Short-wave interference feature extraction method based on ionized layer image processing

Abstract

The invention discloses a short-wave interference feature extraction method based on ionosphere image processing, which comprises the following steps: step 1, carrying out preprocessing and denoising on ionosphere image data by adopting a digital image processing algorithm; step 2, carrying out analysis and calculation according to frequency spectrum, time domain and space domain information of short-wave signals when interference features are extracted; and establishing an interference feature library, and judging whether the short-wave signal is interfered or not through comparative analysis and feature matching. According to the method disclosed by the invention, the short-wave interference features are automatically extracted through an algorithm, manual intervention is reduced, working efficiency and accuracy are improved, and automation is realized; processing and analysis are carried out based on ionized layer image data, multiple interference factors are comprehensively considered, and the systematicness of interference feature extraction is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of ionospheric image processing, and in particular relates to a shortwave interference feature extraction method in this field. Background Art

[0002] Shortwave communication is widely used in wireless communication and has advantages such as strong penetration and long transmission distance. However, since the shortwave frequency band is easily affected by various interference factors such as weather changes, geomagnetic activity, and solar radiation, the quality of shortwave communication is prone to decline, and even communication interruption may occur. In order to solve this problem, researchers began to focus on the extraction of shortwave interference features.

[0003] Shortwave interference feature extraction is to analyze and identify various interference situations that affect the quality of shortwave communication, so as to take corresponding interference suppression or compensation measures. The impact of weather changes on shortwave communication is a common interference factor. When weather conditions change, such as thunderstorms, high humidity, etc., it will cause sudden changes in the ionosphere, causing multipath diffraction and scattering of shortwave signals, interfering with the propagation of the original signal. In addition, geomagnetic activity will also interfere with shortwave communications. Changes in the earth's magnetic field will cause changes in ionospheric parameters, which will in turn affect the transmission of shortwave signals. Solar radiation will produce phenomena such as solar flares and explosions. These solar activities will release a large number of ionospheric particles and affect the reflection and absorption characteristics of the ionosphere, interfering with the propagation of shortwave signals.

[0004] In order to extract the characteristics of shortwave interference, researchers have used a variety of methods and techniques. These include spectrum analysis, time domain analysis, and wavelet transform based on signal processing technology. The spectrum analysis method determines the frequency characteristics of the interference signal by analyzing the spectrum of the shortwave signal, thereby identifying and classifying the interference. The time domain analysis method analyzes the time domain waveform of the shortwave signal to extract the duration, fluctuations and other characteristics of the interference signal, thereby judging and compensating for the interference. The wavelet transform method decomposes the signal into wavelet packets on the time-frequency scale and extracts interference characteristics on different scales to achieve the description and analysis of shortwave interference.

[0005] The research direction of shortwave interference feature extraction covers multiple fields such as signal processing, pattern recognition, and remote sensing. By extracting the characteristic information of shortwave interference, we can better understand and deal with the impact of various interference factors on shortwave communication, and improve the performance and robustness of shortwave communication systems. In the future, with the advancement of technology and in-depth research, shortwave interference feature extraction will become an important research direction in the field of shortwave communication, providing more effective methods and means to achieve reliable shortwave communication.

[0006] In existing research methods, shortwave interference feature extraction is mainly based on empirical rules and expert experience, which lacks systematicity and accuracy. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide a method for automatically extracting shortwave interference features so as to improve the recognition and analysis capabilities of shortwave interference.

[0008] The present invention adopts the following technical solution:

[0009] A shortwave interference feature extraction method based on ionospheric image processing, the improvement of which is that it comprises the following steps:

[0010] Step 1: Use digital image processing algorithms to preprocess and denoise the ionospheric image data:

[0011] Acquire ionospheric image data received by shortwave signals, and use digital image processing algorithms to preprocess and denoise the ionospheric image data to improve data quality and reliability;

[0012] Step 2: When extracting interference features, analyze and calculate based on the spectrum, time domain and spatial domain information of the shortwave signal:

[0013] Extract interference features from ionospheric image data, including spectrum features, time-frequency features, and time-space features;

[0014] Step 3: Establish an interference feature library and determine whether the shortwave signal is interfered by comparative analysis and feature matching:

[0015] An interference feature library is established to store the feature information of various shortwave interferences. Through comparative analysis, the features of the ionospheric image data received in real time are matched with the interference feature library. Based on the matching results, it is determined whether the shortwave signal is interfered with, as well as the type and degree of interference.

[0016] Furthermore, the step 1 is specifically as follows:

[0017] Step 11, obtain the ionospheric image data received by shortwave signal:

[0018] Collect ionospheric imaging data using the Global Ionospheric Mapper or Ionospheric Sounding Radar;

[0019] Step 12: Preprocess and denoise the ionospheric image data using digital image processing algorithms:

[0020] a) Image grayscale adjustment: use histogram equalization to even out the grayscale distribution of the image:

[0021] Calculate the grayscale histogram of the original image, calculate the cumulative histogram, and calculate the grayscale mapping function based on the cumulative histogram:

[0022]

[0023] In the above formula, r represents the original grayscale, L represents the total number of grayscales, N and M represent the width and height of the image respectively, and n j Represents the number of pixels at gray level j;

[0024] b) Image filtering:

[0025] Median filtering: Select a filter window size, which will slide on the image; sort the pixels in the window by value, and select the middle value as the new value of the current pixel; repeat the above steps until the filtering operation is completed on the entire image. new (x,y)=median(I(x,y));

[0026] Gaussian filtering: Create a Gaussian template. The size and standard deviation of the template determine the size and strength of the filter. Apply the template to each pixel of the image and calculate the weighted average as the new value of the current pixel. Repeat the above steps for the entire image. new (x,y)=∑(w*I(x+i,y+j)) / ∑(w);

[0027] In the above formula, I(x,y) represents the pixel value in the original image; I new (x, y) represents the pixel value after filtering; median is the value in the middle of a group of data sorted by size; (x, y) represents the coordinates of the pixel; (x+i, y+j) represents the coordinates of the adjacent pixels in the filter window; w represents the weight of the corresponding pixel;

[0028] c) Image Enhancement:

