RU2787646C1 - Method for measuring roll and pitch angles of an aircraft - Google Patents

Abstract

FIELD: navigation.

SUBSTANCE: invention relates to navigation and is intended for use as an element of the navigation system of an aircraft. The method for measuring the roll and pitch angles of an aircraft consists in using additional stages of preliminary processing of the observed images before calculating the spectra of the observed images, which involves sequential operations of median filtering of the observed images, linear filtering of the observed images in low-pass filters, increasing the contrast of the observed images, transforming the observed images in outline form. And also, additional stages of pre-processing of the spectra of observed images are used before calculating the sums of amplitudes of all harmonics of the spatial spectrum on the images of the spectra, providing for successive operations of median filtering of the images of the spectra of the observed images, linear filtering in the low-pass filters of the images of the spectra of the observed images, increasing the contrast of the images of the spectra of the observed images. In this case, the sums of the amplitudes of all harmonics of the spatial spectrum are calculated in two mutually perpendicular directions corresponding in the local coordinate system to rotation angles in the range from -45° to 45° with a step of at least 1°, after which the directions are determined in which the sum amplitudes of harmonics on each image has a maximum value. These directions correspond to the angles of rotation of the spectra of the observed images relative to the coordinate systems of the onboard optoelectronic cameras, and, consequently, the roll and pitch angles of the aircraft.

EFFECT: increasing the accuracy of determining the angles of roll and pitch of the aircraft, performed by digital processing of the spectra of spatial frequencies of images and the images themselves, formed by onboard optoelectronic cameras.

1 cl, 4 dwg

Description

The invention relates to navigation and is intended for use in the navigation system of an aircraft.

A known method of digital image processing to determine the spatial position of the aircraft [1]. The method for determining the angular position of an aircraft in space, described in the article, consists in constructing a two-dimensional spectrum of the observed image based on a two-dimensional Fourier transform.

The disadvantage of this method is that to determine the pitch and roll angles of an unmanned aircraft, a discrete Fourier transform is used without ways to speed up its computational procedure, which provides insufficient for real-time calculations and the speed of obtaining results. Another disadvantage is the lack of pre-processing of images generated by the onboard optoelectronic cameras, which increases the error in determining the roll and pitch angles.

As a prototype of the proposed invention, a digital image processing method for determining the spatial position of an aircraft, described in article [2], was chosen. According to this method, the values of roll and pitch angles of an aircraft are measured based on the analysis of images of different degrees of contrast and brightness observed by onboard optoelectronic cameras. The values of the roll and pitch angles are determined from the magnitude of the rotation angles of the spectra of the observed images using the discrete Fourier transform with respect to the camera coordinate systems.

The disadvantage of the prototype is the inability to measure in real time, due to the use of only the discrete Fourier transform, which leads to an increase in the calculation time. Another disadvantage is a high error in calculating the rotation angles of the spectrum, since the original image observed by the onboard camera, which can be of different brightness and contrast, is subjected to the Fourier transform.

The problem to be solved by the proposed invention is the development of a method for measuring the angles of roll and pitch of an aircraft, which allows accurate measurements in real time.

The technical result of the invention is to increase the accuracy of determining the angles of roll and pitch of the aircraft, regardless of the value of the brightness and contrast of the images observed by the onboard cameras, and to reduce the calculation time.

The technical result is achieved by introducing additional stages at which the processing of images observed by onboard optoelectronic cameras and the spatial frequency spectra of such images is carried out, as well as by using the fast Fourier transform algorithm to calculate the spatial spectra of the observed images.

The similarities of the proposed method with the prototype are that sequentially calculate: the spatial spectrum of the images observed by the onboard cameras, using the Fourier transform; on the images of the sum of the amplitudes of all harmonics of the spatial spectrum in directions corresponding to the rotation angles in the local coordinate system, the direction along which the sum of the amplitudes of the harmonics has a maximum value; determine the roll and pitch angles corresponding to the directions with the maximum sum of harmonics.

Differences of the proposed method for measuring the angles of roll and pitch of the aircraft from the prototype are as follows.

Firstly, the calculated spatial spectrum is calculated in the directions corresponding to the rotation angles in the local coordinate system in the range from -45° to +45° with a step of at least 1°.

