζ))

[0055] In the above formula, |w(ζ)| represents the degree of polarization information, and φ(ζ) contains the polarization angle information relative to the x-axis direction. The function zero point of w(ζ) can represent the neutral point of the sky polarization pattern. To consider the influence of multiple scattering particles in the atmosphere, a constant A related to the atmospheric turbidity is set. If the projection line of the solar meridian line always coincides with the y-axis, when the solar elevation angle is β s , the projection point of the solar position is ζ s = iy s = i(1 - tan(β s / 2)) / (1 + tan(β s / 2)), the positions of the four neutral points are:

[0056]

[0057] Therefore, the whole sky polarization pattern needs to satisfy the zero point constraint relationship:

[0058] w(ζ)∝(ζ-ζ + )(ζ-ζ - )(ζ+1 / ζ + * )(ζ+1 / ζ - * )

[0059] In the above formula, ζ + and ζ -The neutral point on the left side of the sun vector is -1 / ζ + * The neutral point on the right side of the sun vector is -1 / ζ - * The neutral point on the left side of the sun vector is -1 / ζ * In order to make the degree of polarization |w(ζ)| also satisfy the invariance |w(ζ)| = |w(1 / ζ m )|, the above formula is further modified as

[0060]

[0061] The polarization field singular point model takes into account the sky polarization pattern modeling under the turbid atmosphere. On the one hand, this model is established under the condition that the solar meridian plane coincides with the yoz plane, which is difficult to apply to the acquisition of navigation information. On the other hand, from the simulation results of the polarization field singular point model, it can be seen that although the polarization field singular point model takes into account the neutral point information generated by the depolarization effect, the shape saturation degree is still different from Rayleigh, which will have a certain impact on the heading angle measurement. Therefore, this paper proposes a heading measurement method based on an improved polarization field singular point model. Figure 2

[0062] Set the atmospheric turbidity constant as A, and the solar elevation angle under the polarization sensor system as β, then the sun vector projection point is y m = (1-tan(β / 2)) / (1+tan(β / 2)), and the neutral point coordinates in (4) are written in the vector form of the Cartesian coordinate system:

[0063]

[0064] Define the azimuth angle of the sun vector under the polarization light sensor coordinate system at the current time as α, and the matrix of two-dimensional rotation on the ground plane as

[0065]

[0066] Then rotate the neutral point vector in the ground plane coordinate system to obtain the neutral point coordinates at the current time and its complex representation form as

[0067]

[0068] [ζ1 ζ2 ζ3 ζ4] = ζ x +iζ y

[0069] Then the projection point ζ0=x0+iy0 on the ground plane under the meridian plane coincidence condition is also rotated and transformed to obtain the projection point coordinates ζ

[0070] [xy] T =R Berry [x0 y0] T ,ζ=x+iy

[0071] If the initial observation projection point ζ0 is known, the sky polarization pattern constraints based on the polarization field singular point model with respect to the solar azimuth angle α, altitude angle β, and atmospheric turbidity parameter A can be obtained as follows:

[0072] w(α,β,A)∝(ζ-ζ1)(ζ-ζ2)(ζ-ζ3)(ζ-ζ4)

[0073] In order to make the ∞ shape saturation of the polarization field singular point model closer to the true value, this paper introduces two coefficients k1 and k2 to correct the neutral point position on both sides of the anti-solar position. The initial neutral point coordinates can be expressed as:

[0074]

[0075] Thus, a real-time characterization function of the sky polarization pattern based on the improved polarization field singular point model containing the polarization information of the current observation direction is obtained:

[0076]

[0077] After determining the characterization function of the sky polarization pattern, the navigation information can be solved. As shown in the figure below, an image-based polarization sensor is used to image the sky polarization pattern, and the fisheye camera is calibrated for internal parameters and distortion. In this way, the corresponding relationship between the observation vector and the pixel point (u, v) is obtained. At this time, each pixel (u, v) of the polarization sensor will correspond to an observation vector OP i and the corresponding polarization vector like Figure 3 As shown. Assume that an observation vector OP i The corresponding polarization angle and degree of polarization are:

[0078]

[0079] Where x = [α, β, A, k1, k2] T , represents the solar azimuth and atmospheric turbidity parameter vectors in the current polarization sensor system. imag(·) and real(·) represent the imaginary and real parts of the complex number. To obtain the polarization angle information in the observation meridian coordinate system, a certain conversion is required:

[0080] φ i (x)=ψ i (x)-α OPi

[0081] like Figure 3 As shown, ψ iAOP of the sensor system, a OPi φ represents the azimuth angle of the current observation vector. The AOP measurement of the sensor also needs to be processed mi mi - a OPi , ψ mi is the measured value in the sensor system, and φ mi is the measured value in the local coordinate system. If the number of pixels in the image area participating in the heading calculation is N, the minimization objective function can be constructed

[0082]

[0083] φ represents the azimuth angle of the current observation vector. The AOP measurement of the sensor also needs to be processed mi , where φ represents the actual measurement of the sky polarization pattern of the i-th pixel point. The Powell least squares optimization algorithm is used to solve the objective function, so as to obtain the estimated polarization sensor system azimuth angle a. According to the local geographical position and time, the sun azimuth angle a in the world coordinate system can be obtained s . The heading angle at the current time can be obtained by subtracting the two.