[0029] Histogram Equalization:

[0030]

[0031] In the above formula, T(r) is the output gray level corresponding to the input gray level r, L is the total number of gray levels, MN is the total number of pixels in the image, and p r (j) is the normalized frequency of pixels with gray level j in the input image in the entire image;

[0032] Contrast Enhancement:

[0033] I {text{new}} (x,y)=A*I {text{old}} (x,y)+B

[0034] In the above formula, I {text{new}} (x, y) is the enhanced image pixel value, A and B are the gain and bias parameters respectively, I {text{old}} (x,y) is the original image pixel value;

[0035] d) Image restoration:

[0036] The interpolation algorithm estimates the pixel values ​​of the missing parts by analyzing and interpolating known data points, including linear interpolation, bilinear interpolation, and cubic spline interpolation;

[0037] Model-based methods use prior knowledge of image structure and texture to build mathematical models to describe image features and use these models for repair; partial differential equation-based methods use image gradient information to restore missing or damaged areas; optimization-based methods seek the best repair result by minimizing a certain objective function. Optimization methods include non-negative matrix decomposition, sparse representation, and total variation minimization.

[0038] Step 13, data quality assessment:

[0039] The quality of the ionospheric imagery data was assessed using the following quality metrics:

[0040] a) Signal-to-noise ratio:

[0041] Assuming the average signal strength of the region of interest in the image is S and the surrounding background noise strength is N, the signal-to-noise ratio SNR is expressed as:

[0042] b) Image clarity:

[0043] Assume that the original image is I1, the preprocessed image is I2, and the image clarity C is expressed as: C = f(SSIM(I1, I2), Sharpness(I1, I2));

[0044] SSIM(I1,I2) represents the structural similarity index of the original image and the preprocessed image, Sharpness(I1,I2) represents the sharpness of the original image and the preprocessed image, and f represents a function that comprehensively calculates the structural similarity index and sharpness to obtain the image sharpness;

[0045] Weighted average: C = w1×SSIM(I1,I2)+w2×Sharpness(I1,I2), where w1 and w2 are weights and w1+w2=1;

[0046] c) Uniformity and consistency:

[0047] Assume that the ionospheric image data is I, a certain area in the image is A, and the area of ​​interest in the image is R;

[0048] Uniformity is expressed by the following formula: For area A:

[0049]

[0050] In the above formula, |A| represents the number of pixels in area A;

[0051] The consistency is expressed by the following formula: For different regions B and C:

[0052]

[0053] The formula for evaluating data reliability is expressed as:

[0054] Reliability = f(overall uniformity (R), overall consistency (R)), where f is a function defined according to specific evaluation criteria. If uniformity and consistency are equally important, f is defined as:

[0055] f(uniformity, consistency) = (uniformity + consistency) / 2.

[0056] Furthermore, the step 2 is specifically as follows:

[0057] Step 21, spectrum feature extraction:

[0058] Perform Fourier transform on the ionospheric image data to obtain its spectrum information, analyze the peak value and frequency band range in the spectrum, as well as the frequency distribution, and extract the spectrum characteristics;

[0059] Peak analysis:

[0060] Let the frequency in the spectrum be f, the energy be E(f), and the peak value be defined as: for a given frequency range, there exists fp such that for any f, there is E(f p )≧E(f);

[0061] The number of statistical peaks is expressed as: where [E(f) = E(f p )] is the indicator function, which takes the value of 1 when E(f) is at its peak value, and 0 otherwise;

[0062] The physical phenomenon related to the peak value is expressed as: P(f p )-Φ(f p ), where P(f p ) represents the peak value f p Related physical phenomena, Φ(f p ) represents the research function thereof;

[0063] Frequency distribution analysis: Calculate the statistical indicators of frequency distribution, including the mean, variance and kurtosis of the spectrum;

[0064] Step 22, time-frequency feature extraction:

[0065] Use short-time Fourier transform and continuous wavelet transform to extract the energy change and frequency evolution of the time-frequency image from the time-frequency image, calculate the entropy, variance and kurtosis of the time-frequency image, and use machine learning methods to classify or extract features from the time-frequency image;

[0066] Step 23, spatiotemporal feature extraction:

[0067] a) Statistical indicator analysis: Calculate the mean, variance and standard deviation of ionospheric image data in time series;

[0068] b) Time series correlation analysis: Calculate the correlation coefficient between ionospheric image data to measure the correlation between different time series;

[0069] c) Differential analysis: By performing differential operations on ionospheric image data, the difference values ​​between adjacent time points are calculated to extract the trend characteristics of time series changes;

[0070] Assume that the ionospheric image data is at time point t i The value of I(t i ), then the adjacent time point t i and t i+1 The difference between them is: ΔI(t i )=I(t i+1 )-I(t i ); the rate of change of the data after differentiation is expressed as:

[0071] d) differential relative motion analysis;

[0072] e) Spatiotemporal feature extraction.

[0073] Furthermore, the step 3 is specifically as follows:

[0074] Step 31, establish an interference feature library:

[0075] Collect different types of shortwave interference data, including spectrum, time-frequency and space-time features, extract features from the interference data, extract spectrum features, time-frequency features and space-time features, and establish an interference feature library;

[0076] When establishing an interference feature library, the extracted feature information is stored in a database or data storage structure, and is organized and stored according to interference categories and time dimensions;

[0077] Step 32, real-time feature matching:

[0078] Acquire the real-time received ionospheric image data, extract the features of the real-time data, extract the spectrum features, time-frequency features and time-space features, compare and analyze the features in the interference feature library, calculate the similarity or distance measurement between them, and judge whether the shortwave signal is interfered, as well as the type and degree of interference based on the similarity or distance measurement results;

[0079] Step 33, interference identification and discrimination:

[0080] Based on the results of real-time feature matching, determine whether the shortwave signal is interfered with. If interfered with, the type and degree of interference are determined based on the feature information of the matching results.

