Polarization navigation method in night shielding environment

Abstract

A model is constructed through the summation, denoising, and enhancement of multiple images. The Stokes vectors of detected light are calculated from the model and are then used to determine the direc

Description

TECHNICAL FIELD The present invention belongs to the technical field of polarized light navigation, and is particularly to a polarization navigation method in a night shielding environment. BACKGROUND A high-precision navigation sensor is an important guarantee for an unmanned motion platform to complete a task autonomously, and "inertial + satellite" combined navigation is the most commonly used navigation mode at present. An inertial navigation system is prone to drift in long endurance, and a satellite signal is vulnerable to a natural or human interference. Recent biological researches show that, in nature, sand ants and dung beetles have an ability to navigate by an atmospheric polarization mode, wherein a polarization sensitive unit in a compound eye structure of the sand ants may highly precisely acquire the atmospheric polarization mode to achieve navigation solution; and the dung beetles rely on a polarization visual nerve pathway to enhance a perception system to perceive dim moonlight at night, so as to determine a position to achieve a linear movement, which provides a new solution for autonomous navigation without relying on a satellite. Daytime polarization navigation is widely studied in autonomous navigation, but nighttime polarization is relatively less concerned. A nighttime weak polarized light field may pose a challenge to available navigation information collection, photon emission noise at night is relatively greater than that at day, so that a signal-to-noise ratio of photons at night is low, and it is easily disturbed by noise, and a light intensity at night is 6-8 orders of magnitude lower than that at day, so that long-time exposure is required to achieve an orientation effect equivalent to that at day, but a sampling frequency is seriously reduced. In addition, a bionic polarization compass also faces the problem of local polarization information loss caused by shielding of trees, buildings and the like in a practical application process, especially under a coupling effect of a night weak polarization mode, a polarization orientation precision is reduced sharply. SUMMARY Objective of invention: in order to solve the problems of large noise, long exposure time and low orientation precision of a polarization navigation image in a night shielding environment in the prior art, the present invention provides a polarization navigation method in a night shielding environment. Technical solution: a polarization navigation method in a night shielding environment comprises the following steps: first step: constructing a space-time summation enhancement model, which comprises image denoising and image enhancement; second step: calculating a polarization degree and a polarization angle of an enhanced image by a Stokes vector method, extracting an effective area according to the polarization degree, and searching positive and negative feature points of the effective area according to the polarization angle; and third step: dividing the enhanced image into m rings at equal intervals, continuously adjusting axis positions, and respectively calculating weighted difference sums of positive and negative feature points of effective areas in the m rings under different axes, wherein an axis corresponding to a minimum value of the weighted difference sum is a solar azimuth in a carrier coordinate system; obtaining a solar azimuth in a navigation coordinate system through an astronomical almanac; and calculating a difference between the solar azimuth in the carrier coordinate system and the solar azimuth in the navigation coordinate system to obtain an absolute heading angle. Further, in the first step, the image denoising comprises: denoising a polarization image by double filtering to obtain a denoised image, which comprises smoothing and removing background noise by space filtering, and removing bipolar impulse noise by time filtering. Further, wherein, in the first step, a method for the image enhancement comprises: inputting a plurality of continuous denoised images into a plurality of visual channels correspondingly, and accumulating signals input from the plurality of visual channels to a single cell with a wide dendritic field, so as to enhance perception of light. Further, in the second step, a method for extracting the effective area according to the polarization degree comprises: obtaining an axis L by fitting the feature points, taking a zenith point as a center and taking the axis L as a baseline to rotate by 30° left and right respectively, dividing the image into six areas, and marking the areas in sequence clockwise, wherein a first area, a fourth area, a third area and a sixth area are effective areas, and a second area and a fifth area are interference areas; and setting a 2 polarization angle of a feature point as ” and setting a polarization angle threshold a to select the feature points, defining a feature point with a polarization angle satisfying a< B <90° _ as the positive feature point, and defining a feature point with a _~ >ft >—on0 polarization angle satisfying p - as the negative feature point. Further, in the third step, the weighted difference sum is set as 5, and a calculation formula of 5 is: (1) <y = l- (2) wherein, / is a number of positive feature points of an effective area in an ith ring, is a number of negative feature points of the effective area in the ith ring, i-r is a number of positive feature points in the ith ring, i-r is a number of negative feature points in the ith ring, co represents a symmetry weight of the positive p and negative feature points of the effective area in the ith ring, is a positive feature N point of the effective area, and c" is a negative feature point of the effective area. Further, in the third step, a method for determining the minimum value of the weighted difference sum 5 comprises: controlling the axis L to rotate by an angle 9 each time, and automatically adjusting a precision and calculating a complexity, so that an axis position corresponding to the minimum value of 5 is found quickly, W + X) = Kp* 8(k) + Kd * (W - -1)) (3} K K wherein, p is a proportional coefficient, d is a differential coefficient, and k is an axis L of rotation. Compared with the prior art, the polarization navigation method in the night shielding environment provided by the present invention has the following beneficial effects: the problem of poor orientation precision in the night shielding environment is solved, the double denoising is proposed for the large noise of the polarization image collected at night, the background noise is smoothed by space filtering, and the bipolar impulse noise is processed by time filtering; the image is enhanced by a dung beetlelike visual nerve summation enhancement method; there is no higher requirement for exposure time; the effective area is selected by area division to extract the feature points, which reduces the interference of shielding factors; and the problems of solar meridian fitting deviation and poor orientation precision caused by asymmetry of the positive and negative feature points are solved by a solar meridian fitting method with asymmetrical