RU2638939C1 - Method of radar ship identification - Google Patents

Abstract

FIELD: radio engineering, communication.

SUBSTANCE: invention can be used in on-board radar stations (OBRS) with synthesizing the antenna aperture for identifying surface objects (ships). The specified result is achieved due to the fact that the radar (RL) image of a section of the sea surface with a detected RL mark, indicating the presence of an object on the sea surface, is subjected to processing using reference matrices containing images of RL shadows formed by ships and obtained on the basis of information on the form and location of the ship relative to the OBRS, and based on the results of this processing, a decision is made on the identity of the detected object to one ship out of a given list of ships to be identified.

EFFECT: identification of ships on the sea surface, regardless of the characteristics of direct radar reflections from the elements of its structure, while the characteristic feature of the method is an increase in the probability of correct identification when the sea surface disturbance increases.

7 dwg

Description

The invention relates to radar and can be used to recognize ships by processing radar images (radar) obtained by radar with synthesizing aperture (SA) of the antenna, based on the selection and analysis of radar shadow (radar) formed by the ship on an excited sea surface.

Currently, methods are known for recognizing extended objects of various types (ships, ground objects of artificial and natural origin) based on the processing of radar images obtained in direct SA mode [1, 2, 3] and in inverse SA mode [4]. In these methods, an image of an object is formed due to radar signals reflected directly from its elements, and various methods of image processing based on spatial filtering using standard descriptions of objects are used for its recognition.

In methods using direct SA mode, the image of objects is obtained in the range-azimuth coordinates, that is, a “top view” of the object is formed. In the inverse CA mode, the operation of which is based on the use of the ship's pitching, when the front-side radiation is applied, the image is obtained as a “side view”, that is, a side silhouette. The silhouette of the ship contains more informational signs necessary for recognizing the class of the ship than its representation in plan, formed in direct SA mode. But in the inverse SA mode, there are limitations (the angular position of the ship relative to the radar, weather conditions) that affect its recognition capabilities. In modes with SA, electromagnetic radiation reflected directly from the object in the direction of the radar is used to generate radar images, so the probability of detecting and recognizing extended objects that use technologies to reduce the effective reflective scattering surface ("stele") is reduced.

The technical result that can be obtained using the present method is to increase the likelihood of recognition of surface objects (ships), as well as the ability to recognize ships made using the "stele" technology.

This result is achieved by the fact that as the main information feature used to recognize the object, the form of radar imagery is used, which is formed by an object on an agitated sea surface when it is irradiated with an airborne radar installed on board an aircraft or spacecraft. The shape of the radar is determined by the dimensions and silhouette of the ship, as well as its angular position relative to the radar. The RLT form contains the external structural features of the ship, which increases the likelihood of its correct recognition.

Recognition of surface ships is as follows. After receiving target designation for any marine object that needs to be recognized, the radar in the CA mode forms the radar image of the sea surface with the specified object. In this case, the necessary radar parameters are calculated on the basis of data on the position of the object relative to the radar — slant range (R) and azimuth, as well as the characteristics of the ships stored in the database (in the list of recognizable ships).

Information about the position of the object relative to the radar can be preliminarily obtained using the radar itself, operating in any radar mode for viewing the sea surface, or transmitted over the communication line from other information sources located on other media.

The obtained radar image has the form of a matrix S (m, n) of size M × N, containing the distribution of radar signal levels over the n-th radar range channels (n = 0, 1, 2, ..., N-1) and the m-th azimuth channels ( m = 0, 1, 2, ..., M-1). The linear resolution in the radar image in range - dr and in azimuth - d a can be different and have a value of approximately several meters.

The value of N is chosen so that the length of the section of the sea surface contained in the radar, in range was not less than 4 times the length L _{m of} the longest radar, which is calculated by the formula:

L _m = H _m R / (H ₀ -H _m ),

where H _m - the highest height for ships from the list of recognizable;

H ₀ - the height of the radar carrier above the sea surface;

R - range to the ship.

N≥4L _m / dr.

The value of M is chosen so that the length of the section of the sea surface contained in the radar image in azimuth is not less than 3 times the value of X _m , which is equal to the length of the longest ship from the list of recognizable:

M≥3X _m / d a .

Each list of recognizable ships corresponds to its own values of H _m and X _m , which are stored in the database. The list may include any number of ships, as well as other marine objects.

The range R ₀ to the beginning of the considered section of the sea surface is chosen so that the object itself is in the center of the radar.