[0084]

[0085] The solution of the heading angle has a 180° ambiguity, which can be determined by the value given by other navigation systems.

[0086] To verify the effectiveness of the improved polarization field singularity point model proposed in this paper in the heading measurement and field calibration method, polarization optical sensor outdoor data collection was carried out on March 9, 2022 at 15:30. The experiment was carried out in the Innovation Mansion of Harbin Institute of Technology (longitude 126.6236, latitude 45.7261, altitude 148.74m). The experimental equipment is shown in Figure 4 , mainly composed of polarization optical sensor, fiber-optic inertial navigation system, aluminum alloy support, damping tripod, storage battery, DC stabilized power supply and notebook computer. Among them, the polarization optical sensor and the inertial navigation system are fixed on the aluminum alloy support, and the damping tripod is used as the support of the aluminum alloy support, which is convenient for adjusting the pose of the polarization sensor during data collection. The gyro zero offset of the fiber-optic inertial navigation system used in the experiment is 0.02° / h, and its attitude accuracy can reach 0.1° after combination with GPS. After the system is powered on, the initial alignment of the fiber-optic inertial navigation system is carried out first, and then the leveling is carried out according to the attitude output value of the inertial navigation system, and the pitch and roll misalignment angle is controlled within 1°. Then the polarization optical sensor is rotated manually every 10°, and 10 sky polarization images are collected each time. The heading value of the inertial navigation system for single rotation is shown in Table 1, so as to obtain 360 sky polarization pattern images. The AOP images collected are used for heading measurement experiment

[0087] ​To verify the accuracy of the improved polarization field singular point model compared to the traditional sky polarization characterization model, a full-sky polarization model characterization performance experiment was conducted using four sets of AOP images collected at different heading values. The heading values ​​corresponding to the four sets of images were 0°, 90°, 180°, and 270°, respectively. The actual AOP images detected by the polarization sensor, as well as the theoretical AOP images simulated by the Rayleigh model, the polarization field singular point model, and the proposed model are shown in Figure 2. Figure 5 As shown in the figure, during the polarization pattern simulation, the atmospheric turbidity parameter A of the polarization field singular point model was set to 25°. To ensure consistency in simulation conditions, the atmospheric turbidity parameter A in this model was also set to 25°, and the scaling coefficients k1 and k2 were set to 1.5. It can be seen intuitively that the ∞ shape of the Rayleigh model in the image center region closely matches the measured value. However, because the model does not consider the polarization neutral point, the AOP value in the solar region differs significantly from the measured value. The polarization field singular point model, on the other hand, considers the neutral point, and the AOP zero position represents the neutral point position, which is consistent with the measured value. However, the saturation of the ∞ shape still differs significantly from the measured value. The full-sky polarization pattern characterization model proposed in this paper not only includes the neutral point position information but also controls the neutral point in the anti-solar position through the polarization singular point correction coefficient, resulting in a high degree of consistency with the measured value.

[0088] To further quantitatively evaluate the characterization performance of the polarization model, we first calculate the difference between the simulated and measured polarization angles of each polarization unit in a single polarization image. when When the value is less than 10°, the simulation result of the polarization unit is considered to be similar to the measured value. similar and N total Indicates the number of polarization units and the total number of polarization units for which the result is approximate. Define ε = N similar / N total The similarity coefficient of the sky polarization pattern characterization model proposed in this paper is the similarity coefficient between the simulated data and the measured data. The similarity coefficient of the sky polarization pattern characterization model proposed in this paper is superior to that of the Rayleigh model and the polarization field singular point model under different heading states. The average similarity coefficient of the AOP image is 106.35% higher than that of the polarization field singular point model and 14.08% higher than that of the Rayleigh model, verifying the superiority of the sky polarization pattern characterization model proposed in this paper.