[0081] The beneficial effects of the present invention are:

[0082] The method disclosed by the present invention automatically extracts shortwave interference features through an algorithm, reduces manual intervention, improves work efficiency and accuracy, and realizes automation; processes and analyzes based on ionospheric image data, comprehensively considers multiple interference factors, and improves the systematic nature of interference feature extraction; through real-time matching with an interference feature library, it is timely determined whether the shortwave signal is interfered with, providing timely support for interference identification and processing, and is real-time. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 It is a schematic flow diagram of the method of the present invention;

[0084] Figure 2 It is a schematic flow diagram of step 1 of the method of the present invention;

[0085] Figure 3 It is a schematic flow diagram of step 2 of the method of the present invention;

[0086] Figure 4 It is a schematic flow diagram of step 3 of the method of the present invention. DETAILED DESCRIPTION

[0087] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0088] Embodiment 1, this embodiment discloses a method for extracting shortwave interference features based on ionospheric image processing, such as Figure 1 As shown, the following steps are included:

[0089] Step 1: Use digital image processing algorithms to preprocess and denoise the ionospheric image data, such as Figure 2 As shown, the following steps are included:

[0090] Step 11, obtaining ionospheric image data received by shortwave signals.

[0091] Specifically, in step 11, during the shortwave signal reception process, obtaining accurate ionospheric image data is essential for understanding and processing the propagation characteristics of shortwave signals. Ionospheric image data provides information on the vertical distribution, strength, and changes of the ionosphere, helping to identify the propagation path of shortwave signals, diagnose ionospheric anomalies and interference, and optimize the performance of shortwave communication systems.

[0092] In order to obtain ionospheric image data, a variety of ground or satellite observation equipment are used. The Global Ionospheric Imager (GII) is a commonly used ground-based observation equipment for real-time or near-real-time global monitoring of the ionosphere. GII uses multiple frequency band receivers and antenna arrays that work 24 hours a day to collect shortwave signals from all over the world, and obtains ionospheric image data through data processing and analysis. Ionospheric sounding radar is also an important device that measures the reflection coefficient and scattering characteristics of the ionosphere by transmitting and receiving radar beams, and then infers the state of the ionosphere.

[0093] The data collection methods of these ground or satellite observation equipment include global networks, satellite links, network receiving stations, etc., to ensure coverage of a wide geographical area and diverse ionospheric conditions. The ionospheric image data collected by these devices include parameters such as vertical ionospheric density, frequency delay, phase change, etc., forming ionospheric images with time and space resolution capabilities.

[0094] After obtaining the ionospheric image data, data processing and analysis are required. Common processing methods include model-based inversion technology, signal processing algorithms, etc., to obtain more accurate and detailed ionospheric characteristic information.

[0095] By using global ionospheric imagers, ionospheric sounding radars and other equipment to obtain ionospheric image data, we can better understand the nature and changes of the ionosphere during the shortwave signal reception process, thereby improving the reliability and performance of shortwave communications.

[0096] Step 12: Preprocess and denoise the ionospheric image data using a digital image processing algorithm.

[0097] Specifically, in step 12, in order to improve the quality and reliability of the ionospheric image data, a digital image processing algorithm is used to pre-process and denoise the data. The processing method includes:

[0098] 1. Image grayscale adjustment: adjust the grayscale of ionospheric images to enhance and optimize the display effect.

[0099] Image grayscale adjustment is achieved through different algorithms and methods to enhance and optimize the display effect of ionospheric images. Histogram equalization is used to even out the grayscale distribution of the image and increase the contrast and details of the image.

[0100] The basic idea of ​​histogram equalization is to map the original grayscale of the image to a new grayscale so that the cumulative histogram of the new grayscale is evenly distributed. This is described by the following formula:

[0101] s=T(r)

[0102] Among them, r represents the original grayscale, s represents the adjusted grayscale, and T(.) represents the grayscale mapping function.

[0103] The histogram equalization process is divided into three steps:

[0104] a) Calculate the grayscale histogram of the original image. The grayscale histogram is a distribution function that describes the number of times different gray levels appear in an image.

[0105] b) Calculate the cumulative histogram. The cumulative histogram is the integral of the grayscale histogram, which indicates the cumulative proportion of pixels below each grayscale level.

[0106] c) Calculate the grayscale mapping function based on the cumulative histogram. The commonly used mapping function is to linearly map the grayscale of the original image through the cumulative histogram, so that the grayscale distribution of the output image is more uniform. Use the following formula to calculate the mapping function:

[0107]

[0108] Among them, N and M represent the width and height of the image respectively, n j represents the number of pixels at gray level j, and L represents the total number of gray levels.

[0109] By applying the grayscale mapping function, the grayscale of the original image is converted into the adjusted grayscale, thereby realizing the grayscale adjustment of the image. The adjusted image has better contrast and visual effect, which is helpful for better observation and analysis of ionospheric images.

[0110] Image grayscale adjustment may change the overall tone and brightness of the image. Therefore, when applying the grayscale adjustment algorithm, it is necessary to adjust and control it according to the needs of the specific application and the actual situation to ensure that the adjusted image can still accurately express the meaning of the original data.

[0111] 2. Image filtering: Use filtering algorithms (such as median filtering and Gaussian filtering) to remove noise and interference in ionospheric images.

[0112] Image filtering is a commonly used image processing technology, which is used to remove noise and interference in images to improve the quality and clarity of images. Among them, median filtering and Gaussian filtering are the two most common filtering algorithms.

[0113] a) Median Filter:

[0114] Median filtering is a nonlinear filtering method. Its basic principle is to replace the current pixel value with the median of the pixel values ​​in the neighborhood around the pixel point to achieve the effect of removing noise. The specific steps are as follows:

[0115] Choose a filter window size that slides across the image.

[0116] For the pixels in the window, sort them according to the size of the value and select the middle value as the new value of the current pixel.

[0117] Repeat the above steps until the filtering operation is completed for the entire image.

[0118] Median filtering can effectively remove isolated noise or speckle noise and maintain the details of the image edge, but it will also cause image blur to a certain extent.

[0119] b) Gaussian Filter:

[0120] Gaussian filtering is a linear smoothing filter that reduces the impact of noise by performing pixel-weighted averaging on an image. It uses the characteristics of the Gaussian function to give the center pixel of the filter a higher weight, while the surrounding pixels have a lower weight. The specific steps are as follows:

[0121] Create a Gaussian mask. The mask size and standard deviation determine the size and strength of the filter.