Secondly, before calculating the spatial spectra of images observed by onboard cameras using the Fourier transform, additional stages of pre-processing of images observed by onboard cameras are introduced, on which:

1) median filtering of observed images,

2) linear filtering of observed images in low-pass filters,

3) increasing the contrast of the observed images,

4) transformation of the observed images into a contour form.

Thirdly, the proposed method includes additional stages of processing the spectra of the observed images before calculating the sums of the amplitudes of all harmonics of the spatial spectrum on the images of the spectra, on which the following is carried out:

1) median image filtering of spectra of observed images,

2) linear filtering in the low-pass filters of the images of the spectra of the observed images,

3) increasing the image contrast of the spectra of the observed images.

Fourth, to calculate the spatial spectra of the observed images, the fast Fourier transform algorithm is used, which ensures the implementation of the computational procedure for determining the roll and pitch angles in real time.

Thus, the proposed method for measuring the roll and pitch angles of an aircraft consists of the following stages: median filtering of the images observed by the onboard cameras, linear filtering of the observed images in low-pass filters, increasing the contrast of the observed images, converting the observed images into a contour form, using Fast Fourier transform calculates the spatial spectrum of images, performs median filtering of images of spectra of observed images, performs linear filtering in low-pass filters of images of spectra of observed images, increases the contrast of images of spectra of observed images, calculates on images the sums of amplitudes of all harmonics of the spatial spectrum in directions corresponding to the local coordinate system rotation angles, determine the direction in which the sum of the amplitudes of the harmonics has a maximum value, determine the rotation angles of the spectrum observably th image corresponding to the directions with the maximum sum of harmonics.

Moreover, the calculation of the sums of the amplitudes of all harmonics of the spatial spectrum is carried out in two mutually perpendicular directions corresponding in the local coordinate system to the rotation angles in the range from -45° to +45° with a step of at least 1°, obtaining a range coverage from 0° to 90 ° in steps of at least 1° in the local coordinate system.

The final step of the method is to determine the directions in which the sum of the harmonic amplitudes in each image has a maximum value. These directions correspond to the angles of rotation of the spectrum of the observed image relative to the coordinate systems of the onboard optoelectronic cameras, and, consequently, the roll and pitch angles of the aircraft.

The essence of the invention is illustrated by drawings.

In FIG. 1 shows an example of a two-dimensional median filter with a 3×3 working window.

In FIG. 2a, 2b, and 2c show the rotation angles of the object in the considered image with a tilt of 0°, 45°, and 60°, respectively.

In FIG. 3a shows an example of an object in the original image, in Fig. Figure 3b shows an example of the spatial spectrum of an object in the processed image.

In FIG. 4 shows the result of calculations corresponding to the angles of rotation (roll/pitch).

In accordance with the claimed method, the following stages of processing the images observed by the onboard cameras are performed.

To reduce the noise level in the observed image, the original image is filtered based on order statistics. To perform filtering on the original image, a two-dimensional median filter can be selected, which is a sliding window covering an odd number of image elements (3×3, 5×5, 7×7). The center element is replaced by the median of the image elements in the window. In FIG. 1 shows an example of a two-dimensional median filter with a 3×3 working window.

As a result of the filter operation, the “outlier” with an amplitude of 18 was removed from the array of numbers describing the processed original image. When working with a median filter, the processed image retains greater sharpness of the contours, since this type of filter has less effect on these image elements than a linear digital low-pass filter (LPF). The median filter, based on order statistics, provides particularly effective salt and pepper noise reduction.

To perform linear filtering of images observed by onboard cameras in low-pass filters, a linear digital filter with a transfer function is used:

Processing in such a filter is performed in accordance with a scheme that provides zero phase distortion in the processed image.

The linear digital filter is based on a sliding horizontal working window with a size of 1×3 and coefficients (0.25; 0.5; 0.25) and a vertical working window with a size of 3×1 and coefficients (0.25; 0.5; 0.25). Processing in the vertical direction is the same as processing in the horizontal direction (with the only difference being that image elements adjacent in a column are used for calculation, and not in a row, as in horizontal processing).

A linear digital filter has a zero phase response when processing an image in the following sequence:

- passage of the horizontal filtering window "from left to right" and "from right to left" (the reverse pass ensures the absence of horizontal phase distortions in the processed image);

- passage of the vertical filtering window "from top to bottom" and "from bottom to top" (the reverse pass ensures the absence of vertical phase distortions in the processed image).

