CN-121995338-A - Pulse type radio frequency interference judging method, system, storage medium and electronic equipment for multi-domain characteristic entropy weight fusion

CN121995338ACN 121995338 ACN121995338 ACN 121995338ACN-121995338-A

Abstract

The invention discloses a pulse radio frequency interference judging method, storage medium and electronic equipment for multi-domain feature entropy weight fusion, which comprises the steps of carrying out pulse-by-pulse processing on interference echo data, respectively extracting time domain statistical center deviation degree features, time domain sharpness features, frequency domain kurtosis features and frequency domain skewness first-order difference features, constructing a pulse-by-pulse multi-domain feature vector sequence, then carrying out self-adaptive weighted fusion on the multi-domain features based on an information entropy theory to obtain pulse-by-pulse interference detection score, constructing statistical histogram distribution according to the detection score, carrying out Gaussian smoothing processing, further carrying out segmentation by adopting an OTSU self-adaptive threshold value judging method to realize automatic judgment on the interference pulses containing pulses, and outputting pulse interference judging results. According to the invention, the abnormal characteristics of pulse interference in the multi-domain feature space can be effectively represented under the condition that manual threshold setting is not needed, and the judgment accuracy and stability under the condition of weak pulse interference are improved.

Inventors

LI NING
GAO YUAN
SHU GAOFENG
WU LIN
ZHAO JIANHUI
HUANG YABO

Assignees

河南大学

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (10)