[0122] The template is applied to each pixel of the image and a weighted average is calculated as the new value of the current pixel.

[0123] Repeat the above steps for the entire image.

[0124] Gaussian filtering can better preserve image details and edge information while removing Gaussian noise.

[0125] The two filtering algorithms are expressed as the following formulas:

[0126] Median filter:

[0127] I new (x,y)=median(I(x,y))

[0128] Gaussian filtering:

[0129] Inew (x,y)=∑(w*I(x+i,y+j)) / ∑(w)

[0130] Where I(x,y) represents the pixel value in the original image; I new (x, y) represents the pixel value after filtering; median is the value in the middle position after a group of data is sorted by size; (x, y) represents the coordinates of the pixel; (x+i, y+j) represents the coordinates of the adjacent pixels in the filter window; w represents the weight of the corresponding pixel.

[0131] These filtering algorithms are implemented by calling corresponding functions in image processing software, programming languages ​​or libraries. They are widely used in image processing to purify images, improve image quality, and provide more accurate results for subsequent image analysis and calculations.

[0132] 3. Image enhancement: Use histogram equalization, contrast enhancement and other algorithms to enhance the details and contrast of ionospheric images.

[0133] Image enhancement refers to changing the appearance of an image through a series of algorithms and technical means to make it more visually appealing and interpretable. Among them, histogram equalization and contrast enhancement are commonly used image enhancement methods, which are particularly suitable for improving the details and contrast of ionospheric images.

[0134] a) Histogram equalization is a lossless grayscale image enhancement method that improves the contrast of an image by adjusting the grayscale distribution of the image. The principle is to equalize the grayscale distribution of the image so that the originally densely distributed grayscale is mapped to a wider range, thereby increasing the details in the image. Histogram equalization is expressed by the following formula:

[0135]

[0136] Where T(r) is the output gray level corresponding to the input gray level r, L is the total number of gray levels, MN is the total number of pixels in the image, and p r (j) is the normalized frequency of pixels with gray level j in the input image in the entire image.

[0137] Histogram equalization can significantly enhance the contrast of an image, but in practical applications it may cause brightness distortion or increase in noise. Therefore, it needs to be improved by combining other methods.

[0138] b) Contrast enhancement is to enhance the contrast of the image by adjusting the pixel values ​​in the image and expanding the grayscale range. Contrast enhancement is expressed by the following formula:

[0139] I {text{new}}(x,y)=A*I {text{old}} (x,y)+B

[0140] Among them, I {text{new}} (x, y) is the enhanced image pixel value, A and B are the gain and bias parameters respectively, I {text{old}} (x,y) is the original image pixel value.

[0141] Contrast enhancement controls the contrast of an image by adjusting the gain and bias parameters. The gain parameter is used to adjust the dynamic range of pixel values, while the bias parameter is used to adjust the brightness level of the image. By properly selecting the parameters, the details and contrast of the image can be improved. The details and contrast of ionospheric images can be effectively improved through algorithms such as histogram equalization and contrast enhancement, thereby improving the visualization of the image and the ability to interpret signals.

[0142] 4. Image restoration: For defective or damaged ionospheric image data, image restoration algorithms are used to repair them to make the data more complete and accurate.

[0143] Image restoration is a technology for processing defective or damaged ionospheric image data, aiming to enhance the integrity and accuracy of the data by restoring the missing information. Image restoration algorithms are based on mathematical models and image processing techniques, and reconstruct the visual information of the image by filling in the missing areas or repairing the damaged parts.

[0144] Interpolation is the process of inferring unknown data points through the trends between known data points. For missing ionospheric image data, the interpolation algorithm estimates the pixel values ​​of the missing parts by analyzing and interpolating the known data points. Common interpolation methods include linear interpolation, bilinear interpolation, and cubic spline interpolation. These methods are based on the relationship between neighboring pixels and use local information to fill in the missing parts.

[0145] In addition to interpolation algorithms, other image restoration methods include model-based methods and optimization-based methods. Model-based methods use prior knowledge of image structure and texture to establish mathematical models to describe the characteristics of the image, and use these models to perform restoration. Partial differential equation-based methods use the gradient information of the image to restore missing or damaged areas. Optimization-based methods transform the image restoration problem into an optimization problem, and find the best restoration result by minimizing a certain objective function. Common optimization methods include non-negative matrix factorization, sparse representation, and total variation minimization.

[0146] Mathematical models play an important role in image restoration, understanding the structure and characteristics of images and providing accurate mathematical descriptions for restoration algorithms. Suppose there is a defective ionospheric image data, which is represented as a two-dimensional matrix image, where each pixel value represents the intensity of the ionosphere. By establishing a mathematical model, the location, type, and possible value of the defect are understood, thereby guiding the design and implementation of the restoration algorithm.

[0147] Image restoration is a key image processing task that uses mathematical models and image processing techniques to repair defective or damaged image data to make it more complete and accurate. Various algorithms and methods are selected according to specific application requirements to achieve the best restoration effect.

[0148] Step 13: Data quality assessment.

[0149] Specifically, in step 13, the quality of the preprocessed and denoised ionospheric image data is evaluated to ensure the accuracy and reliability of the data. The quality of the ionospheric image data is evaluated using some quality indicators, such as signal-to-noise ratio, image clarity, etc.

[0150] Quality assessment of ionospheric image data after preprocessing and denoising is an important step to ensure data accuracy and reliability. When evaluating the quality of ionospheric image data, some common quality indicators are used for analysis and comparison.

[0151] 1. Signal-to-Noise Ratio (SNR): The signal-to-noise ratio is a measure of the ratio between the signal and the noise level. In ionospheric image data, noise may be caused by sensor noise, interference, and other environmental factors. The signal-to-noise ratio is calculated by calculating the ratio of the average signal strength of the region of interest (ROI) in the image to the surrounding background noise. A higher signal-to-noise ratio indicates that the image has a stronger signal relative to the noise, and therefore has better quality.

