CN114936576B - Subblock differential coding distribution characteristic extraction method in multi-rotor unmanned aerial vehicle identification - Google Patents

Abstract

The invention belongs to the technical field of unmanned aerial vehicle identification, and particularly relates to a subblock differential coding distribution feature extraction method in multi-rotor unmanned aerial vehicle identification. The method comprises the steps of firstly, carrying out short-time Fourier transform on a radar echo data sequence of an unmanned aerial vehicle target to obtain a time-frequency spectrogram, then, dividing the time-frequency spectrogram into a plurality of sub-blocks which are not overlapped, differentiating the element value of each sub-block from the mean value of the sub-blocks, forming a coding sequence according to positive and negative coding and row sequence of the differential value, counting probability distribution of corresponding numerical values of the coding sequence of each sub-block of the whole time-frequency spectrogram, and taking the probability distribution as the distribution characteristic of the differential coding of the sub-blocks to complete the identification of the multi-rotor unmanned aerial vehicle. Due to the fact that the distribution information of the sub-block structure is fully utilized, local features of related targets can be extracted from radar echo data, detail difference among the targets is increased, and therefore the recognition rate of the targets is improved. The effectiveness of the method is verified according to simulation experiment results of the 4-type multi-rotor unmanned aerial vehicle.

Description

Subblock differential coding distribution characteristic extraction method in multi-rotor unmanned aerial vehicle identification

Technical Field

The invention belongs to the technical field of unmanned aerial vehicle identification, and particularly relates to a subblock differential coding distribution feature extraction method in multi-rotor unmanned aerial vehicle identification.

Background

With the rapid development of unmanned aerial vehicle technology, the application in military and civil fields is more and more extensive, and great threat is brought to the safety of the country, and it is necessary to carry out counterattack battle on the unmanned aerial vehicle. Therefore, the method has very important practical significance for national defense combat by accurately identifying the type of the unmanned aerial vehicle.

At present, a subspace method is a classical method for identifying unmanned aerial vehicles, and a transformation subspace is mainly established through a training data set of targets of the unmanned aerial vehicles, so that the classification features of the targets are extracted. However, the conventional subspace can only extract global structural features, but weakens local structural features, and reduces differences in target details, so that the target recognition rate of the conventional multi-rotor unmanned aerial vehicle subspace recognition method has room for further improvement.

Disclosure of Invention

The invention aims to provide a subblock differential coding distribution feature extraction method, which can extract local structural features reflecting target details by dividing a time-frequency spectrogram into subblocks and counting distribution information of subblock differential coding sequences, thereby increasing the difference degree of the target in the details and improving the identification rate of the target.

The technical scheme of the invention is as follows:

the subblock differential coding distribution feature extraction method in multi-rotor unmanned aerial vehicle identification comprises the following steps:

s1, defining the acquired radar echo training data sequence of the multi-rotor unmanned aerial vehicle as an n-dimensional column vector x _ij ，i＝1,2,…g，j＝1,2…N _i Wherein i represents the category of the drone, g represents the number of categories, N _i The training sample number of the ith unmanned aerial vehicle target is represented, and the total training sample number is

S2, training sample data x _ij Performing short-time Fourier transform to obtain time-frequency spectrogram S _ij ：

S _ij ＝[s _ij,km ] _K×M (1)

Wherein s is _ij,km Representation of a time-frequency spectrum S _ij The elements in (A) K =1,2, \8230AK, M =1,2, \8230AM, K is a row subscript indicating a frequency change direction, M is a column subscript indicating a time change direction, and K is S _ij M is S _ij The number of columns;

s3, time-frequency spectrogram S _ij Dividing the coded block into subblocks to obtain subblock differential coding sequences:

3x3 window in time-frequency spectrogram S _ij Performing non-overlapping sliding to obtain multiple 3x3 sub-blocks to form sub-block set P _ij ：

P _ij ＝[W _ij,1 W _ij,2 … W _ij,L ] (2)

Wherein, W _ij,l Represents a subblock set P _ij The first sub-block matrix of (1), w _ij,l,tf Representing a subblock matrix W _ij,l L represents the total number of subblocks; calculating the subblock matrix W _ij,l Mean value of all elements in

/>

Subtracting the average value from the sub-block elements to obtain:

wherein

For the element subtracted from the mean in a sub-block, pair +>

The following encoding is performed:

wherein, c _ij,l,tf Is that

Corresponding codes, wherein each element code in the sub-block forms a differential coding sequence c of 0, 1 and 2 according to the sequence of lines _ij,l ：

c _ij,l ＝[c _ij,l,11 c _ij,l,12 … c _ij,l,33 ] (7)

S4, extracting the distribution characteristics of the sub-block differential coding:

coding sequence c of difference _ij,l Calculating the numerical value of the time-frequency spectrogram as a ternary number, counting the repeated occurrence times of the differential coding sequence value of each sub-block in the whole time-frequency spectrogram, and forming a vector:

[η _ij,1 η _ij,2 … η _ij,L ] (8)

wherein eta is _ij,l Is a coding sequence c _ij,l The number of repeated occurrences of (D) is the time-frequency spectrogram S _ij Corresponding subblock differential coding distribution feature vector h _ij Comprises the following steps:

after the characteristics are obtained, the target can be classified and identified by using a nearest neighbor classifier.