distribution of positive and negative polarization angles based on PD adjustment, so that a calculation speed is improved, and an environmental adaptability of night polarization navigation is effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a flow chart of a polarization navigation method in a night shielding environment; FIG. 2 is an orientation result of a building shielding experiment in a night weak polarization mode; FIG. 3 is an orientation result of a tree shielding experiment in the night weak polarization mode; and FIG. 4 is a box plot of errors in the building shielding and tree shielding experiments in the night weak polarization mode. DETAILED DESCRIPTION The present invention is further explained and described hereinafter with reference to the drawings and specific embodiments. A polarization navigation method in a night shielding environment, as shown in FIG. 1, comprises the following steps. In first step, a space-time summation enhancement model is constructed, which specifically comprises: denoising a polarization image by double filtering to obtain a denoised image, which comprises smoothing and removing background noise by space filtering, and removing bipolar impulse noise by time filtering; and enhancing the denoised image by a dung beetle-like visual nerve summation enhancement method to obtain an enhanced image. An atmospheric polarization mode signal under ideal conditions is set as , electrical noise and channel noise are classified as Gaussian white noise , and weather noise makes image brightness distribution uneven, which produces bipolar impulse noise nN) , so that the polarization image may be expressed as: ^1(0 = ^(0 + ^1+^(0 m Tl The background noise 1 is smoothed by the space filtering, and a space filter is expressed as: 1 K(n) = . -e V2n<r (2) wherein, g represents a standard deviation of the space filter, and n is an exponent. A coefficient of the space filter is acquired through a second derivative, (3) The filter is normalized: 1 1 _ Vm a Normalized n2 2 1)-e -n1 ve 20-2 n (4) In order to achieve less ripple and narrower filtering, a stacking coefficient c is added for regulation, so as to filter and remove the Gaussian white noise, as shown in Formula (5), Heap (5) A method for removing the bipolar impulse noise by the time filtering is as follows. The impulse noise ^2^) is decomposed first to obtain noise of M times, and the noise of M times is weighted to obtain a first-order modal component: i M ^(0 = -^^(0 M (6) y (f\ A residual signal is calculated: ^1(0 = ^(0-^1(0 (7) component is decomposed The residual signal 1V ’ is decomposed by Formula (6) for M times to obtain a second-order modal component, each order modal repeatedly, and an original signal is expressed as: 1 1 n2 -V (0 = .< / ) + c. ---------+ E, y e^’ (8) A method for enhancing the denoised image comprises: inputting through a plurality of visual channels, which comprises inputting a plurality of continuous denoised images into the plurality of visual channels correspondingly, and accumulating to a single cell with a wide dendritic field, so as to enhance perception of light (inputting the plurality of images into the plurality of channels, and accumulating the outputs of the plurality of channels, so as to enhance the light). In second step, a polarization degree and a polarization angle of the enhanced image are calculated by a Stokes vector method, an effective area is extracted by shielding area division according to the polarization degree, and feature points near an absolute value 90° of the polarization angle are searched according to the polarization angle. A specific method for extracting the effective area comprises: obtaining an axis L by fitting the feature points, taking a zenith point as a center and taking the axis L as a baseline to rotate by 30° left and right respectively, dividing the image into six areas, and marking the areas in sequence clockwise, wherein a first area, a fourth area, a third area and a sixth area are effective areas, and a second area and a fifth area are interference areas; and setting a polarization angle of a feature point as and setting a polarization angle threshold a to select the feature points, defining a feature point with a polarization angle satisfying _ p ~ as the positive feature point, and defining —a >B> —90° a feature point with a polarization angle satisfying ~ p ~ as the negative feature point. In third step, the enhanced image is divided into m rings at equal intervals, axis positions are continuously adjusted, and weighted difference sums of positive and negative feature points of effective areas in the m rings under different axes are calculated respectively, wherein an axis corresponding to a minimum value of the weighted difference sum is a solar azimuth in a carrier coordinate system; a solar azimuth in a navigation coordinate system is obtained through an astronomical almanac; and a difference between the solar azimuth in the carrier coordinate system and the solar azimuth in the navigation coordinate system is calculated to obtain an absolute heading angle. The weighted difference sum is set as 5, and a calculation formula of 5 is: 8=y d. \o)     ,, * P. cd ff* N. ) * a) \ ip_r_eff i_r m r ejj i_r J (9) wherein, !-r-< is a difference between positive and negative feature points of an effective area in an ith ring, is a probability of positive feature points in the P CD ith ring, i-r is a number of the positive feature points in the ith ring, in-r-eff is a , N probability of negative feature points in the i ring, ! is a number of the negative feature points in the ith ring, co is a weight factor, and co is inversely proportional to the difference between the positive and negative feature points of the effective area; D =P -N P reff lyi_r_eff QQ) P wherein, ! r-< is a number of the positive feature points of the effective area in the ith ring, and is a number of the negative feature points of the effective area in the ith ring; co =— ip-r-eff Pi r+Ni CD ,, mreff (11) (12) The weighted difference sum is simplified by Formulas (9), (11) and (12) as follows: * cd (13) 0 = 1- P fe + N jr eff eff (14) P N eff is a positive feature point of the effective area, is a negative feature point of the effective area, and co represents a symmetry weight of the positive and negative feature points of the effective area in the ith ring. A method for determining the minimum value of the weighted difference sum 5 comprises: controlling the axis L to rotate by an angle 9 each time, and automatically adjusting a precision and calculating a complexity, so that an axis position corresponding to the minimum value of 5 is found quickly, W +1) = Kp* 8(^) + Kd * W) (: 5) K K wherein, p is a proportional coefficient, d is a differential coefficient, and &refers to an axis L of rotation. In order to verify the accuracy of the polarization navigation method in the night shielding environment in this embodiment, the following experiments are made respectively. FIG. 2 and FIG. 3 show experimental results of building shielding and tree shielding in a night weak polarization mode respectively, orientation results of this method are compared with orientation results of the least square method and reference values respectively, and FIG. 4 is a box plot of errors in the two experiments. It can be seen that the heading angle obtained by this method has higher accuracy and smaller error.