[0152] Assuming the average signal intensity of the region of interest (ROI) in the image is S and the surrounding background noise intensity is N, the signal-to-noise ratio (SNR) is expressed as:

[0153] If the average signal strength of a region in an ionospheric image is 100 and the surrounding background noise strength is 20, then the signal-to-noise ratio is:

[0154] A higher signal-to-noise ratio (SNR>3) generally indicates that the image signal is stronger than the noise and the image has better quality.

[0155] 2. Image clarity: Image clarity refers to the clarity of details in an image. In ionospheric images, clarity is usually related to the visibility of edges and details. Some metrics for evaluating image clarity are used, such as Structural Similarity Index (SSIM) and Sharpness. These metrics quantify image clarity by comparing the differences in structure, details, and contrast between the original image and the preprocessed image.

[0156] Assume that the original image is I1, the preprocessed image is I2, and the image clarity C is expressed as: C = f(SSIM(I1, I2), Sharpness(I1, I2))

[0157] Wherein, SSIM(I1,I2) represents the structural similarity index of the original image and the preprocessed image, Sharpness(I1,I2) represents the sharpness of the original image and the preprocessed image. f represents a function that comprehensively calculates the structural similarity index and sharpness to obtain the image sharpness.

[0158] Weighted average: C = w1×SSIM(I1,I2)+w2×Sharpness(I1,I2) (where w1 and w2 are weights, and w1+w2=1)

[0159] 3. Uniformity and consistency: The uniformity and consistency of ionospheric image data are important considerations for evaluating its quality. Uniformity refers to the consistency of intensity distribution within the same area of ​​the image. Consistency refers to the consistency of intensity changes between different areas of the image. The reliability of the data is evaluated by analyzing the overall uniformity and consistency of the ROI in the image.

[0160] Assume that the ionospheric image data is I, a certain area in the image is A, and the ROI area is R.

[0161] Uniformity is expressed by the following formula: For area A,

[0162]

[0163] Where |A| represents the number of pixels in area A.

[0164] The consistency is expressed by the following formula: For different regions B and C,

[0165]

[0166] The formula for evaluating data reliability is expressed as:

[0167] Reliability = f(overall uniformity (R), overall consistency (R),

[0168] Where f is a function defined according to a specific evaluation criterion. Uniformity and consistency are equally important, and f is defined as:

[0169] f(uniformity, consistency) = (uniformity + consistency) / 2.

[0170] The quality of the pre-processed and denoised ionospheric image data is evaluated by using quality indicators such as signal-to-noise ratio, image clarity, and uniformity consistency to ensure the accuracy and reliability of the data. These evaluation indicators provide an objective measure of data quality and help optimize the image restoration and processing process to obtain high-quality results.

[0171] Step 2, when extracting interference features, analyze and calculate based on the spectrum, time domain and spatial domain information of the shortwave signal, such as Figure 3 As shown, the following steps are included:

[0172] Step 21: extracting spectrum features.

[0173] Specifically, in step 21, the ionospheric image data is Fourier transformed to obtain its spectrum information. The peak value and frequency band range in the spectrum, as well as the frequency distribution, are analyzed to extract spectrum features. Statistical indicators such as the mean, variance, and kurtosis of the spectrum are calculated, or a spectrum-based feature extraction algorithm, such as fast Fourier transform (FFT) and wavelet transform, is used to extract more specific spectrum features.

[0174] The ionospheric image data is converted from the time domain to the frequency domain through Fourier transform. Fourier transform decomposes the signal into a combination of sine and cosine waves of different frequencies to obtain a spectrum representation. By analyzing the spectrum graph, the peaks and frequency band ranges in the spectrum are observed.

[0175] 1. Peak analysis: Observing the peaks in the spectrum helps determine the frequency components with strong energy in the ionosphere. Peaks indicate high energy at that frequency, which may have an important impact on certain physical processes in the ionosphere. Identify and count these peaks and further study the physical phenomena associated with them.

[0176] Assume that the frequency in the spectrum graph is f and the energy is E(f).

[0177] The peak value can be defined as follows: For a given frequency range, there exists fp such that for any f, E(f p )≧E(f).

[0178] The number of statistical peaks can be expressed as: where [E(f) = E(f p )] is an indicator function, which takes the value of 1 when E(f) reaches its peak value, and 0 otherwise.

[0179] The physical phenomenon related to the peak value is expressed as: P(f p ) ～ Φ(f p ), where P(f p ) represents the peak value f p Related physical phenomena, Φ(f p ) represents the research function on it.

[0180] 2. Frequency band analysis: By observing the frequency band range in the spectrum diagram, the frequency distribution range in the ionosphere can be determined. This helps to understand the distribution of different frequency components in the ionosphere. The ratio of high-frequency components to low-frequency components and their changing trends in different time periods can be analyzed to explore the physical processes and behaviors in the ionosphere.

[0181] 3. Frequency distribution analysis: In addition to observing the peak value and frequency band range, the statistical indicators of the frequency distribution are also calculated, such as the mean, variance, kurtosis, etc. of the spectrum. These statistical indicators provide more specific information about the characteristics of the spectrum. The mean provides the center position of the frequency distribution, the variance describes the degree of dispersion of the distribution, and the kurtosis reveals the steepness of the peak of the distribution.

[0182] A spectrum-based feature extraction algorithm is used to further extract spectrum features. Among them, Fast Fourier Transform (FFT) is a commonly used algorithm that can efficiently perform spectrum analysis. In addition, wavelet transform is also a commonly used spectrum analysis method, which provides better time-frequency locality and is suitable for analyzing the spectrum characteristics of non-stationary signals.

[0183] Step 22: time-frequency feature extraction.

[0184] Specifically, in step 22, the ionospheric image data is subjected to time-frequency analysis using methods such as short-time Fourier transform (STFT) and continuous wavelet transform. Based on the time-frequency analysis results, the features of different time periods and frequency bands are extracted. The features such as energy change and frequency evolution of the time-frequency image are calculated, or other algorithms are used to extract more specific time-frequency features.

[0185] Time-frequency analysis of ionospheric image data is an effective method to help understand the frequency characteristics of the ionosphere that change over time. In time-frequency analysis, a variety of methods are used, among which short-time Fourier transform (STFT) and continuous wavelet transform are commonly used techniques.