1. The pulse radio frequency interference judging method for the multi-domain feature entropy weight fusion is characterized by comprising the following steps of: Step S101, carrying out pulse-by-pulse processing on echo data, and extracting time domain statistical center deviation degree and sharpness characteristics; Step S102, extracting frequency domain kurtosis and skewness first order difference features pulse by pulse, and constructing a multi-domain feature vector sequence; step S103, carrying out self-adaptive weight determination on the pulse-by-pulse multi-domain feature vector sequence based on an information entropy theory; step S104, carrying out weighted fusion on the pulse-by-pulse multi-domain feature vector sequences to obtain an interference detection score sequence; step S105, constructing statistical histogram distribution according to the detection score, and performing Gaussian smoothing; Step S106, adopting OTSU self-adaptive threshold judgment to segment, and realizing automatic judgment of interference pulses containing pulses; Step S107, outputting the impulse interference judgment result.
2. The method for judging pulse radio frequency interference by fusion of multi-domain feature entropy weight according to claim 1, wherein in step S101, echo data is processed pulse by pulse to extract the feature of deviation of time domain statistical center and sharpness, specifically: the time domain statistical center deviation characteristic calculation formula is expressed as follows: ; Wherein, the Represent the first A sample of the magnitude of the signal, The number of bits in the sample is indicated, Representing the absolute mid-level difference of the light beam, A minute positive number introduced to prevent zero removal; Then, for the first The deviation degree of the statistical center is calculated by the amplitude sequence of each pulse, and the maximum value is taken as the characteristic value of the pulse, which can be expressed as follows: ; Wherein the method comprises the steps of , Represent the first Pulse number The distance is directed towards the sample and, Indicating the total number of pulses in the azimuth direction, Is the median absolute deviation; by performing the above processing pulse by pulse, the time domain statistical center deviation characteristic sequence can be obtained : ; Is the first The characteristic value of the deviation degree of the statistical center corresponding to each pulse; the time domain sharpness feature calculation formula is expressed as: ; Wherein, the And Representing the minimum and maximum values within the sample point left sliding window respectively, And Respectively representing the minimum value and the maximum value in the sample point right sliding window, and squaring operation is used for enhancing the response capability to peak mutation; Then, for the first And for highlighting extreme peak characteristics, selecting high-order statistics from the normalized peak characteristics as the time-domain peak characteristics of the pulse, wherein the result can be expressed as: ; Wherein, the , Represent the first The sequence of sharpness corresponding to the individual pulses, A high quantile ratio approaching 1; the high quantile ratio The value range of (0.9, 0.999), and by performing the above processing on a pulse-by-pulse basis, a sequence of time-domain sharpness features for all pulses can be obtained The form thereof can be expressed as: ; By calculating the characteristics pulse by pulse, a complete time domain statistical center deviation characteristic sequence SCD and a complete time domain sharpness characteristic sequence Sharp can be obtained, and an input basis is provided for the subsequent multi-domain characteristic entropy weight fusion.
3. The method for judging the pulse radio frequency interference by fusion of the multi-domain feature entropy weight according to claim 1, wherein in step S102, frequency domain kurtosis and skewness first order difference features are extracted pulse by pulse, and a multi-domain feature vector sequence is constructed, specifically: the kurtosis characteristic calculation formula is expressed as: ; Wherein, the Representing the first in a sequence of spectral magnitudes The point of the sample is taken, , ; Representing the total number of frequency domain sample points, The frequency domain kurtosis coefficient is the ratio of the square of the fourth-order center moment and the square of the second-order center moment; Then, for the first Frequency domain transformation is carried out on each pulse to obtain the frequency spectrum And taking the amplitude value of the frequency spectrum to obtain a frequency spectrum amplitude value sequence On the basis, calculating the corresponding frequency domain kurtosis characteristic The result can be expressed as: ; Wherein, the , Indicating the total number of azimuth pulses; by performing the above-described processing pulse by pulse, a frequency domain kurtosis signature sequence of all pulses can be obtained, which can be expressed in the form: ; the frequency domain bias characteristic calculation formula can be expressed as: ; Wherein, the Representing the first in a sequence of spectral magnitudes A number of sampling points are used to sample the sample, Representing the total number of frequency domain sampling points, The frequency domain skewness is obtained by passing a third-order central moment and a second-order central moment The ratio of the powers characterizes the asymmetry of the spectrum distribution, and the structure has higher sensitivity to frequency domain energy offset and spectrum profile asymmetry change; Then, for the first The time domain echo of each pulse is subjected to frequency domain transformation to obtain the frequency spectrum And taking the amplitude value of the frequency spectrum to obtain a frequency spectrum amplitude value sequence On the basis, the corresponding frequency domain skewness is calculated, and the result can be expressed as: ; Wherein, the , Indicating the total number of azimuth pulses; To further characterize the degree of variation of frequency domain statistics between adjacent pulses, a frequency domain skewness first order difference feature is introduced, defined as the absolute value of the difference between adjacent pulse frequency domain skewness: ; Wherein, the When (when) When the frequency domain skewness first-order difference value is defined as zero; the frequency domain bias first-order difference characteristic sequence corresponding to all the pulses is obtained by executing the processing on a pulse-by-pulse basis The form thereof can be expressed as: ; After obtaining pulse-by-pulse time domain statistical center deviation feature, time domain sharpness feature, frequency domain kurtosis feature and frequency domain deviation first order difference feature, to realize multi-angle joint characterization of pulse interference, uniformly modeling the features of different domains and different behavioural mechanisms to construct a pulse-by-pulse multi-domain feature vector sequence; Specifically, for the first Pulse, the corresponding time domain statistical center deviation degree characteristic is obtained Time domain sharpness feature Kurtosis characteristics in frequency domain Frequency domain skewness first order difference feature Performing joint characterization to form a multi-domain feature vector corresponding to the pulse, wherein the expression can be expressed as follows: ; Wherein, the , Indicating the total number of azimuth pulses; By performing the feature construction process described above for all pulses, a complete pulse-by-pulse multi-domain feature vector sequence can be obtained, the overall form of which can be expressed as: 。