[0089] In order to further verify the heading measurement effect of the improved polarization field singular point model, the heading of the 360 ​​AOP images collected was solved. Three areas with field of view (FOV) sizes of 51.8°, 69.1°, and 88.0° were selected, and the heading information of each image was solved using different sky polarization pattern characterization models. The heading calculation method based on the Rayleigh model refers to the least squares method proposed in the literature. The heading solution error is as follows Figure 6As shown in the solving result of the Rayleigh model, with the increase of the field of view range, the solving result is unstable, and even an incorrect result appears. However, the model of the patent can obtain stable results under different field of view ranges, and the field of view range of the heading measurement is more robust. This is because with the increase of the field of view range, the neutral point interval not considered in the Rayleigh model gradually increases, which destroys the characteristics of the Rayleigh model and leads to incorrect solving. In the polarization mode representation process, the neutral point generated by multiple scattering of atmospheric aerosol particles is considered in the model of the patent, so that the navigation information can be well solved even under the condition that the atmosphere is relatively turbid. In summary, the method provided by the application can realize fast and high-precision alignment in the case that there are mixed Gaussian noise and outliers in the observation.

Claims

1. A bionic polarized light heading measurement method based on an improved polarization field singular point model, characterized in that: The following steps are involved: Step 1: Install the imaging bionic polarization sensor and the high-precision fiber-optic inertial navigation system, complete startup, warm-up preparation, and initial alignment, and evaluate the heading measurement results of the bionic polarization sensor using the heading output of the high-precision fiber-optic inertial navigation system as a reference value; Step 2: The bionic polarization sensor collects images and transmits them to the host computer to solve the observation vector, polarization degree and polarization angle information; the polarization state of light can be described by the Stokes vector In the formula Ipk , k =1,2,3,4 represents the light intensity output of the four polarization channels, then the polarization angle (AOP) and polarization degree of the polarization unit for Step 3: Use the position correction coefficient to correct the neutral point position in the original polarization field singular point model, thereby constructing an improved polarization field singular point model for the full sky polarization mode. wi (x), In the formula i is the index of the current observation vector, Indicates the solar azimuth angle under the polarization sensor system at the current moment α , solar altitude angle β , atmospheric turbidity parameters A , and correction coefficient k 1. k 2 constitutes a vector, The observation vector is xoy Complex representation of a plane projection point, is the initial neutral point coordinate The coordinate value after a certain angle around the z-axis is half the azimuth angle of the sun vector in the rotating sensor coordinate system, that is, α / 2, is the radius of the projected circle on the ground plane corresponding to the current observation vector; Step 4: Dedistort the single pixel of the fisheye polarization camera to obtain the observation vector and pixel point ( u , v ) corresponds to the relationship between the polarization sensor and the pixel ( u , v ) will correspond to an observation vector OPi and the corresponding polarization vector , then an observation vector OPi The corresponding polarization angle and polarization degree are the imaginary and real parts of the complex form of the all-sky polarization model represented by the constructed improved polarization field singular point model; Step 5: In order to obtain the polarization angle information in the observation meridian coordinate system, a certain conversion is required. The polarization angle in the observation meridian is obtained by subtracting the azimuth of the observation vector from the measured AOP. , Ψi Indicates the AOP under the sensor system, Indicates the azimuth of the current observation vector. The AOP measurement value of the sensor also needs to be processed. , Ψmi is the measured value under the sensor system, is the measured value in the meridian coordinate system; Step 6: Use the theoretical AOP value under current parameters and measured AOP values Construct a minimization objective function and use the Powell least squares optimization algorithm to solve the objective function to obtain the estimated solar azimuth angle under the polarization sensor system α , according to the local geographical location and time, the solar azimuth in the world coordinate system can be obtained αs The current heading angle can be obtained by taking the difference between the two. There is an ambiguity of 180° in the solution of the heading angle, which can be determined by the values ​​given by other navigation systems.

2. The bionic polarized light heading measurement method based on the improved polarization field singular point model according to claim 1, characterized in that: The specific method for constructing the improved polarization field singular point model of the all-sky polarization mode in step 3 is as follows: In order to quantitatively describe the polarization information of the sky, a Cartesian coordinate system is used to represent an observation vector on the celestial sphere. OP ( x , y , z ) onto the ground plane is , the complex number of the projection point in the polar coordinate system is expressed as ; In order to make the ∞ shape saturation of the polarization field singular point model closer to the real value, two coefficients are introduced k 1 and k 2. Correct the neutral point positions on both sides of the anti-sun position, and the initial neutral point coordinates can be expressed as: ys =(1-tan( β / 2)) / (1+tan( β / 2)) represents the projection point of the sun’s position on the horizon. β is the solar altitude angle, A is the atmospheric turbidity parameter, for The antisymmetric point of is the neutral point on both sides of the sun position and the anti-sun position, thereby obtaining the real-time characterization function of the sky polarization pattern based on the improved polarization field singular point model, which contains the polarization information of the current observation direction: 。

Images