[0186] 1. Short-time Fourier transform (STFT): STFT is a method for analyzing signals in two dimensions: time and frequency. It divides the signal into multiple time windows and performs Fourier transform on each time window to obtain the spectrum information of the signal in the time window. When performing STFT, different window functions are selected to balance the frequency and time resolution. By superimposing the results obtained by STFT, a time-frequency image is generated to show the energy distribution of ionospheric signals in different time periods and frequency bands.

[0187] 2. Continuous wavelet transform: Continuous wavelet transform is also a common time-frequency analysis method. It uses a set of wavelet functions as basis functions to decompose and reconstruct the signal to obtain the time-frequency information of the signal at different scales and positions. Continuous wavelet transform provides better time-frequency locality than STFT and is suitable for analyzing the time-frequency characteristics of non-stationary signals.

[0188] By performing time-frequency analysis, various features are extracted from the time-frequency image, including but not limited to the following aspects:

[0189] 1. Energy changes in time-frequency images: By calculating the energy of each time period and frequency band in the time-frequency image, the strength changes of ionospheric signals in different time periods and frequency bands are observed. This provides information about ionospheric activity, including the occurrence of energy peaks, energy increases and decreases, etc.

[0190] 2. Frequency evolution: Time-frequency analysis reveals how the frequency components in a signal change over time. By observing the evolution of frequency over time, different frequency patterns in the ionosphere are discovered, such as increases, decreases, or periodic changes in frequency. This helps to understand dynamic phenomena in the ionosphere, such as ionospheric disturbances, waves, and waveguides.

[0191] 3. More specific time-frequency feature extraction: In addition to energy changes and frequency evolution, other algorithms are used to extract more specific time-frequency features. Statistical indicators such as entropy, variance, and kurtosis of time-frequency images are calculated to describe the distribution characteristics of time-frequency images. In addition, machine learning methods such as convolutional neural networks and support vector machines are used to classify or extract features from time-frequency images to further understand the time-frequency characteristics of ionospheric signals.

[0192] Step 23: spatiotemporal feature extraction.

[0193] Specifically, in step 23, the ionospheric image data is subjected to time series analysis to analyze the rules and characteristics of its time series changes. Statistical indicators such as the mean, variance, and time series correlation of the ionospheric image data are calculated to extract time series features. Methods such as difference, differential relative motion, and correlation coefficient of time series data are used to extract spatiotemporal features.

[0194] 1. Statistical indicator analysis: Calculate the mean, variance, standard deviation and other statistical indicators of ionospheric image data in time series to understand its overall average level, range of variation and data volatility.

[0195] 2. Time series correlation analysis: Calculate the correlation coefficient between ionospheric image data, such as the Pearson correlation coefficient, to measure the correlation between different time series. This helps identify whether there are certain time series patterns or trends and explore their interrelationships.

[0196] 3. Differential analysis: By performing differential operations on the ionospheric image data, the difference values ​​between adjacent time points are calculated to extract the trend characteristics of time series changes. The differential data reflects the rate of change of the ionospheric image data, which helps to capture more subtle time series patterns.

[0197] Assume that the ionospheric image data is at time point t i The value of I(t i ), then the adjacent time point t i and t i+1 The difference between them is: ΔI(t i )=I(t i+1 )-I(t i )

[0198] The rate of change of the differential data is expressed as:

[0199] 4. Differential relative motion analysis: Based on the differential data, the relative motion of different time series is further analyzed. The relative displacement and motion speed between different time series are calculated by using methods such as the amplitude correlation function (ARF) or the position trace iteration method (TPI). This reveals the motion and evolution laws of the ionospheric image data in the time series.

[0200] 5. Temporal and spatial feature extraction: In addition to temporal changes, the temporal and spatial characteristics of ionospheric image data are also considered. The temporal data is combined with spatial information, and the spatial correlation of ionospheric image data at different time points is calculated to obtain the temporal and spatial correlation characteristics. This is achieved using correlation coefficient matrices, spatial filtering methods, etc.

[0201] Step 3: Establish an interference feature library, and determine whether the shortwave signal is interfered by comparative analysis and feature matching. Figure 4 As shown, the following steps are included:

[0202] Step 31, establishing an interference feature library.

[0203] Specifically, in step 31, different types of shortwave interference data are collected, including spectrum, time-frequency and space-time features. Feature extraction is performed on the interference data to extract spectrum features, time-frequency features and space-time features. An interference feature library is established to store the extracted feature information in a database or other form of data storage structure.

[0204] For the collected interference data, the following methods are used to extract spectrum features, time-frequency features and space-time features, and establish an interference feature library.

[0205] 1. Spectral feature extraction: Analyze the collected interference data in the frequency domain and extract the spectral features. Spectral features include signal power spectrum density, frequency distribution characteristics, etc. Commonly used spectrum analysis methods include Fourier transform (FFT) and wavelet transform. These methods convert the interference signal in the time domain into energy distribution in the frequency domain, and then extract the spectral features.

[0206] 2. Time-frequency feature extraction: Shortwave interference usually has the characteristics of time variation, so analysis in the time-frequency domain better captures the dynamic characteristics of the interference signal. Time-frequency features are extracted by using time-frequency analysis methods, such as short-time Fourier transform (STFT), continuous wavelet transform (CWT), etc. These methods provide information in time and frequency at the same time, helping to obtain the time-frequency distribution characteristics of the interference signal.

[0207] 3. Temporal and spatial feature extraction: The temporal and spatial features of shortwave interference data include the temporal changes and spatial distribution of interference signals. For temporal changes, the statistical features of interference signals at different time points, such as mean, variance, and temporal correlation, are calculated. For spatial distribution, the interference data is segmented in space, and the spectrum features and time-frequency features of each spatial segmentation area are extracted. In this way, interference features of different areas are obtained.

[0208] In the feature extraction stage, machine learning methods are used to assist in feature selection and extraction. The model is trained using a classifier based on a feature selection algorithm, and the most discriminative features are selected through feature importance evaluation of the model. Common machine learning algorithms are used, such as support vector machine (SVM), random forest, etc.