4. The method for judging pulse radio frequency interference by fusion of multi-domain feature entropy weight according to claim 1, wherein in step S103, adaptive weight determination is performed on pulse-by-pulse multi-domain features based on information entropy theory, specifically: the pulse-by-pulse multi-domain feature vector sequences obtained in the step S101 and the step S102 are jointly characterized according to pulse indexes to form a pulse-by-pulse multi-domain feature matrix: ; Wherein, the The dimension of the feature is represented and, Representing the total number of pulses; In order to eliminate the difference of different features in dimension, amplitude range and statistical distribution and avoid adverse effect of the feature numerical scale on the weight distribution result, the feature matrix is normalized according to the feature dimension, and the expression is: ; Wherein, the , To prevent minor positive numbers introduced by zero denominators, Is a normalization result; Based on the normalized features, taking the values of the features on the pulse sequence as discrete random variables, and constructing probability distribution corresponding to the features: ; further, based on the information entropy theory, carrying out quantitative evaluation on uncertainty and discrete degree of each feature in the pulse dimension, and calculating the first Information entropy corresponding to each feature: ; is a base 10 logarithm; According to the information entropy result, introducing a difference coefficient to describe the effective information contribution degree of each feature to interference discrimination: ; And carrying out normalization processing on the difference coefficient to obtain self-adaptive weights corresponding to the features: 。
5. The method for judging pulse radio frequency interference by multi-domain feature entropy weight fusion according to claim 1, wherein in step S104, the pulse-by-pulse multi-domain features are weighted and fused to obtain an interference detection score sequence, specifically: The multi-domain features are subjected to linear weighted fusion according to the self-adaptive weights to obtain pulse-by-pulse comprehensive interference discrimination scores: ; Wherein, the Is the self-adaptive weight corresponding to each characteristic; The comprehensive discrimination score For characterising the first The interference significance degree of each pulse is used as the input basis of the subsequent Gaussian smoothing histogram.
6. The method for determining the pulse radio frequency interference by fusion of multi-domain feature entropy weight according to claim 1, wherein in step S105, a statistical histogram distribution is constructed according to the detection score, and a gaussian smoothing process is performed, specifically: First, a score sequence is detected for pulse-by-pulse interference And (3) performing linear normalization processing, wherein the expression is as follows: ; Wherein, the A minute positive number introduced to prevent zero denominator; to facilitate subsequent statistical distribution-based threshold analysis, the normalized detection score is mapped to a limited discrete gray space, specifically, to an 8-bit gray interval The expression is: ; On this basis, a statistical histogram is constructed for the mapped detection score sequence, and the histogram count function can be expressed as: ; Wherein, the Indicating that the detection score falls within the first The number of pulses in a single gray scale interval, ; The histogram count sequence is subjected to gaussian smoothing, and the smoothing result thereof can be expressed as: ; Wherein, the Is the standard deviation of the gaussian kernel, is used to control the smoothing intensity, Representing the mean.
7. The pulsed radio frequency interference judging method of multi-domain feature entropy weight fusion according to claim 1, wherein in step S106, the OTSU adaptive threshold decision is adopted for segmentation, so as to realize automatic judgment of pulse-containing interference pulses, specifically: the histogram is normalized and converted into a probability distribution form so as to eliminate the influence of pulse quantity and data scale change on a threshold value calculation result, and the result is obtained: ; on the basis, candidate threshold values are introduced Dividing the detection score sample into a low score and a high score, and respectively corresponding to a potential normal pulse set and an interference-containing pulse set; Further calculating the cumulative probability and the cumulative mean value corresponding to the threshold value: ; The cumulative probability is represented as a function of the probability, Representing a cumulative mean; Meanwhile, defining a global average value of all samples as follows: ; based on the statistics, an inter-class variance function is constructed for measuring the threshold Separation ability for two types of samples: ; Threshold for maximizing inter-class variance by searching in full gray scale Obtaining an optimal self-adaptive segmentation position; mapping the optimal gray threshold back to the normalized detection score domain to obtain a final decision threshold: ; Wherein, the The optimal gray threshold value which is obtained by searching through the OTSU method and maximizes the inter-class variance is obtained, and finally, when the pulse-by-pulse interference detection score is obtained Greater than the adaptive threshold And if not, judging that the corresponding pulse is a pulse-containing interference pulse, otherwise, judging that the pulse is a normal pulse, thereby completing the interference automatic judgment of the pulse-by-pulse layer.
8. The pulse type radio frequency interference judging system with the multi-domain characteristic entropy weight fusion is characterized by comprising: The feature extraction unit is configured to extract time domain statistical center deviation and sharpness features from echo data containing interference of an input system pulse by pulse, and calculate frequency domain kurtosis and frequency domain deviation first order difference features pulse by pulse so as to construct multi-domain feature vectors of each pulse; the feature weight calculation unit is configured to calculate difference coefficients of the features on the basis of an information entropy theory for the multi-domain feature vector sequence, and normalize the difference coefficients to generate corresponding self-adaptive weights; The interference judging unit is configured to carry out weighted fusion on the multi-domain features according to the self-adaptive weights of the multi-domain features to obtain a pulse-by-pulse interference detection score sequence, map the pulse-by-pulse interference detection score sequence to a gray space to construct a histogram, and judge pulses by adopting an OTSU self-adaptive threshold after Gaussian smoothing treatment so as to realize automatic judgment of pulses containing pulse interference; And a result output unit configured to generate a pulse-type interference judgment sequence for use in a subsequent interference suppression process according to the pulse-by-pulse judgment result.
9. A storage device in which a plurality of programs are stored, wherein the program application is loaded and executed by a processor to implement the pulsed radio frequency interference determination method of multi-domain feature entropy weight fusion of any one of claims 1-7.
10. An electronic device comprises a storage device, a processor, the processor is suitable for executing each program, and the memory is used for storing a plurality of programs, and is characterized in that the pulse radio frequency interference judging method for fusion of multi-domain characteristic entropy weights according to any one of claims 1-7 is realized when the memory executes the programs on the processor.