In FIG. Figure 1 shows, for example, the brightness radar image, which contains the marks of the signals reflected from the ship, as well as its radar image (the radar image was obtained using mathematical modeling). In the presented image, the black level corresponds to the minimum signal, and the white level corresponds to the maximum signal contained in the radar image.

The first stage of processing radar data is to exclude from it radar marks corresponding to signals reflected directly from the ship. For this, the average level (S ₀ ) and the maximum level (S _m ) of the radar signals reflected directly from the sea surface are first determined in the obtained radar image. To do this, use the near section of the radar, which obviously does not contain radar, as well as radar signals reflected from the ship (see Fig. 1):

,

,

where N _s = N / 2-X _m / dr.

Then, to remove the ship’s own image from the radar, all elements S (m, n) exceeding the value of S _m are replaced by values equal to S ₀ . The result is a new radar image - matrix S _t (m, n), the form of which is shown in FIG. 2

Next, correlation processing of the obtained matrix S _t (m, n) is carried out using the pre-formed and stored in the database reference matrices T _qab (k, p), which embody the forms of radar generated by recognized ships for different angles at which they are observed from the location of the radar.

Index q of the reference matrices (EM) means the number of the recognized ship (class of the ship) in the list of recognizable ones, and a is the course angle of the ship, given in the number of angular discretes, the value of which Δ a is set with an accuracy of several degrees, and b is the observation angle of the ship in the vertical plane measured in the number of angular discretes, the value of which Δb is set in fractions of a degree.

The numbers k correspond to the range channels (k = 0, 1, 2, ..., K-1), and p to the azimuthal radar channels (p = 0, 1, 2, ..., P-1). The discrete values in range and azimuth for all EMs are set the same as in the matrix S (m, n), and the quantities K and P are determined by the formulas:

,

.

,

.

EMs are prepared in advance and stored in a database. When EM is formed, its elements located inside the RLT are filled with units, and elements outside the RLT are filled with zeros.

In FIG. 3 shows an EM corresponding to the RL scene for which the RLI shown in FIG. 1. EM can be formed using mathematical modeling using 3-D ship models, you can also use any other methods.

Correlation matrices (KM) are calculated by the formula:

,

,

where i = 0, 1, 2, ... M-K-1; j = 0, 1, 2, ... N-P-1;

- средний уровень шумов приемного тракта РЛС.

- the average noise level of the receiving path of the radar.

The total number of CMs obtained for one radar image is equal to the product q a . To compute CM from the database, EMs with index b are selected, which is determined based on the known values of the distance to the object and the own height of the radar carrier by the formula

b = arcsin (H ₀ / R) / Δb.

The average noise level of the radar receiving path, if known and constant, is stored in the database. If this value changes over time, then it is measured in some way before the start of the formation of radar images.

In FIG. 4 shows the CM obtained by processing the X-ray diffraction shown in FIG. 1 using the EM shown in FIG. 3, in the form of a brightness image. In FIG. 5 shows the central part of this CM in the form of a three-dimensional graph. For a better perception, the graph is presented upside down.

The resulting set of CMs is further processed in order to select from the list of recognizable ships, one whose RLT is closest in its form to the RLT contained in the received radar. For this, an algorithm consisting of the following operations is used.

First, in the obtained CM, the values of the average signal level A _qb are calculated:

.

.

For each CM, a signal with a minimum value of B _{q a} is determined, and then the threshold signal level C _{q a is} calculated:

Then, for each CM, the number of its elements D _{q a is} calculated, which have signal levels less than the threshold value C _{q a} , and the coefficient H _{q a is} calculated:

Of all the obtained coefficients H _{q a} selected coefficient having the greatest value and it is concluded that the detected object with an index q, corresponding to the selected coefficient H _{q a,} which has the heading angle corresponding to the selected coefficient and equal to the product and Δ a.

The principle of operation of the described recognition algorithm is explained using the graphs shown in FIG. 6 and FIG. 7, for a simplified case, a two-dimensional representation of RLT images. In FIG. 6 shows the results of processing two RLT images (RLT 1 and RLT 2), which differ in length by four times, using a standard corresponding to RLT 2. At the bottom of FIG. 6 shows graphs obtained as a result of the correlation processing of the RLT and the standard, as well as the corresponding N.

In FIG. Figure 7 shows the results of processing RLT image data using a standard corresponding to RLT 1. In both cases, with the correct choice of the standard, the coefficient H has a value four times greater than with the wrong choice of the standard.