[0209] When establishing an interference feature library, the extracted feature information is stored in a database or other form of data storage structure. It is organized and stored according to different dimensions such as interference category and time for subsequent data retrieval, analysis and application.

[0210] Step 32, real-time feature matching.

[0211] Specifically, in step 32, the ionospheric image data received in real time is obtained. The real-time data is subjected to feature extraction, and the spectrum features, time-frequency features and time-space features are extracted. The features in the interference feature library are compared and analyzed, and the similarity or distance measurement between them is calculated. According to the similarity or distance measurement results, it is determined whether the shortwave signal is interfered with, and the type and degree of interference.

[0212] In order to obtain real-time ionospheric image data, ionospheric monitoring equipment such as ionospheric sounders or satellite receiving devices are used. These devices provide relevant information such as the strength, density, and height of the ionosphere for subsequent analysis and processing.

[0213] Spectral feature extraction: Spectral features are obtained by performing frequency domain analysis on ionospheric image data. This includes the power spectrum density and frequency distribution of the signal. Fourier transform or wavelet transform is used to convert the time domain signal into a frequency domain signal and extract the spectral features.

[0214] Time-frequency feature extraction: The dynamic changes of the ionosphere need to be analyzed in the time-frequency domain to better capture its time-frequency characteristics. The analysis in the time-frequency domain is achieved through methods such as short-time Fourier transform (STFT) and continuous wavelet transform (CWT). These methods provide the characteristics of the changes in ionosphere data in time and frequency.

[0215] Extraction of spatiotemporal features: The spatiotemporal features in ionospheric image data mainly include the temporal changes and spatial distribution of ionospheric parameters (such as density, height, etc.). For temporal changes, the statistical characteristics of ionospheric parameters at different time points are calculated, such as mean, variance, temporal correlation, etc. For spatial distribution, the ionospheric image is segmented, and the spectral features and time-frequency features of each spatial region are extracted to obtain the ionospheric characteristics of different regions.

[0216] After feature extraction is completed, the features will be compared and analyzed with the features in the interference feature library. By calculating the similarity or distance measurement between them, it is determined whether the shortwave signal is interfered with and the type and degree of interference are determined. A higher similarity means that the real-time data is similar to the features in the interference feature library, which may indicate the presence of an interference signal. Conversely, a lower similarity may indicate that the received shortwave signal is relatively normal.

[0217] Step 33: Interference identification and discrimination.

[0218] Specifically, in step 33, it is determined whether the shortwave signal is interfered with according to the result of real-time feature matching. If interfered with, the type and degree of interference are determined according to the feature information of the matching result. According to the result of the determination of the type and degree of interference, corresponding measures are taken, such as locating the interference source, adjusting the parameters of the receiving device, etc., to reduce or eliminate the influence of interference on the shortwave signal.

[0219] According to the result of real-time feature matching, it is determined whether the shortwave signal is interfered. When the matching result shows that the received signal has a high degree of similarity with the features in the interference library, it is determined that the shortwave signal is interfered.

[0220] To identify the type and degree of interference, further analyze the feature information of the matching results. Through feature extraction and comparative analysis, determine the type of interference, such as narrowband noise, broadband noise, FM interference, Doppler-induced frequency deviation, etc. At the same time, evaluate the degree of interference based on the intensity and distribution of the interference features.

[0221] According to the results of the interference type and degree, appropriate measures are taken to reduce or eliminate the impact of interference on shortwave signals. A common measure is to locate the interference source, and calculate the approximate location of the interference source through information such as signal time difference and phase difference of multiple receiving devices, so as to take further countermeasures. In addition, the parameters of the receiving device, such as gain and filter settings, are adjusted to optimize the signal reception quality and resist interference.

[0222] In addition to positioning and parameter adjustment, other technical means are also used, such as frequency hopping, protocol optimization, coding error correction and other methods to enhance the robustness of shortwave signals and reduce the impact of interference on signals.

[0223] In summary, the present invention provides a shortwave interference feature extraction method based on ionospheric image processing, which is characterized by automation, high efficiency and accuracy, and effectively improves the ability to identify and analyze shortwave interference. The method has a wide range of application potentials and can be used in the fields of monitoring and maintenance of shortwave communication systems, radio spectrum management, and electromagnetic environment monitoring, and has important application and promotion value.

Claims

1. A shortwave interference feature extraction method based on ionospheric image processing, characterized in that: The steps include: Step 1: Use digital image processing algorithms to preprocess and denoise the ionospheric image data: Acquire the ionospheric image data received by shortwave signals, and use digital image processing algorithms to preprocess and denoise the ionospheric image data; Step 2: When extracting interference features, analyze and calculate based on the spectrum, time domain and spatial domain information of the shortwave signal: Extract interference features from ionospheric image data, including spectrum features, time-frequency features, and time-space features; Step 3: Establish an interference feature library and determine whether the shortwave signal is interfered by comparative analysis and feature matching: An interference feature library is established to store the feature information of various shortwave interferences. Through comparative analysis, the features of the ionospheric image data received in real time are matched with the interference feature library. Based on the matching results, it is determined whether the shortwave signal is interfered with, as well as the type and degree of interference.