Claims

1. A polarization navigation method in a night shielding environment, comprising the following steps:first step: constructing a space-time summation enhancement model, which comprises image denoising and image enhancement;second step: calculating a polarization degree and a polarization angle of an enhanced image by a Stokes vector method, extracting an effective area according to the polarization degree, and searching positive and negative feature points of the effective area according to the polarization angle; andthird step: dividing the enhanced image into m rings at equal intervals, continuously adjusting axis positions, and respectively calculating weighted difference sums of positive and negative feature points of effective areas in the m rings under different axes, wherein an axis corresponding to a minimum value of the weighted difference sum is a solar azimuth in a carrier coordinate system; obtaining a solar azimuth in a navigation coordinate system through an astronomical almanac; and calculating a difference between the solar azimuth in the carrier coordinate system and the solar azimuth in the navigation coordinate system to obtain an absolute heading angle.

2. The polarization navigation method in the night shielding environment according to claim 1, wherein, in the first step, the image denoising comprises: denoising a polarization image by double filtering to obtain a denoised image, which comprises smoothing and removing background noise by space filtering, and removing bipolar impulse noise by time filtering.

3. The polarization navigation method in the night shielding environment according to claim 1 or 2, wherein, in the first step, a method for the image enhancement comprises: inputting a plurality of continuous denoised images into a plurality of visual channels correspondingly, and accumulating signals input from the plurality of visual channels to a single cell with a wide dendritic field, so as to enhance perception of light.

4. The polarization navigation method in the night shielding environment according to claim 1 or 2, wherein, in the second step, a method for extracting the effective area according to the polarization degree comprises: obtaining an axis L by fitting the feature points, taking a zenith point as a center and taking the axis L as a baseline to rotate by 30° left and right respectively, dividing the image into six areas,and marking the areas in sequence clockwise, wherein a first area, a fourth area, a third area and a sixth area are effective areas, and a second area and a fifth area are interference areas; and setting a polarization angle of a feature point as and setting a polarization angle threshold a to select the feature points, defining a feature point with a polarization angle satisfying _ p ~ as the positive feature point, and defining —a >B> —90°a feature point with a polarization angle satisfying - p - as the negative feature point.

5. The polarization navigation method in the night shielding environment according to claim 1 or 2, wherein, in the third step, the weighted difference sum is set as 5, and a calculation formula of 5 is:69 = 1-(1)-Nr_eff i _r _effP ff + N ,,.eff eff(2)wherein, / is a number of positive feature points of an effective area in anith ring, ! r cl1 is a number of negative feature points of the effective area in the ith P Nring, i-r is a number of positive feature points in the ith ring, i-r is a number of negative feature points in the ith ring, co represents a symmetry weight of the positivepand negative feature points of the effective area in the ith ring, is a positive featureNpoint of the effective area, and c" is a negative feature point of the effective area.

6. The polarization navigation method in the night shielding environment according to claim 1 or 2, wherein, in the third step, a method for determining the minimum value of the weighted difference sum 5 comprises:controlling the axis L to rotate by an angle 9 each time, and automatically adjusting a precision and calculating a complexity, so that an axis position corresponding to the minimum value of 5 is found quickly,W + X) = Kp* S(k) + Kd * tf(k) - 5(k -1)) (3}K Kwherein, p is a proportional coefficient, d is a differential coefficient, and k is an axis L of A111 rotation.T +44(0)30 0300 2000A

Images