The method has the advantages that the distribution information of the sub-block structure is fully utilized, the local characteristics of related targets can be extracted from radar echo data, the detail difference among the targets is increased, and the target recognition rate is improved.

Detailed Description

The following simulations were combined to demonstrate the effectiveness and progress made by the present invention:

adopt 4 types of unmanned aerial vehicle to simulate, including three rotor unmanned aerial vehicle, four rotor unmanned aerial vehicle, six rotor unmanned aerial vehicle, eight rotor unmanned aerial vehicle, its simulation parameter is shown as table 1. The simulated radar parameters comprise: the radar carrier frequency is 24GHz; the pulse repetition frequency is 100KHz; the distance between the target and the radar is 200m; the unmanned aerial vehicle has a pitch angle of 10 degrees and an azimuth angle of 30 degrees relative to the radar

Each type of target records 10s of radar echo signals and divides the radar echo signals into segments with fixed length of 0.05s (at least including one rotation period), the overlap between the segments is 50%, each segment includes 0.05 × 100000=5000 radar echo sampling data points, and each type has 400 segments. In the 400 segments, 200 segments are randomly selected as training data sets, and the other 200 segments are used as test data sets, so that the training data set of the 4 types of targets comprises 800 segments in total, and the test data set comprises 800 segments. For the selected training data set, the subblock differential coding distribution characteristics are extracted by using the text method, a template library is established, the subblock differential coding distribution characteristics of the test sample are extracted, a nearest neighbor classifier is adopted for classification, and the average correct recognition rate of the 4 types of multi-rotor unmanned aerial vehicles reaches 95%. Wherein the signal-to-noise ratio is 15dB.

Table 1 simulation parameters of four unmanned aerial vehicles

/>

Claims (1)

1. Sub-block differential coding distribution feature extraction method in multi-rotor unmanned aerial vehicle identification is characterized by comprising the following steps:

s1, defining the acquired radar echo training data sequence of the multi-rotor unmanned aerial vehicle as an n-dimensional column vector x _ij ，i＝1,2,…g，j＝1,2…N _i Where i represents the unmanned aerial vehicle class, g represents the number of classes, N _i The training sample number of the ith unmanned aerial vehicle target is represented, and the total training sample number is

S2, training sample data x _ij Performing short-time Fourier transform to obtain time-frequency spectrogram S _ij ：

S _ij ＝[s _ij,km ] _K×M

Wherein s is _ij,km Representation of a time-frequency spectrum S _ij The elements in (A) K =1,2, \8230AK, M =1,2, \8230AM, K is a row subscript indicating a frequency change direction, M is a column subscript indicating a time change direction, and K is S _ij M is S _ij The number of columns;

s3, comparing the time frequency spectrogram S _ij Dividing the coded block into subblocks to obtain subblock differential coding sequences:

3x3 window in time-frequency spectrogram S _ij Performing non-overlapping sliding to obtain multiple 3x3 sub-blocks to form sub-block set P _ij ：

P _ij ＝[W _ij,1 W _ij,2 …W _ij,L ]

W _ij,l ＝[w _ij,l,tf ] _3×3 ,t,f＝1,2,3

l＝1,2,…L

Wherein, W _ij,l Represents a set of subblocks P _ij The first sub-block matrix of (1), w _ij,l,tf Representing a subblock matrix W _ij,l L represents the total number of subblocks; calculating the subblock matrix W _ij,l Mean value of all elements in

Subtracting the average value from the sub-block elements to obtain:

wherein

For the element from which the mean value has been subtracted in the sub-block, pair->

The following encoding is performed:

wherein, c _ij,l,tf Is that

Corresponding codes, wherein each element code in the sub-block forms a differential coding sequence c of 0, 1 and 2 according to the sequence of lines _ij,l ：

c _ij,l ＝[c _ij,l,11 c _ij,l,12 …c _ij,l,33 ]

S4, extracting sub-block differential coding distribution characteristics:

coding sequence c of difference _ij,l Calculating the numerical value of the time-frequency spectrogram as a ternary number, counting the repeated occurrence times of the differential coding sequence value of each sub-block in the whole time-frequency spectrogram, and forming a vector:

[η _ij,1 η _ij,2 …η _ij,L ]

wherein eta is _ij,l Is a coding sequence c _ij,l The number of repeated occurrences of (D) is the time-frequency spectrogram S _ij Corresponding subblock differential coding distribution feature vector h _ij Comprises the following steps:

/>