2. The shortwave interference feature extraction method based on ionospheric image processing according to claim 1 is characterized in that: The step 1 is specifically as follows: Step 11, obtain the ionospheric image data received by shortwave signal: Collect ionospheric imaging data using the Global Ionospheric Mapper or Ionospheric Sounding Radar; Step 12: Preprocess and denoise the ionospheric image data using digital image processing algorithms: a) Image grayscale adjustment: use histogram equalization to even out the grayscale distribution of the image: Calculate the grayscale histogram of the original image, calculate the cumulative histogram, and calculate the grayscale mapping function based on the cumulative histogram: In the above formula, r represents the original grayscale, L represents the total number of grayscales, N and M represent the width and height of the image respectively, and n j Represents the number of pixels at gray level j; b) Image filtering: Median filtering: Select a filter window size, which will slide on the image; sort the pixels in the window by value, and select the middle value as the new value of the current pixel; repeat the above steps until the filtering operation is completed on the entire image. new (x,y)=median(I(x,y)); Gaussian filtering: Create a Gaussian template. The size and standard deviation of the template determine the size and strength of the filter. Apply the template to each pixel of the image and calculate the weighted average as the new value of the current pixel. Repeat the above steps for the entire image. new (x,y)=∑(w*I(x+i,y+j)) / ∑(w); In the above formula, I(x,y) represents the pixel value in the original image; I new (x, y) represents the pixel value after filtering; median is the value in the middle of a group of data sorted by size; (x, y) represents the coordinates of the pixel; (x+i, y+j) represents the coordinates of the adjacent pixels in the filter window; w represents the weight of the corresponding pixel; c) Image Enhancement: Histogram Equalization: In the above formula, T(r) is the output gray level corresponding to the input gray level r, L is the total number of gray levels, MN is the total number of pixels in the image, and p r (j) is the normalized frequency of pixels with gray level j in the input image in the entire image; Contrast Enhancement: I {text{new}} (x,y)=A*I {text{old}} (x,y)+B In the above formula, I {text{new}} (x, y) is the enhanced image pixel value, A and B are the gain and bias parameters respectively, I {text{old}} (x,y) is the original image pixel value; d) Image restoration: The interpolation algorithm estimates the pixel values ​​of the missing parts by analyzing and interpolating known data points, including linear interpolation, bilinear interpolation, and cubic spline interpolation; Model-based methods use prior knowledge of image structure and texture to build mathematical models to describe image features and use these models for repair; partial differential equation-based methods use image gradient information to restore missing or damaged areas; optimization-based methods seek the best repair result by minimizing a certain objective function. Optimization methods include non-negative matrix decomposition, sparse representation, and total variation minimization. Step 13, data quality assessment: The quality of the ionospheric imagery data was assessed using the following quality metrics: a) Signal-to-noise ratio: Assuming the average signal strength of the region of interest in the image is S and the surrounding background noise strength is N, the signal-to-noise ratio SNR is expressed as: b) Image clarity: Assume that the original image is I1, the preprocessed image is I2, and the image clarity C is expressed as: C = f(SSIM(I1, I2), Sharpness(I1, I2)); SSIM(I1,I2) represents the structural similarity index of the original image and the preprocessed image, Sharpness(I1,I2) represents the sharpness of the original image and the preprocessed image, and f represents a function that comprehensively calculates the structural similarity index and sharpness to obtain the image sharpness; Weighted average: C = w1×SSIM(I1,I2)+w2×Sharpness(I1,I2), where w1 and w2 are weights and w1+w2=1; c) Uniformity and consistency: Assume that the ionospheric image data is I, a certain area in the image is A, and the area of ​​interest in the image is R; Uniformity is expressed by the following formula: For area A: In the above formula, |A| represents the number of pixels in area A; The consistency is expressed by the following formula: For different regions B and C: The formula for evaluating data reliability is expressed as: Reliability = f(overall uniformity (R), overall consistency (R)), where f is a function defined according to specific evaluation criteria. If uniformity and consistency are equally important, f is defined as: f(uniformity, consistency) = (uniformity + consistency) / 2.

3. The shortwave interference feature extraction method based on ionospheric image processing according to claim 1 is characterized in that: The step 2 is specifically as follows: Step 21, spectrum feature extraction: Perform Fourier transform on the ionospheric image data to obtain its spectrum information, analyze the peak value and frequency band range in the spectrum, as well as the frequency distribution, and extract the spectrum characteristics; Peak analysis: Let the frequency in the spectrum be f, the energy be E(f), and the peak value be defined as: for a given frequency range, there exists fp such that for any f, there is E(f p )≧E(f); The number of statistical peaks is expressed as: where [E(f) = E(f p )] is the indicator function, which takes the value of 1 when E(f) is at its peak value, and 0 otherwise; The physical phenomenon related to the peak value is expressed as: P(f p )-Φ(f p ), where P(f p ) represents the peak value f p Related physical phenomena, Φ(f p ) represents the research function thereof; Frequency distribution analysis: Calculate the statistical indicators of frequency distribution, including the mean, variance and kurtosis of the spectrum; Step 22, time-frequency feature extraction: Use short-time Fourier transform and continuous wavelet transform to extract the energy change and frequency evolution of the time-frequency image from the time-frequency image, calculate the entropy, variance and kurtosis of the time-frequency image, and use machine learning methods to classify or extract features from the time-frequency image; Step 23, spatiotemporal feature extraction: a) Statistical indicator analysis: Calculate the mean, variance and standard deviation of ionospheric image data in time series; b) Time series correlation analysis: Calculate the correlation coefficient between ionospheric image data to measure the correlation between different time series; c) Differential analysis: By performing differential operations on ionospheric image data, the difference values ​​between adjacent time points are calculated to extract the trend characteristics of time series changes; Assume that the ionospheric image data is at time point t i The value of I(t i ), then the adjacent time point t i and t i+1 The difference between them is: ΔI(t i )=I(t i+1 )-I(t i ); the rate of change of the data after differentiation is expressed as: d) differential relative motion analysis; e) Spatiotemporal feature extraction.

4. The shortwave interference feature extraction method based on ionospheric image processing according to claim 1 is characterized in that: The step 3 is specifically as follows: Step 31, establish an interference feature library: Collect different types of shortwave interference data, including spectrum, time-frequency and space-time features, extract features from the interference data, extract spectrum features, time-frequency features and space-time features, and establish an interference feature library; When establishing an interference feature library, the extracted feature information is stored in a database or data storage structure, and is organized and stored according to interference categories and time dimensions; Step 32, real-time feature matching: Acquire the real-time received ionospheric image data, extract the features of the real-time data, extract the spectrum features, time-frequency features and time-space features, compare and analyze the features in the interference feature library, calculate the similarity or distance measurement between them, and judge whether the shortwave signal is interfered, as well as the type and degree of interference based on the similarity or distance measurement results; Step 33, interference identification and discrimination: Based on the results of real-time feature matching, determine whether the shortwave signal is interfered with. If interfered with, the type and degree of interference are determined based on the feature information of the matching results.

Images