Information sources

1. Tsivlin I.P. Automatic recognition of radar images in an airborne radar // Radio engineering. 2002. No. 9. Issue 65. "Radio-electronic complexes", No. 2. S. 43-50.

2. RF patent 2423722. A method for recognizing surface ships on an agitated sea surface / Verba B.C., Neronsky LB, Osipov I.G. and etc.; Claim 04/07/2010. Publ. 07/10/2011. Bull. No. 19.

3. Ksendzuk A.V., Evseev I.A. Features of detection of objects in bistatic and multi-position SAR // Aviation and space technology and technology. 2005. No2 (18). S. 62-68.

4. Menon M, Bourdreau E., Kolodzy P. An Automatic Ship Classification System for ISAR Imagery // The Lincoln Laboratory Jornal, Vol. 6, Num. 2, 1993, p. 289-308.

Claims (1)

A method for radar recognition of ships, including obtaining by means of an on-board radar station (ARL) with synthesizing an aperture of an antenna of a radar image (RLI) a portion of the sea surface with an object to be recognized by determining its belonging to one of the ships included in a given list of recognizable ships, using known information about the location of the recognized object - azimuth and slant range R relative to the radar carrier, and in radar imagery, comprising a matrix of radar (RL) signal levels S (m, n), the rows correspond to the nth radar range channels with linear resolution dr, the total number of which is N (n = 0, 1, 2, ..., N-1) , the columns correspond to the m-th azimuth radar channels with linear resolution d a , the total number of which is equal to M (m = 0, 1, 2, ..., M-1), and the length of the surface area of the sea surface should be no less than 4 times the length L _{m of} the longest radar shadow (RLT), which can be formed by a ship from a list of recognizable ones and is calculated by the formula L _m = H _m R / (H ₀ -H _m ), where H _m is the highest height for ships from the list of recognizable; H ₀ is the height of the radar carrier above the sea surface, with N≥4L _m / dr, the length of the section of the sea surface contained in the radar, should be at least 3 times the azimuth of X _m , which is equal to the length of the longest ship from a list of recognizable, while M≥3X _m / d a , and each list of recognizable objects has its own values H _m and X _m that are stored in the database, and the range R ₀ to the beginning of the considered surface area is chosen so that the object itself was in the center of radar, floor The radar image is subjected to a sequence of operations, including the exclusion from the radar signal of signals reflected directly from the ship, for which the maximum level S _m and the average level S _{0 of the} radar signal reflected from the sea surface are determined in the radar signal, using for this purpose a radar signal section that is obviously not RLT and RL signals reflected from the ship, according to the formula

, where N _s = N / 2-X _m / dr, and elements of the matrix S (m, n) exceeding the value of S _m are replaced by values equal to S ₀ to obtain the matrix S _t (m, n), processing the obtained matrices S _t (m, n) using the pre-formed and stored in the database reference matrices T _qab (k, p) that store the forms of radar, formed by recognizable ships for different angles at which they are observed from the location of the radar, with obtaining correlation matrices

where

- the average noise level of the radar receiving path, the q index means the number of the recognized ship in the list of recognizable, and - the course angle of the ship, specified in the number of angular discretes, the value of which Δ a is set with an accuracy of several degrees, b - the observation angle of the ship in the vertical plane, measured among the angular discretes, the value of which Δb is set with an accuracy of fractions of a degree, the numbers k and p in the reference matrices (EM) correspond to the range and azimuth channels, and the discrete values in range and azimuth in the EM are the same as in the radar, K = L _m / dr, P = X _m / d a , k = 0, 1, 2, ..., K-1, p = 0, 1, 2, ..., P-1, i = 0, 1, 2, ... , MK-1, j = 0, 1, 2, ..., N-P-1, while the EM elements located inside the RLT have values equal to one, and the elements outside the RLT are zero, the index b used by the EM is determined based on the known values of the distance to the object and the own height of the radar carrier according to the formula b = arcsin (H ₀ / R) / Δb, in the obtained correlation matrices the values of the average signal level are calculated

, find the minimum signal values B _{q a} , calculate threshold signal levels

calculate the number of elements D _{q a} that have signal levels less than the threshold value C _{q a} , calculate the coefficients

, of all the obtained coefficients H _{q a} , the coefficient having the greatest value is selected and it is concluded that an object with the index q corresponding to the selected coefficient H _{q a is} recognized, which has a heading angle corresponding to the selected coefficient H _{q a} and equal to the product a Δ a .

Images