RU2537696C1 - Method of selection of moving targets - Google Patents

Abstract

FIELD: physics, navigation.

SUBSTANCE: invention relates to the field of radiolocation and is intended for usage in radars for detection of moving targets against the background of reflections from the ground surface. The difference between current and previous radar frames is determined, several previous radar frames are averaged, the obtained difference radar frame is divided into the fields with the given size and in the resulted fields the points of maximum density of brightness marks in the frame is determined iteratively using the special formula.

EFFECT: decrease of probability of identification of false targets and probability of target loss.

4 dwg

Description

Technical field

The invention relates to the field of radar and is intended for use in radar stations for detecting moving targets against a background of reflections from the earth's surface.

State of the art

The invention is most consistent with the methods of selection of moving targets, called inter-period subtraction of signals (Bakulev P.A., Stepin V.M. Methods and devices for selecting moving targets. - M .: Radio and communications, 1986 - 288 s).

Known methods for detecting moving targets by finding the difference between the current and previous (previous) radar frames and subsequent threshold processing of the resulting difference (Fukunaga K., Hostetler L., D. The Estimation of the Gradient of a Density Function, with Applications in Pattern Recognition. - IEEE Transactions on Information Theory (IEEE), vol. IT-21, NO. 1.01. 1975. - pp. 32-40).

The disadvantage of such methods is that these methods determine the probability of detecting false targets and the probability of missing targets, which are unacceptable for a number of applications.

A known method of signal processing against the background of strong pulsed interference in the receiving channel of a pulse-Doppler radar station, including incoherent accumulation, post-detector incoherent processing of received signals is performed within a single sliding analysis window by sequentially performing the following operations: first weighing in accordance with the law of determining weight coefficients, which is inversely proportional to the law describing the antenna radiation pattern (BOTTOM); pairwise selection of at least two for weighted signals in adjacent taps; the second weighing from the exits of the selections for a minimum of two, in which the law of determining the weighting coefficients is directly proportional to the square of the law describing the BOTTOM; summation after the second weighing (Patent RU 2334247 C1, publ. 09/20/2008).

A disadvantage of the known technical solution is that its implementation uses an expensive coherent-pulse radar. One of the most significant drawbacks of such radars is the presence of “blind speeds”. This disadvantage manifests itself in the fact that targets moving at speeds that are multiples of the probing pulse repetition rate, or moving along a tangential trajectory around the radar, cannot be detected.

SUMMARY OF THE INVENTION

An object of the present invention is to provide a method for selecting moving targets, which makes it possible to distinguish moving targets from data from both coherently pulsed and non-coherently pulsed radars.

The technical result achieved in the implementation of the proposed solution is to reduce the likelihood of detecting false targets and the probability of missing targets.

The technical result is achieved by the fact that in the method for selecting moving targets, including finding the difference between the current and previous radar frames with subsequent threshold processing, according to the proposed solution, several previous frames are averaged, the resulting differential radar frame is divided into areas of a given size and iteratively determined points of maximum density of brightness marks in the frame according to the formula:

, $$m\left(x\right)=\frac{{\displaystyle \sum {{}_x}_{{}_i}{}_{\in N\left(x\right)}K\left(x_i-x\right)x_i}}{{\displaystyle \sum {{}_x}_{{}_i}{}_{\in N\left(x\right)}K\left(x_i-x\right)}}$$

,

where N (x) are the points of brightness marks from a neighborhood of the point x for which K (x) ≠ 0;

K (x) is the Gaussian core;

; $$K\left(x_i-x\right)=e^{-\mu {\left(x_i-x\right)}^2}$$

;

µ is a constant scale factor;

x is the reference point of the image area whose position in the iteration loop does not change;

x _i are the points of the image area that are sequentially sorted to calculate the sums in the estimate m (x).

In the analyzed literature, this combination of distinctive features was not found. This set of distinctive features does not follow explicitly for a specialist from the prior art, and this technical solution can be successfully used in industry. Thus, the proposed technical solution meets the criteria of the invention of "Novelty", "Inventive step", "Industrial application".

Brief Description of the Drawings

The invention is illustrated by drawings.

Figure 1 presents the structural diagram of the implementation of the method;

Figure 2 shows one of the frames of the brightness radar image obtained from the radar station;

Figure 3 presents the differential radar image obtained by subtracting the current image from the averaged image;

Figure 4 shows the result of determining the points of maximum density, that is, the result of detecting a moving target.

The structural diagram for implementing the method includes random access memory 1 (RAM 1), one output of which is connected to the input of the second storage device 2 (RAM 2), the second output of RAM 1 and the output of the second RAM 2 are connected to the inputs of the adder 3, the output of which is connected to the correlator four.

The implementation of the invention

In an example of a specific embodiment, random access memory devices RAM 1 and RAM 2 are implemented on an SRM 20100 LMT chip, an adder 3 is implemented on a K525PS3 chip, and a correlator 4 is implemented on an ADSP 2105 signal processor.

During the review period in RAM 1, a frame of the brightness radar image is accumulated (Fig. 2), from where it is copied and summed with the contents of RAM 2. Thus, in RAM 2 an averaged radar image is formed. The adder 3 reads the images in RAM 1 and RAM 2, and sums them pixel by pixel, and the pixel values read from RAM 2 are taken with a negative sign. As a result, a differential image is formed (Fig. 3). The differential image is fed to the correlator 4, is processed according to the formula:

. $$m\left(x\right)=\frac{{\displaystyle \sum {{}_x}_{{}_i}{}_{\in N\left(x\right)}K\left(x_i-x\right)x_i}}{{\displaystyle \sum {{}_x}_{{}_i}{}_{\in N\left(x\right)}K\left(x_i-x\right)}}$$

.

At the beginning of the jth iteration cycle, the position of the point x is corrected by substituting the position estimate obtained from the previous iteration j-1. If the density maximum is in the region, then the point x and the estimate m (x) converge. In this case, the inequality

, $$\left|x_j-m\left(x_{j-\mathrm{one}}\right)\right|<\epsilon$$

,

where ε is the predetermined accuracy in advance.

The maximum density in the segmented image is indicated by points of higher intensity than the points of the region where the above inequality is not satisfied. Consequently, regions that do not contain density maxima where inequality is not satisfied are marked by low-intensity points.

As a result, a segmented image is formed (Fig. 4), bright dots of high intensity of which indicate found moving targets.

Claims (1)

A method for selecting moving targets, including finding the difference between the current and previous radar frames with subsequent threshold processing, characterized in that several previous frames are averaged, the resulting differential radar frame is divided into areas of a given size and the points of maximum density of brightness marks in the frame are iteratively determined according to the formula:
$$m\left(x\right)=\frac{{\displaystyle \sum {{}_x}_{{}_i}{}_{\in N\left(x\right)}K\left(x_i-x\right)x_i}}{{\displaystyle \sum {{}_x}_{{}_i}{}_{\in N\left(x\right)}K\left(x_i-x\right)}}$$

,
where N (x) are the points of brightness marks from a neighborhood of the point x for which K (x) ≠ 0,
K (x) is the Gaussian core;
$$K\left(x_i-x\right)=e^{-\mu {\left(x_i-x\right)}^2}$$

µ is a constant scale factor;
x is the reference point of the image area whose position in the iteration loop does not change;
x _i are the points of the image area that are sequentially sorted to calculate the sums in the estimate m (x).

Images