Description

Pulse type radio frequency interference judging method, system, storage medium and electronic equipment for multi-domain characteristic entropy weight fusion Technical Field The application relates to the technical field of signal processing, in particular to a pulse type radio frequency interference judging method, a system, a computer readable storage medium and electronic equipment for multi-domain characteristic entropy weight fusion. Background Synthetic aperture radar (SYNTHETIC APERTURE RADAR, SAR) is an active microwave imaging system capable of achieving all-weather, all-day high resolution ground observation. SAR obtains high distance resolution by transmitting broadband signals, combines platform motion to realize high azimuth resolution, and is widely applied to the fields of topographic mapping, disaster monitoring, agricultural monitoring, military reconnaissance and the like. However, in increasingly complex electromagnetic environments, SAR systems are inevitably affected by radio frequency interference (Radio Frequency Interference, RFI), which is a typical form of RFI. PRFI are typically derived from ground based radar or communication equipment with frequencies overlapping the SAR signal, and have unidirectional propagation, high power, short duration, and short pulse periodicity characteristics, resulting in bright line or spike structures in the SAR echo, severely affecting the effective reception and subsequent imaging quality of the echo signal. Currently, the judging method for impulse interference mainly focuses on two types of time domain and transform domain. The time domain method characterizes transient characteristics of interference pulses by means of amplitude mutation, peak value or differential analysis and the like, but is easy to miss detection or generate false detection in weak interference or noise environments, and the transform domain method analyzes interference characteristics by relying on spectral energy concentration, spectral asymmetry or short-time Fourier transform, can reflect concentration and non-stable change of interference in a frequency domain, and usually ignores dynamic change information among pulses. In addition, the existing method mostly adopts fixed threshold value judgment, and is difficult to adapt to different interference intensities and complex scenes, so that the detection result is unstable or the error is larger. Especially in the weak energy interference scene, the traditional method is easy to generate missed judgment, and in the strong interference scene, a large number of false positives can be generated, which directly restricts the subsequent interference suppression and the improvement of SAR imaging quality. Disclosure of Invention The application aims to provide a pulse radio frequency interference judging method, a system, a storage medium and electronic equipment for multi-domain feature entropy weight fusion, so as to solve or alleviate the problems in the prior art. In order to achieve the above object, the present application provides the following technical solutions: The application provides a pulse radio frequency interference judging method of multi-domain feature entropy weight fusion, which comprises the steps of carrying out pulse-by-pulse processing on echo data, extracting time domain statistical center deviation degree and sharpness features, carrying out pulse-by-pulse extracting frequency domain kurtosis and deviation degree first order difference features, constructing a multi-domain feature vector sequence, carrying out self-adaptive weight determination on the pulse-by-pulse multi-domain features based on an information entropy theory, carrying out weighted fusion on the pulse-by-pulse multi-domain features to obtain an interference detection score sequence, carrying out step S105, constructing statistical histogram distribution according to detection scores, carrying out Gaussian smoothing processing, carrying out step S106, adopting OTSU self-adaptive threshold judgment to carry out segmentation, realizing automatic judgment of pulse-containing interference pulses, and outputting a pulse interference judging result, wherein the step S107 is carried out. Preferably, in step S101, echo data is processed pulse by pulse, and features of time domain statistical center deviation and sharpness are extracted, specifically: the time domain statistical center deviation characteristic calculation formula is expressed as follows: Wherein, the Represent the firstA sample of the magnitude of the signal,The number of bits in the sample is indicated,Representing the absolute mid-level difference of the light beam,A small positive number is introduced to prevent zero removal. Then, for the firstThe deviation degree of the statistical center is calculated by the amplitude sequence of each pulse, and the maximum value is taken as the characteristic value of the pulse, which can be expressed as follows: Wherein the method com