With this filtering method, phase distortions are not introduced into the processed image.

In accordance with the claimed method, the contrast of the original image is increased in order to equalize the amplitude characteristics of the spatial frequency spectrum formed from the processed image.

An image has maximum contrast when its white level is 255 and its black level is 0. Uniform contrast is performed according to the following expressions:

where:

- элемент массива элементов преобразованного изображения;

- element of the array of elements of the transformed image;

- элемент массива элементов исходного изображения;

- element of the array of elements of the source image;

Y _min and Y _max - limit desired values of the minimum and maximum brightness of the processed image;

X _min and X _max - real values of the minimum and maximum brightness of the original image.

The spectrum is calculated using the fast Fourier transform. For the discrete two-dimensional case, the Fourier transform always exists and can be expressed by the following equality:

where f (x, y) is an array of brightness values of the current image frame - a function of two variables with dimensions M×N; x=0, 1, 2, ..., M-1, y=0, 1, 2, ..., N-1, u=0, 1, 2, ..., M-1, v=0, 1, 2, …, N-1;

x and y are spatial or image variables;

u and v are transformation variables or frequency variables.

This expression must be evaluated for all u and v. When working with images, the frequencies that make up the spectrum are usually called spatial frequencies, which numerically characterize the number of repetition periods of the brightness values of image elements that change according to a sinusoidal law per unit length.

The type and structure of the two-dimensional amplitude spatial spectrum of the image allows you to highlight the presence of vertical and horizontal boundaries and lines in the image and give a general idea of their spatial orientation. As can be seen in figure 2, the orientation of the two-dimensional amplitude spatial spectrum of the image characterizes in the local coordinate system the direction of the boundaries and lines in the original image.

Next, the operations of median filtering of the spectra of the observed images, linear filtering in the low-pass filters of the spectra of the observed images and increasing the contrast of the images of the spectra of the observed images are carried out.

In conclusion, the operation of determining the roll and pitch angles is performed. The direction is determined in two mutually perpendicular directions (Fig. 3a, 3b, 4), corresponding in the local coordinate system to the angles of rotation in the range from -45° to 45° with a step of at least 1°, which increases the accuracy of determining the angles of rotation spectrum.

The lines corresponding to the maximum sum of harmonic amplitudes form a deviation from the vertical axis of the local coordinate system.

Thus, the use of additional stages, at which the processing of images observed by onboard optoelectronic cameras and the spatial frequency spectra of such images, makes it possible, regardless of the brightness and contrast values of images observed by onboard cameras, to increase the accuracy of determining the roll and pitch angles of the aircraft, as well as to reduce calculation time by using the fast Fourier transform algorithm to calculate the spatial spectra of the observed images.

List of sources:

[1] Makarenko A.A. Application of digital image processing to determine the spatial position of the aircraft / A.A. Makarenko, L.S. Turnetsky, A.G. Karmanov // Information and space. - 2013. - No. 1. - S. 30-34.

[2] Makarenko A.A. Algorithms for digital image processing in tasks of visual navigation of an unmanned aerial vehicle / A.A. Makarenko, L.A. Vinokurov // Radio industry. - 2020. - №3. - S. 75-85.

Claims (1)

A method for measuring the roll and pitch angles of an aircraft, which consists in calculating the spatial spectrum of the image using the Fourier transform, calculating the sum of the amplitudes of all harmonics of the spatial spectrum on the images, determining the direction along which the sum of the harmonic amplitudes has a maximum value, determining the roll and pitch angles corresponding to the directions with the maximum sum of harmonics, characterized in that the spectra are calculated using the fast Fourier transform method and before calculating the spectra of the observed images, median filtering of the observed images, linear filtering of the observed images in low-pass filters are performed, the contrast of the observed images is increased, and the observed images are converted into a contour form, and before calculating the sums of the amplitudes of all harmonics of the spatial spectrum on the images of the spectra, median filtering of the images of the spectra of the observed images, a linear filter low-pass filters of the images of the spectra of the observed images, increase the contrast of the images of the spectra of the observed images, while the sums of the amplitudes of all harmonics of the spatial spectrum are determined in two mutually perpendicular directions corresponding to the rotation angles in the local coordinate system in the range from -45° to +45° s steps of at least 1°.

Images